Trajectory controller

IS 2 255 271 T3

DESCRIPTION

Path controller.

The present invention relates to a novel flow control system, and more particularly to a flow control system for use in motion control applications. Even more particularly, the present invention relates to a novel motion control system for use in friction welders.

In most electromagnetic systems, the transfer of energy from one component of the system to another is critical for the proper functioning of the system. In many electromagnetic systems, this energy transfer is achieved by properly exciting one component of the system to establish a magnetic flux that interacts with another component of the system to transfer energy from the excited component to the other component. Despite the fact that energy transfer is achieved by flow, in known electromagnetic systems the flow of the system is not directly controlled. Instead, the current and / or voltage applied to the driven element is controlled and, based on assumed relationships between current, voltage, and flux, it is assumed that current and / or voltage control based on these assumed relationships will produce the proper flow. This current and / or voltage control is typically implemented because, to date, the prior art has not provided an efficient, low-cost, and easy-to-implement system for directly controlling flow in an electromagnetic system.

A disadvantage of current and / or voltage control systems as described above is that the relationships between current, voltage and flow cannot be represented mathematically in a simple way and vary non-linearly depending on a variety of variables. For example, the particular characteristics of each piece of magnetic material in a system will result in voltage, current, and flux relationships that vary from system to system, and even within a certain system, from one section of the system to another. Due to these different relationships of voltage, current and flow, it is difficult to precisely and correctly control currents and / or voltages to produce the desired flow and thus the desired energy transfer. As such, the prior art is limited in its ability to provide an electromagnetic system in which flow is directly controlled.

The lack of a suitable flow control system in the prior art is particularly evident in electromagnetic systems where it is desired to ultimately control the force exerted by one component of the system on another component of the system. In such systems, the actual force produced by the system is related to the flow established by the excited component of the system. However, as described above, since the prior art cannot directly and precisely control flow, it cannot, therefore, precisely control the force produced by such systems.

The inability of the prior art to precisely control established forces in an electromagnetic system is particularly severe in applications where the movement of at least one component of the system must be precisely controlled. An example of such an application is a friction or vibration welder to drive a thermoplastic part to be welded with a linear, orbital, rotational or arbitrary vibratory movement relative to another thermoplastic part, with both parts in forced contact along the surfaces of the same to be welded so that the relative movement of the parts to each other causes friction to heat the parts along the intersections of the same so that when the movement ceases, the pieces cool down and are welded together.

Friction welders are especially suitable for use in welding thermoplastic parts by means of linear, rotational or orbital vibratory forces that induce frictional heating in the parts. This frictional heating at the interface of the surfaces to be welded causes the parts to melt at their contact surfaces and bond together as they cool. Although the vibratory forces that generate frictional heating can be created through mechanical coupling means, it is common to employ an electromagnetic system to generate the necessary controlled motion.

There are numerous electromagnetically or hydraulically actuated friction welders commercially available to operate in a linear vibratory mode. However, the motion of these friction welders is not ideal. Due to the linear or reciprocating motion of the weld component, the frictional forces at the interface of the weld materials are translational and descend to zero speed each time the weld components reverse. When the components are at zero speed, no heat is produced, since friction welding is a resistance process that provides heat proportional to the product of the resistance forces and the average of the square of the relative velocity of the components in the interface.

In addition, many linear motion welding systems use electromagnetic systems or actuators that use the well-known Scott-T magnetic circuit to convert three-phase electrical energy into a single-phase mechanical movement. In such systems, due to the electromagnetically linked nature of the actuator component of the system, it is difficult to precisely control the movement of the moving element in all directions and to limit the zero speed ranges for the moving element. Accordingly, alternative movements and controllers have been developed for friction welding components that seek to reduce or minimize zero speed components and simplify control circuits.

IS 2 255 271 T3

One such alternative is rotary welding, in which the welding components are rotated about an axis, and rotational forces, not linear motion, produce frictional heating at the interface. However, the rotational forces are proportional to the radial distance from the center of rotation, and consequently, neither the speed of the components nor the resulting heating is radially uniform. Furthermore, friction welding is generally limited to applications where the parts to be welded have a circular geometry.

A second alternative is to produce an orbital motion of the welding components electromagnetically. During orbital motion, the velocity of the components remains constant as the parts rub, thus generating the same amount of frictional heating that is generated by linear motion, but with less force required and less relative displacement of the welding components.

Despite the heating advantages of orbital welding, there are parts that are not amenable to orbital motion welding, but which are amenable to either linear or rotary motion welding. Accordingly, electromagnetically actuated friction welders have been developed capable of producing either linear motion or orbital motion. Such a friction welder is disclosed in US Patent No. 5,378,951 to Snyder. The electromagnetic drive system of these friction welders is similar in several respects to that of an electromagnetic motor.

In such systems, three coupled magnetic coils are positioned equidistant around the circumference of the welder, in a plane parallel to the plane of motion. The coupled magnetic coils are electrically connected in either a delta or a Y-shaped connection to primarily form an orbital motor stator component. A triangular armature or rotor component separately formed as a single body of magnetic material is centrally positioned relative to the stator component, such that each face of the triangular armature is adjacent to a magnetic coil. The armature is held in the horizontal orbital plane by a bending spring support system connected to a large stationary frame. The orbital movement of the armature results from the application of a controlled three-phase AC current to the coupled magnetic coils, producing a force in the armature proportional to the flux generated. This armature motion can be decomposed into vectors of displacement, velocity, and acceleration proportional in amplitude to the sine and cosine of twice the AC power line frequency. Linear motion of the armature is produced by adding a second orbital motor or by dividing each coupled magnetic coil of a single orbital motor into two sections and selectively applying current to several sections in combinations either in series or in parallel.

Several disadvantages arise in producing orbital motion using coupled magnetic coils. First, the use of coupled magnetic coils reduces the overall performance of the system, since the force generated in one direction always generates neutralizing force elements in the opposite direction due to the coupling of the flux paths in the magnetic circuit. Second, the system is incapable of producing motion that is neither orbital nor linear, that is, pure arbitrary motion. It is desirable to produce arbitrary motion of the welding components when the system needs to compensate for uneven armature mass distribution or when random orbits are desired.

Eventually, the control systems to produce the armature's orbital or linear motion become complex. When coupled magnetic coils are used in an orbital motor, the magnetic flux within the system is forced to sum to zero. If in addition it is also forced to add zero to the AC phase currents, there are not enough degrees of freedom in the magnetic system to generate the arbitrary forces to produce an arbitrary motion. However, if the phase currents are not forced to sum to zero, there are sufficient degrees of freedom in the magnetic system to produce the arbitrary forces, but continuous flux operation is required to generate those arbitrary forces.

The flux through each air gap between the magnetic coils and the adjacent faces of the armature in such systems is a function of all three-phase and non-linear magnetic currents. There is no unused or unexcited magnetic coil at any time. This limits the ability to use any form of precise flow control in such systems.

An electromagnetic drive unit for a friction welder in which the magnetic flux is monitored is known from US-A-4 715 523.

The invention is defined in claim 1. Some optional features of the invention are defined in the dependent claims.

The objects, features and advantages of the invention as well as the presently preferred embodiments thereof will become more apparent from a reading of the following description in conjunction with the accompanying drawings.

Figure 1 illustrates a novel flow control system in accordance with certain aspects of the present invention.

Figures 2A and 2B illustrate in greater detail an exemplary embodiment of a flow control system such as that illustrated in Figure 1.

Figure 2C illustrates an exemplary embodiment of a zero return error control circuit.

IS 2 255 271 T3

Figure 3A generally illustrates an exemplary embodiment of an open loop flow observer that can be used in the control system of Figure 1.

Figure 3B generally illustrates a circuit using a low pass filter that provides a signal corresponding to flux in an electromagnetic system.

Figure 4 generally illustrates an electromagnetic actuator that can be used with the novel flow control system of Figure 1.

Figure 5 generally illustrates an example of a force control system that implements a desired force-to-flow transformation constructed in accordance with certain teachings of the present invention.

Figures 6A-6D generally illustrate, through the use of pseudo-code blocks, an exemplary form of a force-to-flow transformation that can be implemented through the use of a microcontroller or program microprocessor.

Figure 7A generally illustrates a representation of the various phase coils of the actuator of Figure 4 that should be driven to produce a force for a desired force factor in terms of X and Y.

Figure 7B generally illustrates a group of functional blocks which, when implemented, generate the appropriate flow commands to implement the flow-to-force transformation reflected by Figure 7A.

Figures 8A-8C generally illustrate functional blocks for a force-to-flow controller that can be used to produce linear motion of a moving member.

Figure 9 provides a high-level block diagram of an exemplary position control system useful in understanding the present invention. It can be used to build a vibration welding apparatus.

Figure 10A generally illustrates functional blocks that can be used to implement a position controller for use in the apparatus of Figure 9, useful in assisting in an understanding of the present invention.

Figure 10B illustrates functional blocks similar to those of Figure 10A that are adjusted to reflect a sampling-based controller useful in assisting in an understanding of the present invention.

Figure 11A generally illustrates an exemplary set of functional blocks that can be used to implement a position command generator useful to aid in an understanding of the present invention.

Figure 11B illustrates a position command generator similar to that of Figure 10A that operates on a sampling basis to produce position command signals in response to sampled input signals, useful in assisting in an understanding of the present invention.

Figures 12A-12D generally illustrate the operation of a position controller of the type illustrated in Figure 9 for various modes of operation, useful in assisting in an understanding of the present invention.

Figure 13 generally illustrates one form of a QD control system constructed in accordance with certain teachings of the present invention, useful in assisting in an understanding of the present invention.

Figures 14A-14D generally illustrate functional blocks that can be used to implement the exemplary QD controller of Figure 13, useful in assisting in an understanding of the present invention.

Corresponding reference characters indicate corresponding parts throughout the various views of the drawings.

Turning to the drawings, and in particular to Figure 1, a novel flow control system 10 constructed in accordance with certain aspects of the present invention is shown. In general, the novel system 10 includes a flow controller 12 that receives on an input flow command 13 and outputs a phase coil drive signal 14. The phase coil drive signal 14 is provided, through suitable means, to an electromagnetic system 15. A flow observer 16 provides a feedback signal to the flow controller 12 that corresponds to the flow in the electromagnetic system 15. In general, flux controller 12 provides phase coil drive signals 14 that drive electromagnetic system 15 so that flux in electromagnetic system 15 meets flux command 13.

The electromagnetic system 15 can be any electromagnetic system that includes at least one phase coil and a core, in which the phase coil can be excited through the application of electrical energy (for example, through the application of a voltage and / or a controlled current) to establish a flow in the core. For example, the electromagnetic system 15 can be as simple as an iron core inductor, with a phase coil wound around a core, or a transformer having primary and secondary windings each wound around a core.

IS 2 255 271 T3

The flow control system 10 of the present invention is believed to have particular application to electromagnetic systems 15 constituting electromagnetic actuators in which excitation of system 15 causes movement of a moving element. For example, the flow control system of Figure 1 is believed to have particular applicability to electromagnetic systems that include a main core (or a plurality of magnetically decoupled cores) around which one or more phase coils are wound. . In such embodiments, a movable element is typically provided that interacts with the core when the nucleus is excited such that excitation of the core causes movement of the movable element. The moving element can be a piece of paramagnetic material (eg steel) or a stack of laminations of such materials that interacts with the excited core in a similar way to the interaction between the rotor and the stator of a reluctance machine. Alternatively, the moving element may include one or more permanent magnets (or electromagnets, or even induction coils) that interact with the excited core in a manner similar to the interaction between the rotor and stator of a permanent magnetic machine. Although the following discussion is performed in the context of one or more specific electromagnetic systems 15, those skilled in the art will understand that the described flow control system 10 can be used with a number of electromagnetic systems 15 and that the discussion herein it is to illustrate and explain the present invention and not to limit the scope of the claims presented herein.

In Figure 1 the electromagnetic system 15 is illustrated as having a single phase coil. In applications where the electromagnetic system 15 includes a plurality of phase coils, the drive signal 14 may take the form of a drive vector that includes different drive signals for each of the individual phase coils. In such applications, the flux command 13 may take the form of a flux command vector that includes different flux command signals for each of the various phase coils. In such applications, the flux associated with each of the phase coils can be controlled independently. For the sake of clarity, the following discussion initially focuses on the structure and operation of the flow controller 12 in the context of a system having an electromagnetic system 15 that includes a single phase coil. Those skilled in the art will appreciate that a multiphase flow controller 12 can be constructed simply by duplicating the single phase flow control system described herein.

Figures 2A and 2B show in greater detail an embodiment of a flow control system 10. Figure 2B shows an electric drive circuit for driving the electromagnetic system 15 and Figure 2A shows a control circuit that provides the control signals for the actuator of Figure 2B. The actuator of Figure 2B will be discussed first.

In the illustrated embodiment, the electromagnetic system 15 is illustrated as a single phase coil wound around a core. The phase coil defines two ends. The two ends of the phase coil are coupled through a high voltage DC bus (V_DCa) by a switching bridge comprising an upper switching device 17, a lower switching device 18, a lower flyback diode 19 and an upper flyback diode 20. The switching devices 17 and 18 can be any suitable controllable switching device such as MOSFET, BJT, BIFET, IGBT, MCT, standard FET power transistors or any other suitable switching device that can be controlled through the application of a signal. of control. In the illustrated embodiment, the upper switching device 17 is controlled by the control signal SU and the lower switching device 18 is controlled by the control signal SL. The flyback diodes 19 and 20 can be any device that exhibits current characteristics (eg, unidirectional) similar to those of a diode.

Generally, the switching signals SU and SL can be activated to produce three switching states:

(i) a first state in which both upper and lower switching devices 17 and 18 become conductive;

(ii) a second state in which only one of the switching devices becomes conductive; and (iii) a third state in which none of the switching devices becomes conductive. When the switching signals SU and SL are in the first state, so that both the upper and lower switching devices 17 and 18 become conductive, the phase coil 18 will be coupled across the V bus_CCa, thus causing the electric current to flow from the positive rail of the V bus_CCa (for example, + V_CCa), through the phase coil of the electromagnetic system 15, to the ground of the V bus_CCa. This current flow will cause power to be transferred from the V bus_CCa up to electromagnetic system 15, thus resulting in increased flux from system 15.

When the switching signals SU and SL are in the second state, so that only one of the switching devices 17 or 18 becomes conductive and the other becomes non-conductive, the voltage applied across the phase coil of the system 15 will be approximately zero. Under this second switching state, any current in the phase coil will "flow freely" through the conductive switching device and one of the flyback diodes. For example, if there is current in the phase coil 15 and the signals SU and SL are such that the upper switching device 17 becomes conductive and the lower switching device 18 is non-conductive, the current in the phase coil "will flow. freely ”from the positive rail of the V bus_CCa, through the phase coil, and back to the positive rail of the V bus_CCa across the top return diode 20. The voltage applied across the phase coil under such circumstances will be the voltage drop across the flyback diode 20 plus the voltage across the device 17, or a voltage of approximately zero. When the lower switching device 18 becomes conductive and the upper switching device 17 non-conductive, a similar substantially zero voltage state will be obtained. Under such circumstances, current will flow freely through the lower switching device 18 and the lower flyback diode 19. When the current in the phase coil is in a free-flowing state, the flux established by the phase coil will remain substantially constant or decrease

IS 2 255 271 T3 slightly. As such, the energy in the electromagnetic machine, and hence the flux, will remain substantially constant or will decrease slightly.

When the switching signals SU and SL are in the third state, such that both upper and lower switching devices 17 and 18 are open, any current in the phase coil at the initiation of this switching state will continue to flow because the phase coil it is an inductive element and the current cannot drop to zero instantaneously. However, since the upper and lower switching devices 17 and 18 are non-conductive, the path for this current flow will be from the ground rail of the V bus._DCa, through the lower flyback diode 19, through the system phase coil 15 and to the positive rail of the V bus_CCa across the top return diode 20. Therefore, in this third state, the phase coil of system 15 will be negatively coupled across the V bus._CCa so that the negative of the V bus_CCa It is applied through the phase winding. This application of the negative of the V bus_CCa the phase winding will tend to drive any flux, and therefore current, in the phase winding down to zero. Therefore, when the signals SU and SL are in the third state, the energy stored in the electromagnetic system 15 will be dissipated or returned to the V bus_CCa, and the energy in the system, and therefore the flow, will decrease.

Those skilled in the art will appreciate that the particular switching scheme illustrated in Figures 2A and 2B is just one such scheme that can be used to control the driving of a phase coil. Other switching systems can be used, for example switching systems using a complete H-bridge with four switching devices. In general, any switching arrangement can be used to implement the systems described herein that allow the excitation of the phase coil to be controlled so that the flux in the system 15 is increased, remains substantially constant, or is reduced.

In the embodiment of Figures 2A and 2B, the flow controller 12 provides the switching signals SU and SL. A schematic representation of an exemplary flow controller 12 is provided in Figure 2A.

With reference to Figure 2A, the illustrated flow controller 12 comprises an addition connection 25 which receives, at a positive input, a flow command signal 13 corresponding to the desired flow level and, at a negative input, a signal from flux feedback from flux observer 16 corresponding to flux in electromagnetic system 15. The addition connection 25 subtracts the flow feedback signal from the flow command 13 to produce an error signal that varies with the difference between the actual flow and the desired flow. When the flow error signal is positive, the flow command is greater than the flow feedback, and it will be necessary to increase the flow in system 15 to bring the flow to the level of the flow command. When the error signal is negative, the feedback flow is greater than the flow command, and it will be necessary to decrease the flow in the system 15 to bring the flow in the system to the level of the flow command. The error signal from the add connection 25 is first amplified by an amplifier 26 and then provided to a drive control circuit 21.

The excitation control circuit 21 generates switching signals SU and SL to increase, reduce or stabilize the flux in the electromagnetic system 15 depending on the magnitude of the error signal.

The drive control circuit 21 can take many forms. For example, the controller may use a conventional form of pulse width or pulse frequency modulation to control the drive of the phase coil. Alternatively, the drive control circuit 21 can take the form of a controller that will close both switching devices 17 and 18 when the error signal is positive; it will open both switching devices when the error signal is negative; and it will open one switch device and close the other when the error signal is zero. According to one embodiment of the phase controller 21, the switching signals SU and SL are controlled so that the error signal is reduced to zero and allowed to circulate freely until the error signal falls outside a controlled hysteresis band. Such a time-hysterical controller is referred to herein as a "return-to-zero error" controller or RZE. The RZE controller described is advantageous because it provides, for applications where the desired flow command is not changing rapidly and the available energy is large enough to force a follow-up relatively quickly, a fast follow-up of the command signal from flow so that the "lag" in the flow reference tracking is less than a sampling period of the control system.

A detailed schematic illustrating one embodiment of an RZE control circuit as described herein is illustrated in Figure 2C. Generally, the RZE controller 21 of FIG. 2C includes an upper comparator 22A and a lower comparator 22B that together establish two adjacent asymmetric hysteresis bands centered around zero error. Each of the comparators handles only one error polarity. The outputs of the two comparators are provided to inverters 23a and 23b and the outputs of the two inverters correspond to the signals SU and SL.

Generally, when the magnitude of the error signals from amplifier 26 is large and positive (indicating that there is a need to increase flux in the electromagnetic system), the outputs of both comparators 22a and 22b will be low. Assuming that the upper and lower switching devices 17 and 18 are of the type that becomes conductive when a positive voltage is applied to the control gates of the devices, the low outputs of the comparators 22a and 22b will be inverted by the inverters 23a and 23b to produce high-level SU and SL signals, thus placing the switching devices in the first switching conduit, so that the flux in the electromagnetic system 15 will tend to increase.

IS 2 255 271 T3

When the magnitude of the error signals from amplifier 26 is large and negative, the outputs of both comparators 22a and 22b will be high. These high signals will be reversed by inverters 23a and 23b, resulting in low voltage SU and SL signals that will cause both the upper and lower switching devices 17 and 18 to become non-conductive, thus reducing the current (and therefore the flux) in the electromagnetic system.

When the error signal from amplifier 26 is between large positive values ​​and large negative values, the state of the switching signals SU and SL will depend on the magnitude of the error signal, as it is compared to the + V voltages._H and V_H hysteresis. In the illustrated embodiment, the application of the voltages + V_H and V_H hysteresis is controlled by an activation signal T so that when T (24a) is logically high, the voltage + V_H hysteresis is applied to the positive input of comparator 22a and no hysteresis voltage is applied to the lower comparator 22b, and that when T (24b) is logically low, the voltage -V_H Hysteresis voltage is applied to lower comparator 22b and no hysteresis voltage is applied to upper comparator 22a. As explained below, the drive signal T changes state to control the application of the hysteresis voltages + VH / -VH to comparators 22a and 22b.

Assuming that there is an initial state in which the trigger signal is logically high, and the flux in the electromagnetic system 15 is below the command flux so that the error signal of the amplifier 26 is positive and large and the outputs of comparators 22a and 22b are both low (resulting in logically high SU and SL signals). Under such conditions, the flux in system 15 would begin to increase, and thus, the magnitude of the error signals from amplifier 26 would begin to decrease. At some point, the magnitude of the error signal will begin to decrease from a large positive value towards zero. The instant the error signal reaches and exceeds zero, the output of the lower comparator 22b (which has no hysteresis feedback) will change state, thus resulting in a transition of the signal SL from high to low. This will cause the switching arrangement to be in the second switching state, such that a current in the electromagnetic system 15 will circulate freely, resulting in a constant or slightly decreasing flow. This change of state of the signals SU and SL will also cause a change in the state of the activation signal T, which will therefore result in the application of the voltage -V_H feedback to the lower comparator. If the flux decreases to the point where the amplifier error signal 26 now falls below the hysteresis value, the output of the lower comparator 22b will change state again, resulting in the signal SL changing low. to high and therefore apply electricity back to the electromagnetic system 15 causing an increase in flux in the system 15, and there will be a chopping in the error signals between -VH and zero.

Assuming there is a second set of initial conditions in which the drive signal is logically low, and the flux in the electromagnetic system 15 is above the command flux so that the error signal of the amplifier 26 is negative and large and the outputs of comparators 22a and 22b are both high (resulting in logically low SU and SL signals). Under such conditions, the flux in the electromagnetic system 15 would begin to decrease, and thus the magnitude of the error signal from the amplifier 26 would begin to increase. At some point, the magnitude of the error signal will increase from a large negative value to zero. By the time the error signal reaches and exceeds zero, the outputs of both the upper and lower comparators 22a and 22b will have changed state, thus resulting in a transition of both SU and SL signals from logically low to logically high. . This will cause the switching distribution to be in the first switching state, so that the current in the electromagnetic system will increase, resulting in an increasing flux. This change in the state of the signals SU and SL will also cause a change in the state of the drive signal T, thus resulting in the application of the feedback voltage + VH to the upper comparator 22b. If the flux increases to the point where the error signal from amplifier 26 now exceeds the hysteresis value, the output of the upper comparator 22b will change state again, resulting in the signal SU changing from logically high to logically high. short. The signal SL will logically remain high, and therefore, the switching arrangement will be in the free-flow state, causing the flow in system 15 to remain constant or to decrease slightly, and there will be chopping in the error signals between zero and + VH.

Thus, as described above, the hysteresis controller of Figure 2C can control the flow in the electromagnetic system 15 so that the flow: (i) increases when the flow is below the desired level by a negative hysteresis amount. ; (ii) decrease when the flow is above the desired flow level by a positive hysteresis amount; and (iii) is allowed to remain substantially constant or to decrease slightly when the flow is between the positive and negative hysteresis values. In US Patent No. 5,530,333 entitled "Control of an Inductive Load" by Michael J. Turner, issued June 25, 1996, which is incorporated herein by reference in its entirety, additional details may be found regarding the structure and operation of a hysteretic controller of the type described in connection with FIG. 2C.

Those skilled in the art having the benefit of this disclosure will appreciate that the flow controller 21 of Figure 2C is but one of several flow controllers that can be used to implement the flow control system of Figure 1.

Referring again to FIG. 1, it can be seen that in the illustrated flow control system 10, a flow feedback signal from a flow observer 16 provides an indication of the level of flow in the electromagnetic system 15. The flow observer 16 can take the form of a flow sensor (eg, a magnetometer); a Hall effect probe such as a thin film Hall device; a superconducting quantum interference device (SQUID); or a flux calculator that uses, for example, the measurement of the curved flux air gap of a given coil.

IS 2 255 271 T3

For systems in which the phase coils of the electromagnetic system 10 are excited so that there are regular periods during which each phase coil experiences a state of zero flux (that is, each coil has zero flux inside it for a nonzero interval), the flux associated with each coil can be estimated by an open loop flux observer that is zeroed during a known zero flux interval for that phase coil. Such an open loop flux observer can provide an indication of flux through the application of the known relationship between flux associated with a coil and the applied voltage and current in that coil. That known relationship is reflected in the following equation 1:

Flux = Integral of [V_phase_n (t) - i_phase_n (t) * R] / Nt Equation 1 in which V_phase_n (t) is a signal corresponding to the phase coil voltage as a function of time; i_phase_n (t) is a signal corresponding to the phase coil current as a function of time; R is a value corresponding to the phase coil resistance; and Nt is the number of turns that the phase coil comprises.

To determine the approximate flux for each phase coil, a conventional open-loop integrator can be used. To avoid drift problems, the open loop integrator can be zeroed during known zero flux states to minimize the build-up of uncertainty in the integrator over time. The zeroing of the integrator can occur either in a timed manner (in which the zero flow states occur at known time intervals) or the integrator can be a simple stabilized integrator. When the electromagnetic system 15 is excited such that only unipolar flux is established in the system, a stabilized integrator that is zero stabilized can be used.

Figure 3A generally illustrates an exemplary embodiment of an open loop flow observer 30 that may be used in the control system 10 illustrated in Figure 1. Referring to Figure 3A, the exemplary open loop flow observer comprises a stabilized integrator circuit 31 that receives as its input a signal that is equal to the sum of four voltage input signals. The four voltage input signals that determine the input to the integrator circuit 31 are, from top to bottom in Figure 3: (i) a voltage + V INTCP that corresponds to a constant voltage drop that is associated with the operation of the power switching devices 17 and 18; (ii) a voltage signal corresponding to the magnitude of the phase current i; (iii) a + V signal_DC which is switchable coupled to the input of inverter 31; and (iv) a -V signal_DC which is switchable coupled to the input of inverter 31. The + V signals_DC and V_DC correspond to DC voltage levels that are related in a known way to the positive and negative levels associated with high voltage V_CCA applied to the system phase coil 15. Normally, the + V signals_DC and V_DC they will be proportionally less than the + VCCA and -VCCA signals associated with the high voltage DC bus.

In the embodiment of Figure 3A, the + VCC signal is coupled to the input of integrator 31 through a controlled switch 32a that becomes conductive when both the SU and SL signals are logically low (that is, when the coupled switching arrangement to the phase coil is actuated so that the + VCCA bus is applied to the phase coil). Similarly, the -VCC signal is coupled to the input of integrator 31 through a controlled switch 32b that becomes conductive when both the SU and SL signals are logically high (that is, when the switching arrangement is coupled to the coil of so phase is actuated that the -VCCA bus is applied to the phase coil). The output of the stabilized integrator 31 is -1 * the flux (the signal is inverted). An amplifier with gain = -1 can be used at the output of the stabilized integrator to view the flux estimate. The reversal of the flux estimate is exploited in the flux controller error amplifier. The error amplifier is an adding amplifier with the flux reference and flux estimate as inputs resulting in the gain K (flux reference - flux estimate) at its output.

Due to the known relationship between the + VCC and -VCC signals and the VCCA bus, the voltage appearing at the input of integrator 31 will correspond directly to the voltage applied to the phase coil. As such, an integration of the input voltage applied to an integrator will produce a signal that corresponds directly to the flux in the electromagnetic system. The illustrated circuit has a stabilized flow estimate output> 0.

The use of the switching devices 32a and 32b and the + VCC and -VCC signals to provide a signal representing the actual voltage applied to the phase coil is believed to be beneficial because the magnitude of the actual DC bus value is normally relatively high (on the order of several hundred volts more). As such, large and expensive components would be required to directly integrate the relatively high voltages that are applied to the phase coil. Using the approach of Figure 3A, cheaper, lower-voltage devices can be used to provide an accurate indication of flow in the system 15. Those skilled in the art who have the benefit of this description will appreciate that the actual voltages applied to the Phase coil could be used to generate the input voltage for integrator 31. Alternatively, searchcoils using a phase voltage ratio could be used and directly integrated to form the coil.

In the exemplary circuit of Figure 3A, phase current is applied to a resistor 33 to provide a voltage signal that is intended to correct the input to integrator 31 for coil resistance. The value of resistor 33 used for the correction operation described above can be selected in a number of different ways. For a relatively simple correction factor, the value of resistor 33 can be selected as an unalterable value representing an estimate of the phase coil resistance over operating conditions.

ES 2 255 271 T3 expected from the associated electromagnetic system. Since R will vary with temperature and other operating factors, the selected R-value will only be a rough estimate of the actual R-value for each phase winding. The current offset can be improved if the R-value is measured / estimated / calculated through the use of a thermal model and electrical measurements or techniques currently used to estimate actual resistances, such as DC voltage injection.

Again referring to Figure 3A, it will be seen that even through actuator switching, the SU and SL signals are used to develop a low voltage signal corresponding to the high voltage signal actually applied to the phase coil. , where the actual phase current i, not the switching currents, is the current that is used to obtain the current correction factor. The actual phase current should be used to provide a more accurate current correction factor, as the switching current will not necessarily correspond to the phase current. In embodiments where the actual voltages applied to the phase coils are estimated using switching signals from the actuator and a low voltage bus, the current correction factor should be appropriately scaled before subtracting the current correction factor of the voltage corresponding to the voltage applied to the phase coils.

Although actual integrators can be used to obtain an estimate of the flow in system 15 as described in connection with Figure 3A, a more simplistic approach can be used in many applications. In particular, when the flux in system 15 is bipolar zero average (discontinuous or continuous), the flux of the system can be roughly estimated by simply low-pass filtering the voltage applied to the phase coil (V - iR) (or filtering by low-pass a voltage that, like the voltage applied to integrator 31 of Figure 3A, corresponds to the phase coil voltage). Such estimation of flux in system 15 through the use of low pass filters is beneficial in certain applications as it can minimize drift and random behavior problems associated with certain integrators. Figure 3B generally illustrates such an approach in which a signal corresponding to the phase coil voltage (which can be corrected for coil resistance) is applied to a low-pass filter 34 to provide a signal corresponding to the flux associated with the phase coil. The signal provided will also reflect a device voltage drop that is always opposed to the drive voltage and that would change sign in applications involving bipolar drive currents.

When using low-pass filters instead of integrators to estimate the flux for a given phase coil, the filter time constant should be greater than the period associated with the fundamental frequency of flux drive for that coil. For example, if the phase coil is driven at a fundamental frequency of 100 Hz (a period of 10 ms), the time constant for the low pass filter for that phase coil should be approximately 50 ms or greater. Given the 5: 1 (or greater) ratio of the filter time constant to the fundamental frequency of the voltage and the zero average voltage of each coil voltage, the low-pass filters will roughly integrate the coil voltage to provide a useful estimation of coil fluxes.

The novel flow control system 10, and its various components, described above in connection with Figures 1-3B, can be beneficially used in a number of different applications. For example, the flow control system 10 can be used to control the flow in a rotating electrical machine, such as a conventional induction motor, a universal motor, a switched reluctance motor, or a permanent magnet motor, or a hybrid motor ( for example, IP and RC). The novel flow control system described herein can also be used in various calibration devices in which the flow through a core must be controlled to a predetermined desirable level. Furthermore, the flow control system described herein can be used to control electromagnetic actuator devices in which the movement of a moving element is controlled through control of the flow passing through one or more paramagnetic cores.

Figure 4 generally illustrates an electromagnetic actuator 40 that can be used with the novel flow control system 10 described above. In general, the electromagnetic actuator 40 comprises a stationary outer assembly 41 and a movable element positioned within a gap defined by the stationary assembly 41. In the illustrated embodiment, the stationary outer assembly 41 comprises three substantially identical magnetically decoupled flux generators comprising cores 43A, 43B and 43C in E. Each E-core comprises a stack of substantially identical laminations of a paramagnetic material (e.g. steel) defining an E-shaped core having a central arm and two secondary arms, in which the secondary arms are positioned on either side. of the central arm. A fork part couples the central arm to the secondary arms. In the illustrated embodiment, for each E-core, the width of the central leg is greater than the width of the secondary legs. In one embodiment, for each E-core, the widths of the two secondary arms are substantially identical and are approximately half or slightly less than half the width of the central arm. In general, the same construction techniques used to construct the stator cores of switched reluctance machines can be used to construct cores 43A, 43B, and 43C at E.

Placed around the center arms of each of the cores 43A, 43B and 43C at E is a phase coil. In the illustrated embodiments, each of the phase coils A, B, and C has the same number of turns and is formed in the same way, such that the three phase coils A, B, and C are "symmetrical."

IS 2 255 271 T3

Each phase coil is positioned around the central arm of its respective E-core, so that when electrical energy is applied to the phase coil, a current will be established in the phase coil that will establish a flux through the core at E. In figure 4 approximates of the flux paths that will be established when coils A, B and C phase cores 43A, 43B and 43C are provided by curved lines in E. As reflected in the figure, the flow paths for the three cores at E are substantially identical and the flow path for each core at E defines a central flow path through the central arm and two secondary flow paths through of the secondary arms.

Referring again to Figure 4, it can be seen that the arrangement of the three cores 43A, 43B and 43C at E is such that the cores at E define a generally triangular web. Placed within this core is a movable element 42 with a substantially triangular shape. In the illustrated embodiment, the moving element 42 comprises a stack of substantially identical laminations of paramagnetic material (eg, steel), although alternative embodiments are envisioned in which the moving element includes permanent magnets (induction coils) or electromagnets.

As those skilled in the art having the benefit of this disclosure will appreciate, in the electromagnetic actuator 40 of FIG. 4, driving one of the phase coils of one of the E-cores will produce a force on the moving element that it will tend to cause the moving element to move toward the excited E-shaped core. This is because whenever a phase coil is energized, the moving element will tend to move to a position where the reluctance of the energized phase coil is minimized. Thus, if phase coil A associated with core 43A at E is energized, moving element 42 will tend to move downward toward core 43A at E to minimize reluctance of energized coil A.

In the embodiment illustrated in Figure 4, the three cores 43A, 43B and 43C at E are arranged so that the movement of the movable element can be controlled along two degrees of freedom. For example, using the X, Y reference indicators of Figure 4, the movable element can be controlled to move in both the positive and negative X directions (a first degree of freedom) and in both the positive and positive Y directions. negative (a second degree of freedom), as well as along any path defined by points X and Y. Therefore, the use of the three nuclei in E allows to control the moving element along two degrees of freedom.

Although the specific actuator of Figure 4 allows the movable element to be controlled along two degrees of freedom, those skilled in the art will appreciate that different numbers and arrangements of the E-cores can be used to control movement along more or less less degrees of freedom. For example, if an actuator was desired that was capable of moving along a single degree of freedom (for example, only along the positive and negative X-axis), then a substantially rectangular moving element could be used with as few as two nuclei in E. Generally, when the movement of the moving element is caused by the tendency of the moving element to move to a position in which the reluctance of an energized coil is minimized, N + 1 phase coils will be required to control N degrees of freedom.

In alternative embodiments where permanent magnets or electromagnets are placed on the movable element (and thus where energizing a coil with positive or negative current can produce both positive and negative forces over a degree of freedom ), only N coils are required to control N degrees of freedom. Referring again to actuator 40 of FIG. 4, it can be seen that the flow paths associated with the three cores 43A, 43B, and 43C at E are independent of each other. In other words, the three E-nuclei illustrated are "magnetically uncoupled." An advantage of this feature is that the flow in each of the cores in E can be controlled independently. Furthermore, in the embodiment of Figure 4, the phase coils A, B, and C associated with the three cores at E are separately excitable. In other words, the phase coils are "electrically decoupled". As such, the phase currents in each phase coil can be controlled independently of the currents in the other phase coils. As explained in more detail below, this ability to independently control the flux and current in each E-core independent of the flux and currents in the other E-cores is advantageous in many respects.

Referring again to Figure 4, it can be shown that for each E-core, the force exerted on the movable member that tends to move the member toward the excited E-core is proportional to the square of the flux passing through the central arm. of the nucleus in E and generally corresponds to the following equation 2:

2—0S^(flow2) * μο * S where po is a constant that reflects the magnetic permeability of the air, S is a value corresponding to the cross-sectional area of ​​the central arm of the core at E that is parallel to a face of a moving element 42, and flux is a signal that corresponds to flux in the core at E of interest. The flow value may be provided by a flow observer of the type described above in connection with the description of the flow observer 16.

Using matrix representations, the relationship between the flows in the three nuclei 43A, 43B and 43C in E and the forces exerted on the movable element that tend to move the movable element towards the nuclei 43A, 43B and 43C in E can be represented by the equation matrix 1:

IS 2 255 271 T3

FA 10 0 flow_A^TO2

FB = (1 / [2 * po * S]) * 0 10 * flow_BA2

FC 0 0 1 flow_CA2 in which FA, FB and FC represent the forces that tend to move the moving element 42 towards the moving arms of the nuclei 43A, 43B and 43C in E, respectively, and flow A, flow B and flow C represent the actual or estimated fluxes in the corresponding E nuclei.

By means of simple geometric calculations, the forces that will be exerted on the moving element can be transformed from the coordinates FA, FB and FC into forces in coordinates X and Y using the matrix equation 2:

FX = 0 γ3 / 2 -γ3 / 2 * FA; FA, FB, FC> 0

FY -1 / / FB

FC

It can be appreciated from the matrix equations n^you 1 and 2 that for a certain arbitrary desired force, in terms of a suitable reference frame (e.g. FX and FY), there exists a significant number, more potentially an infinite number, of solutions of flow_A, flow_B and flow_C that are capable of producing said desired force. Furthermore, due to the decoupled nature of the E-cores and phase coils of the actuator 40 of FIG. 4, there is no limitation that would make any of these potentially infinite solutions illicit. As such, due to the nature of the actuator illustrated in the figure, the optimal solution of flow_A, flow_B and flow_C can be selected. Once the desired flow solution is selected, it can be implemented through the use of flow control systems 10 of the type described above in connection with Figures 1-3A. Typically, a separate flow control system 10 will be required to control the flow in each of the three cores 43A, 43B, and 43C in E. In general, an optimal “force into flow” solution or transformation to convert a desired arbitrary force X, Y to the values ​​of flow_A, flow_B, and flow_C needed to produce that force will be the solution that minimizes the net flow of the system, and therefore, minimize the amount of energy required to establish that flow. Such a solution will usually provide the most energy efficient approach to setting the desired force. Furthermore, for many electromagnetic actuators, there will be force-to-flow transformations that will require the actuators to operate so that a discontinuous unipolar flow is established in the actuator core or cores. Such a discontinuous flow operation allows the use of the open loop flow observers described above. An advantage of the novel electromagnetic actuator illustrated in figure 4 is that the optimal transformation of force into flow requires the operation of the actuator in such a way that a discontinuous unipolar flow is established in the three cores 43A, 43B and 43C in E during the operation of the actuator. .

The optimal force-to-flow transformation for a given system can be obtained by: (i) establishing the relationship between the forces associated with the different actuating elements (for example, the cores in E) and the flows associated with those actuating elements (for example, determining the relationship reflected by the matrix equation 1); (ii) establishing the relationship between the desired forces in a given reference frame (eg X, Y) and the forces associated with the different actuating elements (eg determining the relationship reflected by matrix equation 2); and (iii) solving for the actuator element flows in terms of the desired forces in the given reference system and selecting the solution that is the minimum norm solution. Although any minimum norm solution can be used, it has been found beneficial to select the force-to-flow solution or transformation that is the minimum Euclidean norm solution. The concept of a minimum Euclidean norm solution will be understood by those of skill in the art and is discussed generally on page 166 of Modern Control Theory (3^to Ed.) By William L. Brogen.

Once the desired force-to-flow transformation is established for a given system, it can be implemented in practice through a force control system. An exemplary embodiment of such a force control system is shown in Figure 5.

With reference to figure 5, a force control system 50 is shown that includes an electromagnetic actuator 40, of the type illustrated in figure 4, which is driven by a flow controller 12 'that receives as input some commands cmd_flux_A, cmd_flux_B and desired flow_flow_cmd and a flow_st_flow, flow_B_est, and flow_C_est flow feedback signals. The flow estimation commands are provided by a flow estimator 16 'which, per flow coil, can take the form of any of the flow observers described above in connection with the flow observer 16 of FIG. 1. The flux controller 12 ', for each phase, compares the flux command with the flux estimate in a comparator 19', amplifies the error signal in an amplifier 26 ', and generates appropriate phase coil drive signals across the use of an excitation control circuit 21. The construction of the controller 12 'may, per phase, follow the description provided above in connection with the flow controller 12 of Figure 1. The drive circuits required to drive phase coils A, B, and C are not illustrated in Figure 5, although it will be apparent to those skilled in the art that they have the benefit of this disclosure.

IS 2 255 271 T3

Coupled to the flow controller 12 'is a flow force controller 51 which receives at its inputs desired force commands in a given reference frame (the X, Y and Z reference frame in Figure 5) and provides at its inputs. Output appropriate flow commands in terms of cmd_flow_A, cmd_flow_B, and cmd_flow_C. The flow controller 12 'receives and acts on the flow commands in a manner similar to that described above in connection with the flow controller 12 of Figure 1.

Although the force-in-flow controller 51 can be constructed from exclusively analog circuitry, in the illustrated embodiment of Figure 5, the force-in-flow controller 51 includes a digital circuit, such as a microprocessor or microcontroller, that is programmed in appropriate way to implement a desired force-to-flow transformation. The use of a digital circuit to build the force-in-flow controller can be beneficial because the force-in-flow transformation can be easily implemented by mathematical relationships that are easily implemented in digital circuits and because the use of digital circuits allows for easy modification of the transformation. force in flow. In addition, when digital circuitry is used to implement the force-in-flow controller 51, the input force commands can be either digital or analog. If they are analog, some form of analog-to-digital conversion will be required to transform the force commands into proper digital values. Similarly, when using a force-in-flow controller 51, it may be required to convert the desired flow commands (if digital) into analog signals if an analog flow controller is used. The construction and programming of a force-in-flow controller 51 will be within the ability of one of ordinary skill in the art having the benefit of this description.

Figures 6A-6D illustrate, through the use of a pseudo-code block, an exemplary form of a force-in-flow controller 51 that can be implemented through the use of a programmed microcontroller or microprocessor. The force-to-flow transformation implemented by the controller illustrated by Figures 6A-6D corresponds to the minimum Euclidean norm solution for the actuator 40 of Figure 4. Those skilled in the art who have the benefit of this description will appreciate that although the description of the controller 51 is in terms of discrete pseudo-code functional blocks, the controller can be implemented through one or more programmed processors, analog circuits, or a combination of both. .

With reference to FIG. 6A, the general operation of the force-in-flow controller 51 is divided into three high-level functional blocks 60, 61, and 62. Typically, the functional block 60 labeled FXY_a_FABC receives the desired force commands in terms of a certain reference frame (in this case the XY reference frame) and converts the force commands to force commands that are consistent with the layout. physics of actuator 40 (here, the arrangement of cores 43A, 43B, and 43C at E). Function block 61 receives the force commands in terms of the arrangement of actuator 40 and converts those force commands to flow commands corresponding to the phase coils of actuator 40. Function block 62 is an optional function block not required for all implementations of controller 51 that receives the flow commands from block 61 and modifies the commands to ensure that each of the flow commands is zero for a finite time interval . The use of the "zero_flow" block 62 ensures that the flows in the actuator 40 are discontinuous, thus allowing for the use of some of the beneficial forms of flow observers described above in connection with the flow observer 16. . Each functional block will be discussed in more detail below.

Figures 6B1-6B4 illustrate the structure and operation of a functional block 60 that transforms the input force commands in terms of FX and FY into force commands in terms of the forces FA, FB and FC that can be generated directly by actuator 40. In the illustrated embodiment, functional block 60 first uses the FX and FY commands to obtain six intermediate control signals F_a1, F_a2, F_b1, F_b2, F_c1, and F_c2 in functional blocks 63a, 63b, and 63c. Functional blocks 63a-63c essentially multiply each of the FX and FY commands by a 2X2 state matrix that, per phase coil, implements a positive or negative inverse of each possible partition of the matrix equation 2. The correct solution is then selected from among the possible solutions. The 2X2 state matrices used by the functional blocks 63a-63c are illustrated, respectively, in Figures 6B2, 6B3 and 6B4. Each of the functional blocks 63a, 63b, and 63c provides two output force values ​​because implementing the inverse of matrix equation 2 will produce two solutions for each of FA, FB, and FC. The suitable solution for the system is selected by functional blocks 64a, 64b and 64c.

Referring again to FIG. 6B1, each of the functional blocks 64a, 64b, 64c receives as its inputs the two force solutions of its associated functional block 63a, 63b, or 63c. Due to the nature of the functional blocks 63a, 63b, and 63c, at least one of the force solutions will be positive. Functional blocks 64a, 64b, and 64c first eliminate all negative force solutions by setting the corresponding intermediate force command to zero and then selecting the larger of the two adjusted force commands. In the specific embodiment of Figure 6B1, the selected force commands from functional blocks 64a, 64b, and 64c are limited to a maximum value by a functional block 65 to produce the force commands Fa, Fb, and Fc per phase coil. limited. Embodiments are envisioned in which the limiting function implemented by block 65 is removed or implemented as a function of some other system parameter.

Referring back to FIG. 6A, once the functional block 60 generates the Fa, Fb, and Fc commands, those commands are processed by the functional block 61 to produce the flow volts_flow_A, volts_flow_B, and volts_flow_C commands. Functional block 61 is illustrated in greater detail in FIG. 6C. The operation of the functional block 61 represents a simple implementation of the bounded inverse (> 0) of the matrix equation 1

ES 2 255 271 T3 per phase coil. Specifically, any negative Fa, Fb, and Fc command is set to zero, since the actuator 40 cannot produce negative force. From the adjusted force commands FA, FB, and FC, the required phase coil flux is decided by taking the square root of the adjusted force command for a given phase multiplied by a constant value corresponding to

See equation 2 above. The outputs of function block 61 are the flow command volts_flow_A, volts_flow_B, and volts_flow_C.

In the particular embodiment of Figure 6A, the flow commands from the functional block 61 are applied as inputs to a flow zeroing functional block 61 which guarantees that the flow commands are zero for a finite time interval. The use of such a flow zero block is not required for most applications and is only essential for linear X motion. The precise shape of the functional block 62 will vary slightly depending on whether the force-in-flow digital controller 51 is operating on a sample or continuous basis. Figure 6D1 illustrates an exemplary construction of a functional block 62 for a digital controller that samples the various control parameter values ​​regularly. Figure 6D2 illustrates a similar functional block 62 'for a continuously operating controller 51.

With reference to the two figures 6D1 and 6D2, the functional blocks 62 and 62 'receive the commands volts_flow_A, volts_flow_B and volts_flow_C of flow and then, in a functional blocks 66a, 66b and 66c for figure 6D1 and a blocks 66a', 66b 'and 66c' functional for FIG. 6D2, generate an average flow command signal that corresponds to a moving average of the flow command signal over a certain time interval. The different way in which these mean flow command values ​​are determined constitutes the most significant difference between the sampled controller reflected in Figure 6D1 and the continuous controller reflected in Figure 6D2. Any form of analog or digital averaging filter could be used with the proper file constant.

Both the actual flow commands and the averaged flow commands are provided to a functional block 67. Function block 67 compares the actual flow commands to a fraction of the mean flow commands and if the actual flow command for a given phase is less than a fraction of the mean flow command, adjusts the flow command to be a minimum flow value. If the actual flow value is greater than the fraction of the mean flow command, then the actual flow command is not adjusted. Thus, the outputs of function block 67 constitute the flow command volts_flow_A, volts_flow_B, and volts_flow_C from the force-in-flux controller 51. Referring to Figure 5, these outputs are then processed by flow controller 12 'to control flow in actuator 40.

Those skilled in the art having the benefit of this disclosure will recognize that the particular force-to-flow transformation reflected in Figures 6A-6D is but one example of a force-to-flow transformation that can be implemented by a force-to-flow controller. built according to certain teachings of this description. For example, an alternate force-to-flow transformation may be based on an angle described by the force commands FX and FY and the physical arrangement of the E-cores that comprise the actuator 40.

Figure 7A generally illustrates a representation of which of the actuator 40 phase coils should be driven to produce force for a desired force vector in terms of X and Y. Basically, Figure 7A serves as a window for which of the coils they will be lit as a function of the direction of the desired force in terms of a vector comprising components FX and FY. In this force-to-flux transformation, as in the previous one, only two phase coils are excited at a given instant. With reference to Fig. 7A, it can be noted that any commanded force corresponding to an angle between 30 degrees and 150 degrees can be generated by a combination of the forces FB and FC with a zero force FA. Similarly, any desired force having a direction between 150 degrees and -90 degrees can be generated from the forces FA and FC without any force FB, and any desired force with a direction vector between -90 degrees and 30 degrees can be generated with FA and FB forces and without FC forces.

Figure 7B generally illustrates a functional block for generating the appropriate flow commands to implement the force-to-flow transformation reflected by Figure 7A. First, the function block illustrated in the figure calculates the proper force angle as a function of the FX and FY commands. Then, using that angle, the block determines in which category of Figure 7A the angle falls, and based on this determination, determines the proper flux commands for the two phase coils to be driven for that category. These generated flow commands are then applied to flow controller 12 ', and the system operates as described above.

Additional force-to-flow transformations are anticipated. For example, the particular force-to-flow transformations described above occurred in the context of an actuator 40 having three disengaged E-cores that is designed to control the motion of the movable member 42 through two degrees of freedom. Alternative embodiments are possible in which the actuator comprises only two E-cores arranged for linear movement of the movable element. In such applications, the only input force command would be an FX (or FY) command. Additionally, four cores could be used in E and FX and FY commands could be provided. The

ES 2 255 271 T3 Figures 8A-8C generally illustrate functional blocks for a flow force controller 80 that can be used in a linear system with two E-cores, each set of two controlling an axis (degree of freedom).

Referring to Figure 8A, the illustrated controller 80 receives a force FX command and outputs flux commands for the two phase coils, designated A and B. In general, a functional block 81 transforms the FX command first. into force commands FA and FB corresponding to the linear actuator and a functional block 82 transforms these force commands into appropriate flow volts_flow_A and volts_flow_B commands.

Details of functional block 81 are provided in Figure 8B. In general, the function block 81 receives the FX command and, if the FX command is negative, it assigns a value of zero to FA and a value equal to the magnitude of FX to FB. On the contrary, if FX is positive, the functional block assigns a zero value to FB and the absolute value of FX to FA. The FA and FB commands are then restricted to a maximum value. Function block 81 provides the FA and FB commands to a function block 82 which determines the appropriate flow commands, volts_flow_A and volts_flow_B.

Functional block 82 is illustrated in more detail in FIG. 8C. Referring to Figure 8C, the functional block 82 basically sets the corresponding flow command if the force command is equal to or less than zero or calculates the appropriate flow command using the inverse of equation 1. The commands volts_flow_A and volts_flow_B Flows are then applied to drive the system to a suitable flow controller.

The force control systems described above in connection with Figures 5-8C have several advantages that are not generally available from known systems for controlling an electromagnetic actuator to produce a desired force. For example, in known control systems, the control variable used to obtain a desired force is either the voltage or the current applied to the phase coils. These systems are inherently limited because, if a voltage control is used, the forces that will be produced will be a non-linear function of the controlled voltage. Such nonlinear control problems are, in practice, difficult and expensive to implement and cannot accommodate changed operating conditions and / or manufacturing tolerances. If a current control is used, the system is unstable in open loop and, for reasonably acceptable performance, the non-linear characteristics of the actuator core must be addressed. This results in the same non-linear difficulties associated with voltage control systems.

The use of a flow control system such as that described herein significantly reduces or eliminates many of the difficulties associated with voltage or current control systems. This is because, when using flow control, the force generated through each air gap of each actuator element is proportional to the square of the flow in the air gap. As such, non-linearities in magnetic material and non-linearities in electrical dynamics need not be accounted for to provide acceptable control. Thus, the force and flow control systems described herein can be used to implement simpler and more efficient control schemes that are easier and more economical to design and implement. Furthermore, when coupled with the use of a simple open-loop flow observer such as that described herein, flow control can result in an extremely efficient and elegant system.

Although the flow control system of Figure 5 allows efficient control of the force exerted on the moving element 42 of the actuator 40, in many applications it is the movement of the moving element (e.g., the position / velocity or the trajectory of the element mobile), not the forces exerted on it, which you want to control. For such applications, the system of Figure 5 can be enhanced to incorporate a position / speed control loop that generates the desired force commands FX and FY in such a way that the movement of the movable element is controlled in a desired manner. The applications of a position / speed control system of the type described above are numerous. For example, such a position / speed control system can be used to build electromagnetic bearings, as well as vibrating and screening apparatus. A specific application of such a position / speed controller is in the field of orbital welders. Such welders typically operate by propelling a first part to be welded (for example, a thermoplastic part) with an orbital motion relative to a second part to be welded such that the relative movement of the first and second parts causes a friction to heat the parts along their intersection, so that when relative motion ceases, the parts will cool and become welded together.

In such welders, a first part is usually placed on a bracket. A second piece is also placed on a support. The parts are then clamped in forced contact with each other along an interface between the parts and an impulse is provided to drive the second part along a predetermined repeating path relative to the first part to frictionally heat the parts to along the interface, so that at the end of the repetitive movement, the pieces are welded together.

Conventional orbital welders are capable of moving the first part to be welded such that the relative movement between the parts is restricted to either linear or orbital movement. These restrictions on the movement of the welder are limiting because, for many applications, a different and arbitrary relative movement is desired so that the relative movement of the parts to be welded can closely correspond to the shape of the parts. Such "match to shape" can significantly increase the strength of the resulting weld. Through the use of the position / speed control system described herein, a welding apparatus can be constructed that is capable of establishing arbitrary relative movement between the parts.

ES 2 255 271 T3 to be welded, including linear motion, orbital motion, rotary motion, or any arbitrary motion of the part to be welded. Those skilled in the art will appreciate that the position / speed control system described herein is but one example of a control system that can be constructed in accordance with the teachings herein and that other features and combinations of features can be controlled. characteristics (for example, acceleration, velocity, position, or any combination thereof).

Figure 9 provides a high-level block diagram of an exemplary position / speed control system 90 that can be used to construct a welding apparatus such as that described above.

Although the following discussion is in the context of a welding apparatus, those skilled in the art who have the benefit of this disclosure will appreciate that the position control system disclosed can be used to control the position / speed of a moving member in other apps.

With reference to Figure 9, the position / speed control system 90 is identical in many respects to the force control system of Figure 5. In particular, within the box 50 of dashed lines, the position control system 90 / speed includes all of the components of the force control system 50 of FIG. 5, all of which function as described above in connection with FIG. 5. Since the actuator 40 of the system is part of a large vibration welding machine, the movable element of the actuator can be attached to a suitable welding arm or other suitable welding tool (not shown).

In addition to including the components of the force control system 50, the position / speed control 90 includes two additional major components. First, in the embodiment of FIG. 9, the force command signals FX and FY are generated by a position / speed controller 91. The position / speed controller 91 receives at its inputs: (i) position command signals 92 representing the desired position / speed (or trajectory) of the moving element of the actuator 40; and (ii) feedback signals from a position / velocity observer 93 representing current position / velocity or trajectory information associated with the movable element of the actuator 40. In general, the position / speed controller 91 compares the position / speed commands with the position / speed feedback information to produce position / speed error signals and uses these position / speed error signals to generate the appropriate force commands that they will tend to bring the moving element to the position / speed indicated by the position / speed commands 92 or to move along the path defined by the path commands 92, when path commands are provided.

The position / speed controller 91 can take many forms depending on how the desired position / speed and / or trajectory of the moving element is defined. In one embodiment, the position / speed controller 91 can be constructed to control the position and speed of the moving element of the actuator 40 in terms of the position of the moving element in an X, Y reference frame and in terms of the speed of the element. mobile in terms of X and Y. In such a system, the two input position commands provided to the position / speed controller 91 will be in terms of cmd_X, cmd_X_dot (X speed), cmd_Y, and cmd_Y_dot (Y speed), and the two feedback commands from observer 93 of position / velocity (described in more detail below) will be in terms of the actual or estimated X and Y positions and velocities of the moving element (eg, real_X, X_dot_real, real_Y, Y_dot_real).

Like the force-in-flow controller 51 described above, the position / speed controller 91 can be implemented through the use of a programmed digital processor, such as a microprocessor or microcontroller. In certain applications, the position / speed controller 91 can be implemented using the same programmed processor that resulted to implement the force-in-flow controller. Analog implementations are also envisioned. In operation, the position / speed controller 91 receives the command signals X, X_dot, Y, and Y_dot from the position / speed command generator and compares the command signals with feedback signals representing the actual or estimated position and speed. of the moving element (real_X, X_dot_real, Y_real, Y_dot_real). As a result of this comparison of the instantaneous position and velocity command and the feedback signals, four error signals (eX, eY, eX_dot and eY_dot) are generated corresponding to the differences between the command signals and the feedback signals. The four error signals can then be multiplied by suitable control gains that can be optimized for each application. Then the error signals from X (eX and eX_dot) are added together and the signals from Y (eY and eY_dot) are added together to produce command signals FX and FY required to place the moving element in the desired position and with the desired speed. Before being provided to the force control system 50, the force FX and FY commands may be band-pass filtered to reduce bias and noise content. The filtered force command signals FX and FY are then provided to the force-in-flow controller 51 of the flow control system 50. Depending on the parasitic modes of the system and other factors, the filter and filtering operation may or may not be necessary. In one embodiment of the position / speed controller 91, feedback signals representing the position and velocity X, Y of the actuator moving element 40 are generated through the use of a position / speed observer 93 that includes conventional accelerometers that they are positioned relative to actuator 40 so that they can provide X and Y acceleration information to controller 91. In that embodiment, the position and velocity in terms of X and Y are obtained by the position / velocity controller 91 using low-pass filters that are, for practical purposes, at frequencies above 170 Hz, integrating. Using low-pass filters instead of pure integrators eliminates the drift and variance build-up problems of open-loop integrators. Exemplary function blocks for implementing such a position controller are provided in FIG. 10A.

IS 2 255 271 T3

Referring to FIG. 10A, the acceleration signals from the accelerometers X (d_2_x_dt_2) and Y (d_2_y_dt_2) of the position / velocity observer 93 are applied as inputs to the functional blocks designated 100X and 100Y. When using a digital controller 91, any analog acceleration signals should be converted to digital values, and the digital values ​​should be applied as inputs to controller 91.

The functional blocks 100X and 100Y respectively include conventional mathematical transformations that transform the acceleration signals into estimates of the velocity of the moving element in terms of X and Y (eg, X_dot_est and Y_dot_est). The X and Y velocity estimates are then applied, respectively, to functional blocks 101X and 101Y that transform the velocity estimates into position / velocity estimates. The mathematical nature of the transformation is illustrated in Figure 10A. The X and Y position and velocity estimates are applied respectively as inputs to gain blocks 102X and 102Y and scaling blocks 103X and 103Y that adjust the estimate signals. The position and speed signals are then subtracted from the X and Y position and speed command signals in functional blocks 104x and 104Y to produce X and Y position and speed error signals. Functional blocks 105X and 105Y combine the X and Y error signals to produce FX and FY signals that are modified by a low pass filter and gain multiplier combinations 106X and 106Y to produce FX and FY commands that are used to control the system as described above.

The exemplary position / speed controller of FIG. 10A can be used in systems in which the acceleration information from the position / speed observer 93 is continuously sampled and the FX and FY commands are continuously generated. Alternative embodiments are envisioned in which the operation of the controller 91 is not constant, but rather operates on sampled information. Function blocks for such a sampling controller 91 are illustrated in FIG. 10B. In general, the controller 91 of Figure 10B is similar to that of Figure 10A except that the mathematical nature of the filters and transformations has been adjusted to reflect the sampling nature of the controller.

In the examples of Figures 10A and 10B, the position / velocity observer 93 comprises X and Y accelerometers, and the position and velocity estimates for the actuator movable member 40 are obtained from the outputs of the accelerometers. Alternative embodiments are envisioned in which conventional position sensors (eg proximity sensors) are used to detect the X and Y position directions. In such alternative embodiments, velocity information can be obtained using a reduced order observer, such as a standard deterministic observer, an integral error observer, or a Kalman filter observer. In general, any type of full-order or reduced-order position / velocity observer 93 that can provide position and velocity information on the movable element of actuator 40 can be used to construct a position / velocity observer 93. In alternative embodiments, pickup coils with a position observer could be used. Pickup coils measure speed. In one embodiment, the position / speed observer 93 can be removed, and the position and speed information about the actuator moving member 40 can be obtained from the electrical characteristics of the actuator 40 itself. phase of actuator 40 are energized so that, at all times or at certain instants, at least one of the phase coils is not energized, the unexcited coil can be used as a proximity sensor. In such an embodiment, the unexcited phase coil can be driven with high intensity flux pulses and the resulting current can be detected. The magnitude of that current will have a simple algebraic relationship to the magnitude of the flow gap. That magnitude of the air gap, when determined, will be totally deterministic of the X or Y position.

Unfortunately, when phase coils are required to provide a force vector having a direction between +30 degrees and +120 degrees (see Figure 7A), only the X position can be determined from the unexcited coil. In such cases, the X position can be determined from actual or estimated current and flow information. The relationship between estimated flow and current is, again, simple and algebraic. In systems having four phase coils, the position of the moving element could be fully determined by analyzing the electrical characteristics of the unexcited coils.

Since the force-to-flow transformation used in the position / speed control system of Figure 9 eliminates non-linearity in the force production mechanism, the error signals generated and used by the position / speed controller 91 have dynamic Linear and the rest of the control system can be designed using linear methods. Any multiple input or multiple output linear control design method can be used to tune the system and thereby establish the system response, such as pole placement, LQG, robust, and so on. This ability to instantly control the position and speed of the moving element is very beneficial in vibration welding applications.

In one embodiment, the pole placement is used to adjust the control system in which the four control gains place the closed-loop poles, resulting in a proportional control system. Although there will be a finite steady state error and phase error between the X and Y commands when using this form of pole placement, the orbit shape of the moving element will not be affected.

Conventional vibration welding systems use an average-based control system in which the moving element of the system is controlled to move either in a circle or in an ellipse and the mean radius of the circle or ellipse is controlled. When asymmetric welding tools are used or moving element disturbances are encountered, such averaging control systems cannot compensate quickly enough for this.

ES 2 255 271 T3 tabulate the system. In contrast, the novel control system disclosed in this document instantly controls the X position, Y position, X speed and Y speed of the movable element. Furthermore, using instantaneous control, the coupling of the X and Y movement is rejected due to asymmetries of the tools and load disturbances.

When position / velocity commands are provided in a stationary reference frame (e.g. X, Y), a slower average radius control loop can be placed around the instantaneous control loop to ensure zero steady-state error . Such an external control loop is especially beneficial in ensuring that the moving element follows the desired trajectory, even under overload conditions. In general, the average radio control loop produces an average radio error that is provided as input to a PI control law controller. The PI control law will then eliminate any steady state error, although there will still be a phase error between the X and Y commands. Since the average radius will be a DC variable, the PI control law controller will reject constant load disturbances. The output of the PI control law controller is then added to the radio command signal to increase the total radio command signals to the proportional instantaneous command signal.

In the illustration of Figure 9, the position / speed input commands are provided in terms of the desired position and speed of the actuator moving member 40 in terms of X and Y position and speed. In many vibration welding applications, the desired movement of the movable element will not correspond to the arbitrary X and Y commands, but will instead correspond to X and Y commands that will tend to produce an elliptical motion of the movable element of the actuator 40. For applications of this type, the X and Y position / speed commands can be generated with a novel position / speed command generator that generates the required X and Y position and speed commands to produce the desired elliptical path. Such a position / speed command generator would have its outputs coupled to the input of the position / speed controller 91.

FIG. 11A generally illustrates an exemplary set of function blocks that may be used to implement an exemplary position / speed command generator 110. With reference to the figure, the position / speed command generator 110 receives command signals at its inputs that define a desired elliptical orbit for the moving element of the actuator 40. The command signals received by the position / speed command generator 110 are particularly: (i) a major axis command (command_r) corresponding to the major axis of the desired elliptical orbit along the X axis; (ii) a major axis to minor axis relationship command or eccentricity command (command_exc) that defines the relationship of the major axis along the X axis to the minor axis along the Y axis; (iii) an angle command (command_angle) that defines the angle of angular displacement of the major axis of the desired path from the X axis; and (iv) an operating frequency command (cf) that defines the frequency at which the moving element traverses the desired elliptical path.

Referring to Figure 11A, the various input commands described are filtered using the appropriate filters 111, 112, 113 and 114 shown in the figure. If an outer radio control loop is used as described above, the filtered radio command (r_filtered) can be modified using a radio error correction value on an add connection 115. Then the modified radius command (r_error_total) is limited to falling within certain limits in a limiting block 116 and the limited r command is applied to an elliptical transform block. Then, through functional blocks 118, 119a and 119b, the input operating frequency command is converted to time varying angular values ​​that are applied as inputs in elliptical transform block 117. The other filtered input signals (filtered_exc, filtered_angle) are applied directly to elliptic transform block 117. Elliptical transform block 117 receives the signals referenced above and generates the X and Y velocity and position commands (cmd_X, cmd_X_dot, cmd_Y, and cmd_Y_dot) using the mathematical relationships set forth in Figure 11A.

Figure 11A illustrates functional blocks that can be used to implement a position / speed command generator 110 that operates continuously in response to continuous input signals. FIG. 11B illustrates a similar position / speed command generator 110 'which is sampled to produce the appropriate position / speed commands in response to sampled input signals. As those skilled in the art who have the benefit of this specification will appreciate, position / speed command generator 110 is substantially identical to position / speed command generator 110 'with the exceptions being that filters 111', 112 ', 113 'and 114' of generator 110 'are different from those of filters 111, 112, 113 and 114.

Those skilled in the art having the benefit of this disclosure will appreciate that the position / speed control system of Figure 9 could have position / speed commands other than those described above. In general, any suitable position / velocity command signal can be expanded in Fourier series to describe any periodic waveform that, at its limit, can be described as a rectangular wave. For example, the input X and Y velocity and position commands could be generated as a function of time according to the following relationships: cmd (t) _X = r * cos (omega * t); cmd (t) _X_dot = -r * omega * sin (omega * t); cmd (t) _Y = r * exc * sin (omega * t) and cmd_Y_dot = r * exc * omega * cos (omega * t); where r is the major axis, exc is the desired eccentricity, omega corresponds to the desired operating frequency and t is time.

Furthermore, the desired elliptical path described above could be made to match a "superelipse" defined by ((X / A^TOn + (Y / B)^TOn) = r, where n is an integer> = 2. Such a superelipse path can be traversed with an angular velocity of omega with Fourier series of command reference signals.

IS 2 255 271 T3

The position / speed control system would then attempt to track the position / speed waveforms to the limit of the actuator capabilities and bandwidth.

Figures 12A-12D illustrate the operation of a position / speed controller of the type described above in connection with Figure 9. Each figure illustrates the X and Y movement of the actuator moving element 40, the FX and FY commands, the flux and flux values ​​of the three phase coils A, B and C. Figure 12A illustrates the operation of the system for linear motion along the X axis in a no-load state and Figure 12B illustrates the operation of the system for the same type of motion under a fully loaded state. Similarly, Figures 12C and 12D illustrate the operation of the system for circular motion in no-load and fully-loaded states, respectively.

As figures 12A-12D make clear, the flow established in actuator 40 by a flow controller constructed as described herein will be discontinuous and will rigorously follow flow commands, for both no-load and full-load states.

The position / speed control systems described above are all based on a stationary reference system (eg X and Y). Alternative embodiments are envisioned in which the reference system used for control purposes is based on rotary coordinates. The use of such a rotating reference frame, for a given performance objective, can reduce the required sample rate and provide improved phase and amplitude tracking. Furthermore, the use of such a reference system can minimize and / or guarantee a steady-state zero error through the use of a PI control law controller, since, for a reference system of this type, the variables of control using such a “rotary” control system are DC quantities.

According to an embodiment of the present invention, a rotary position / speed controller can be constructed that is based on a rotary reference system QD in which the movement of the moving element is defined, not in terms of position and speed X and Y, but in terms of the rotating vectors Q and D each having a corresponding vector whose elements are quantities PC in steady state. Such a position / speed control system QD operates in a manner similar to that described above in connection with the XY control system with the general exceptions described below.

When using a rotary position / speed controller, both the position / speed commands to the position / speed controller and feedback signals from the position / speed observer 93 reflecting the position / speed can be provided in terms of the QD reference. speed of the moving element of the actuator. Alternatively, the position / velocity commands and feedback signals may be provided in terms of XY position and velocity, in which case the signals and position / velocity commands must be transformed from the XY reference frame to the QD reference frame. These transformations can be carried out using a suitably programmed digital processor.

Figure 13 generally illustrates one form of a constructed QD control system 130 that is useful in helping to understand the present invention. In general, the QD control system includes a QD controller 132 that receives at its inputs position command signals that define the desired movement of the actuator moving member 40 in a predefined QD reference frame. Particularly, in the illustrated embodiment, the input QD position commands are: (i) cmd_q, which defines the desired instantaneous magnitude and the sign of the vector Q; (ii) cmd_d, which defines the desired instantaneous magnitude and the sign of the desired vector D; (iii) cmd_dot_q, which defines the desired rate of change in the magnitude of the vector Q; (iv) ref_dot_d, which defines the desired rate of change in the magnitude of vector D; and (v) cf, which defines the desired operating frequency of the system. In general, Q and D are quasi-static variables that can be> 0 or <0.

The exemplary controller 130 of FIG. 13 is adapted for use in an orbital welding apparatus and as such is specially adapted to control the elliptical movement of the actuator moving member 40. Thus, QD control operations are carried out in a rotating elliptical QD reference system. To define the proper rotary elliptical QD reference frame, the 130 QD controller requires two additional variables. The first variable (command_exc) like the command_exc discussed above, defines the relationship between the major axis of the ellipse along the X axis and the minor axis along the Y axis. The second variable (delta) defines the angular extent in where the major axis of the rotating elliptical system is offset from the stationary X axis.

Other inputs of the QD controller 132 include feedback signals that provide information about the position / velocity of the actuator moving member 40. In general, these signals are provided by a position / velocity observer 93 who, like the position observer 93 / speed mentioned above, includes an X and Y accelerometers. The X and Y position / velocity information from the position / velocity observer 93 is converted into signals q, d, q_dot and d_dot corresponding to the QD parameters of the movable element in the appropriate rotating elliptical QD reference frame. This transformation is achieved by an XY to QD transformation operation, reflected in a block 131. The position / speed information QD from block 131 is provided to the QD controller 132 as another set of inputs. The QD controller then compares the QD information for the moving element with the command information QD to generate error signals QD that are used to generate desired Q and D force commands, FQ and FD. These force commands are then transformed into corresponding force commands in the reference frame FX and FY by means of a transformation operation from QD to XY

ES 2 255 271 T3 reflected in a block 133. The signals FX and FY are then applied to a force control system 50 which controls the forces applied to the actuator 40 in the manner previously described in relation to the force control system of figure 5.

The QD controller 132 of FIG. 13 and blocks 131 and 133 can be implemented using a programmed digital processor. Figures 14A-14D generally illustrate functional blocks that may be used to implement an exemplary QD controller 132 and blocks 131 and 133.The illustrated controller operates continuously although those skilled in the art who have the benefit of this disclosure will appreciate that a QD controller Sampling can also be constructed from the teachings provided herein. Referring to Figure 14A, an overview of QD controller 132 and blocks 131 and 133 is provided. In the illustrated embodiment, the QD controller shown receives X and Y acceleration information from a position / position observer 93. speed comprising X and Y accelerometers. Functional blocks 100X, 100Y, 101X and 101Y process the X and Y acceleration information to generate X_est, X_dot_est, Y_est and Y_dot_est signals in the manner described above in connection with FIG. 10A. Next, the estimated XY velocity and position information is provided as an input to a transform functional block 140 that transforms the XY information from functional blocks 101X and 101Y into the appropriate rotating elliptical QD reference frame. FIG. 14C provides a more detailed illustration of the operation of the functional block 140.

Alternative embodiments are envisioned in which X, X_dot, Y and Y_dot are measured directly, estimated as described above or estimated from either X, Y, or X_dot, Y_dot.

Additionally, aX, aY could be directly transformed into aQ, aD and an observer could be used for Q, Q_dot, D, D_dot in the rotating frame of reference.

Referring to FIG. 14C, functional block 140 comprises three sub-functional blocks 141, 142 and 143. These functional blocks transform the XY information from the position / velocity observer 93 into QD information from the desired rotating elliptical reference system. Initially, the signals X_est, X_dot_est, Y_est, and Y_dot_est are provided to functional block 141 which transforms the XY signals into corresponding signals in an XY reference frame that rotates with respect to the standard reference frame by an amount corresponding to the value of the delta variable described above. The outputs of functional block 141 thus constitute XY position and velocity data in an XY reference frame that is offset relative to the stationary XY reference frame by an amount defined by the input variable delta.

Delta-adjusted XY information from functional block 141 is applied as input to functional block 142, which receives the delta-adjusted XY signals and transforms those signals to an XY reference frame that has been adjusted to accommodate the eccentricity of the reference frame. desired elliptical. Generally, the functional block 142 adjusts the delta-adjusted XY signals for the differences between the major X axis and the major Y axis of the ellipse. As reflected in FIG. 14C, one input of functional block 142 is command_exc.

The signals set for delta and for eccentricity from function block 142 (X_c, X_dot_c, Y_c, and Y_dot_c) are applied as inputs to function block 143. In general, the functional block 143 transforms its inputs (in terms of a stationary XY reference frame) into QD signals in terms of a rotating circular reference frame. However, since the inputs to XY circular QD functional block 143 are XY signals adjusted for delta and eccentricity, the signals q_est, d_est, q_dot_est, and d_dot_est of functional block 143 will be equivalent to signals in the elliptical QD reference frame. rotary. As reflected in Figure 14C, the cf command is used in functional block 143 as a "theta" variable. The variable "theta" is defined as 2 * pi * cf * t, where t is time.

In certain implementations of the illustrated controller, the specific order in which the transformations of Figure 14C are performed is believed to be especially important because it ensures that the proper adjustments are made to accurately transform the input XY commands into QD commands that correspond to the appropriate rotating elliptical reference system.

Referring again to FIG. 14A, the q_est, d_est, q_dot_est, and d_dot_est signals of functional block 143 are applied to a group of functional blocks indicated globally as 144. The functional blocks 144 compare the estimated QD signals, which reflect the position / speed of the moving element of the actuator 40, with the command signals QD to generate error signals Qd. Functional blocks 144 may also implement any suitable control law, such as a multiple input, multiple output PI control law. The error signals QD from the functional blocks 144 are suitably summed by an array addition block 145 to obtain the force command signals FQ and FD. Examples of variables for addition matrix 145 are provided in Figure 14B.

In the illustration of FIG. 14A, low-pass filters 146a and 146b low-pass filter the force command signals FQ and FD. Function block 147 then transforms the filtered FQ and FD signals into the corresponding FX and FY commands. A general illustration of the subfunctional blocks that can be used to implement functional block 147 is provided in Figure 14D.

IS 2 255 271 T3

Referring to FIG. 14D, the filtered FQ and FD signals are first transformed into the corresponding XY force command signals FX_circ and FY_circ. In the illustrated embodiment, the transformation performed by functional block 148 is a simple circular transformation from QD to XY that does not take into account the elliptical nature of the QD reference frame used by the controller or the delta offset of that reference frame. from the stationary XY reference system.

Then the commands FX_circ and FY_circ from function block 148 are applied as inputs to function block 149. Function block 149 also receives as input the command_exc. In general, functional block 149 transforms the FX_circ and FY_circ commands into FX and FY (FX_elip, FY_elip) commands that are adjusted to account for the elliptical nature of the rotating elliptical reference frame used by the illustrated QD controller. However, the FX and FY commands in function block 149 are not adjusted to reflect the delta offset of the rotating elliptical reference frame used by the QD controller from the stationary XY reference frame. This adjustment is made by function block 150 which receives the signals from function block 149 and a signal corresponding to the delta variable and, in response to those inputs, generates command signals FX and FY. As with the transformations reflected in Figure 14C, the transformations of Figure 14D should be carried out in the special order described above to ensure proper transformation of the signals.

Referring again to FIG. 14A, gain amplifiers then amplify these FX and FY signals from functional block 150 and apply them to the force controller to control the forces applied to the actuator moving member 40 as described above.

Although the various transformation operations described above for: (i) converting feedback XY signals to a rotating elliptical QD reference frame; (ii) implement a control law function in the rotating elliptical QD reference frame to generate QD force commands; and (iii) transforming the force commands of the rotating QD reference frame into XY force commands, may seem complicated, provide several advantages. Particularly because of the transformations described above, the control variables actually used in the functional blocks 144 to implement the system control law are DC quantities. As such, a multiple-input, multiple-output PI, or other zero steady-state error control law can be used. The use of such a control law can result in better system performance than is available from conventional control approaches. This is particularly true in orbital welding applications where the use of the described transformations is novel. Specifically, the use of the transformations and the control law described above eliminates the need for an external radio control loop since the control functions that a loop of this type would execute are handled by the control law implemented by the blocks. 144, 145 and 146 functional.

Although the invention has been described in connection with the illustrative embodiments discussed above, those skilled in the art will appreciate that many variations can be made without departing from the present invention. For example, the novel actuator 40 is described herein primarily in systems using a flow control system. It will be appreciated that the actuator 40 can be driven by a number of different apparatus, eg, a conventional inverter, which would produce sinusoidal flux in the E-cores of the actuator 40. Although such an application of the novel actuator 40 would not potentially be as elegant or perform the same as an application in which the actuator is driven by the novel flow controller described herein, it would still provide many of the advantages as a consequence of the decoupled nature of the E-nuclei and the phase coils associated with those nuclei. In light of variations of the type described above, it will be understood that the above description is made by way of example and not by way of limitation.