Method of operational control of an inertial centre of a mobile

Title of the invention

Functional control method of an inertial unit of a mobile.

Field of the invention

The present invention relates to the field of functional control of an inertial unit of a mobile on board a movement simulator.

Prior art

In general, a vehicle such as an airplane, a rocket or another type of mobile is equipped with a navigation and piloting system comprising an on-board computer, an inertial unit and piloting means for providing a spatial position, control and guidance of the mobile.

More particularly, the piloting means (for example aerodynamic control surfaces) will direct the mobile towards a designated destination or along a determined trajectory thanks to the commands received from the on-board computer which itself receives information of an inertial nature from the inertial unit. Indeed, from this inertial information, the on-board computer will estimate the spatial position of the mobile, and as a function of this, it will give orders or commands to the piloting means so that the mobile continues to operate. route to its destination.

However, if the inertial unit has a fault, the computer will make its estimates of the spatial position of the mobile from the inaccurate inertial information. The on-board computer will therefore make errors at each calculation step and send erroneous commands to the control means. Consequently, the real trajectory of the mobile will be very different from the trajectory estimated by the on-board computer. Thus, if the inertial unit has a serious fault, the orders sent by the on-board computer will be so inappropriate that the moving part risks being destabilized.

Thus, in order to control the functionality of inertial units, motion simulators are often used. More particularly, hybrid simulations are used implementing real subsets (for example, inertial unit, movement simulator, on-board computer of the mobile, certain elements of the mobile, etc.) and mathematical models for other subsets ( for example, other elements of the mobile, atmosphere, etc.).

Currently, in hybrid simulations using inertial units, two cases may be encountered concerning the information coming from the accelerometers of the inertial unit.

According to the first case, the information coming from the accelerometers is not used and is replaced by information coming from a mathematical model. This is due to the fact that the motion simulator does not allow the translation movements to be reproduced. Thus, the information from the accelerometers of the inertial unit is incomplete and does not include any information concerning the linear displacement of the mobile. Consequently, the on-board computer of the mobile cannot take into account information coming from the accelerometers of the inertial unit to determine the location, control and guidance of the mobile. Any anomalies potentially present in this information are therefore not detected by the hybrid simulation. In other words, the contribution of the accelerometers of the real inertial unit is absent, which does not make it possible to detect the slightest potential fault on the accelerometers.

According to the second case, the information coming from the accelerometers is supplemented by information representative of the translational movements, calculated by a mathematical model. However, the information coming from the accelerometers of the inertial unit is measured at a fixed point corresponding to the coordinates of the simulation laboratory. Thus, this information is not entirely representative of the information that these accelerometers would provide, for the same physical origin, during the course of the trajectory of the mobile around the terrestrial globe. For example, the gravity exerted at the fixed point of the laboratory is invariable, while that felt by the accelerometers of the inertial unit on board the mobile moving around the terrestrial globe is variable according to the altitude and the latitude. Thus, this difference distorts the trajectography of the mobile obtained by using the inertial unit at the laboratory fixed point, and makes it difficult to interpret the results. As a result, this method only makes it possible to detect a gross defect in one or more accelerometers of the inertial unit.

Moreover, the information coming from the gyrometers of the inertial unit are not entirely representative of what they are during the actual movement of the mobile around the terrestrial globe.

Indeed, the decomposition of the earth's rotation on the axes of the gyros is different depending on whether the inertial unit is located at a point of fixed coordinates (case of hybrid simulation in the laboratory) or whether it is on board a mobile moving around. of the terrestrial globe. The impact of this incomplete representativeness is such that it complicates the analysis of the results obtained in hybrid simulation.

In addition, the motion simulator must have angular dynamic performance greater than or equal to the need throughout the course of the trajectory of the mobile, which may require a very sophisticated and very expensive motion simulator.

Purpose and summary of the invention

The present invention relates to a method of functional control of an inertial unit of a mobile on board a movement simulator and providing in real time, inertial measurement data representative of a movement applied by said movement simulator, said movement simulator. process comprising:

-a theoretical modeling of the inertial unit on board the motion simulator, said theoretical modeling being fed in real time by kinematic data actually executed by the motion simulator and providing in real time theoretical inertial data representative of said supposed inertial measurement data be measured by the inertial unit on board the motion simulator,

a simulation modeling comprising a modeling of the inertial unit in a real navigation environment, the simulation modeling being fed in real time by piloting commands and providing, in real time, inertial simulation data representative of the output data of the inertial unit in said real navigation environment, said simulation modeling taking into account said piloting commands to unwind a trajectory of the mobile implementing the inertial unit,

-calculation of said control commands as a function of the inertial measurement data, the inertial simulation data and the theoretical inertial data, and

validation of the inertial unit, by comparing the trajectory of the mobile obtained by implementing the inertial unit, with a predetermined reference trajectory.

The method according to the invention makes it possible to develop a trajectography of the mobile very close to the reality not tainted with the problem linked to the measurements taken at a fixed point and thus making it possible to check whether the inertial unit has characteristics in accordance with the need then a gain in quality and cost.

Advantageously, the inertial measurement data comprise accelerometric information originating from the accelerometers of said inertial unit, and in that the theoretical inertial data comprise theoretical accelerometric information.

Thus, it is possible to detect any faults originating from the accelerometers of the inertial unit.

Advantageously again, the inertial measurement data further comprises gyrometric information originating from the gyrometers of said inertial unit, the theoretical inertial data comprises theoretical gyrometric information, and the simulation inertial data comprises simulation gyrometric information and accelerometric information from simulation.

Thus, the accelerometric and gyrometric information used to calculate the piloting commands are representative of those supplied by the inertial unit on board by the mobile during the course of its trajectory around the terrestrial globe. This is due to the fact that this accelerometric and gyrometric information includes accelerometric and gyrometric information coming from the inertial unit and a complement calculated in real time by the theoretical and simulation models.

According to a particular embodiment of the present invention, said control commands are calculated as a function of the inertial data I defined by a sum between the inertial measurement data R and the inertial simulation data T2 minus the theoretical inertial data T1 (i.e. i.e., I = T2 -Tl + R).

Thus, it is possible to use an undersized motion simulator for transients with high trajectory dynamics. Indeed, by virtue of the principle I = T2 -Tl + R, as R and Tl depend on the accomplishment performed by the motion simulator, they remain consistent, even if the latter does not correctly execute the command. This makes it possible to carry out a precise trajectory of the mobile at a lower cost.

The inertial unit can be considered valid when a differential between the inertial data I and the inertial simulation data T2 is bounded by a predetermined threshold value (i.e., | I-T2 | <ε).

Thus, it is possible to ensure the validity of the inertial unit efficiently and with very high precision.

The method according to the invention further comprises phase advance modeling fed in real time by kinematic input commands from the simulation modeling and providing output kinematic commands to the motion simulator to compensate for a delay of execution inherent to said motion simulator.

Thus, it is possible to compensate for the execution delays, which allows all the quantities to remain in phase in order to guarantee the representativeness of the implementation of the inertial unit and of any other sensors simultaneously on board the movement simulator.

Advantageously, said kinematic input commands have a profile synchronous with that of the kinematic data, and in that the amplitudes of said kinematic data and said kinematic input commands CO are consistent.

Thus, the kinematic data are in agreement with the movements actually executed by the movement simulator making it possible to obtain stimulation of the real inertial unit, in phase with the kinematics of the mobile. This guarantees the representativeness of the implementation of the inertial unit and ensures synchronism with respect to the kinematics of the moving body, for other sensors that can be onboard simultaneously with the inertial unit, on the movement simulator.

Advantageously, the method according to the invention further comprises a masking modeling downstream of the simulation modeling and upstream of the phase advance modeling so that the masking modeling is fed in real time by the kinematic commands. inputs and providing masked kinematic controls to the phase advance modeling to mask at least part of the phases of motion.

Thus, it is possible to achieve a trajectory for which the angular movement of the mobile is greater than that authorized by the movement simulator.

According to one feature, at least part of the masked kinematic commands depends on a law internal to the masking modeling and which is independent of the kinematic input commands.

This makes it possible to develop a trajectory where the mobile can carry out several loops or circuits with a movement simulator having a limited angular movement.

The invention also relates to a functional control system of an inertial unit of a mobile on board a movement simulator and providing in real time, inertial measurement data representative of a movement applied by said movement simulator, said system. comprising:

a theoretical model comprising a modeling of the inertial unit on board the motion simulator, the theoretical model being fed in real time by kinematic data actually executed by the motion simulator and providing in real time theoretical inertial data representative of said inertial data of measurements supposed to be measured by the inertial unit on board the motion simulator,

a simulation model comprising a modeling of the inertial unit in a real navigation environment, the simulation model being fed in real time by piloting commands and providing, in real time, inertial simulation data representative of the output data of the inertial unit in said real navigation environment, said simulation model taking into account said piloting commands to unwind a trajectory of the mobile implementing the inertial unit,

- calculation means for calculating said control commands as a function of the inertial measurement data, the inertial simulation data and the theoretical inertial data, and

validation means to ensure the validity of the inertial unit, by comparing the trajectory of the mobile obtained by using the inertial unit, with a predetermined reference trajectory.

The calculation means are intended to determine said control commands as a function of the inertial data I defined by a sum between the inertial measurement data R and the inertial simulation data T2 minus the theoretical inertial data T1.

The validation means are intended to consider the inertial unit as being valid when a differential between the inertial data and the simulation inertial data is bounded by a predetermined threshold value.

The system further includes a phase advance model fed in real time by kinematic input commands from the simulation model and providing output kinematic commands to the motion simulator to compensate for an execution delay inherent in said motion simulator. movement.

The system further comprises a masking model interposed between said simulation model and said phase advance model so that the masking model is fed in real time by the input kinematic commands and providing masked kinematic commands to the model. phase advance to mask at least part of the phases of the movement.

The invention is also aimed at a computer program comprising code instructions for executing the steps of the control method according to at least one of the above characteristics, when it is executed on a computer.

Brief description of the drawings

Other features and advantages of the device and of the method according to the invention will emerge better on reading the description given below, by way of indication but not limiting, with reference to the appended drawings in which:

FIG. 1 schematically illustrates a functional control system of an inertial unit of a mobile, according to the invention;

FIG. 2 very schematically illustrates an example of a functional control system according to FIG. 1;

FIGS. 3 to 5 illustrate different embodiments according to the invention; and

FIG. 6 schematically illustrates an example of an unfolding of a trajectory of a mobile.

Detailed description of embodiments

FIG. 1 schematically illustrates a functional control system 1 of an inertial unit of a mobile. This system 1 comprises a movement simulator 3 connected to a computer device or computer 5 used for the execution of the instruction codes of a computer program designed to implement the method according to the invention.

The motion simulator 3 can perform angular movements around a roll axis A1, a pitch axis A2 and a yaw axis A3. Thus, by receiving an inertial unit 7, the movement simulator 3 can apply angular movements to the latter along the roll axes A1, pitch A2 and yaw A3. It will be noted that the inertial unit 7 can be on board alone on the movement simulator 3 or when it is included in the mobile (not shown), or when it is included in at least part of the mobile.

The functional control system 1 further comprises a computer 9 on board the mobile which is connected to the inertial unit 7 and to the computer 5.

It will be noted that its various connections between the computer 5, the movement simulator 3, the on-board computer 9, and the inertial unit 7 can be made via electrical or optical cables, by radio or by other means.

According to the invention, FIG. 2 very schematically illustrates an example of a functional control system 1 of an inertial unit 7 of a mobile (not shown). It will be noted that FIG. 2 is also an illustration of the main steps of the control method according to the invention.

This method or system is implemented in a real hybrid simulation in closed loop using real subassemblies (inertial unit 7, motion simulator 3, computer 9 onboard the mobile, and possibly at least part of the mobile) and models digital other sub-assemblies (for example, propulsion of the mobile) and the environment (atmosphere). The outputs from the real subsets are the inputs of the digital models, and the outputs of the digital models are the inputs of the real subsets.

The inertial unit 7 (alone or included in at least part of the mobile) is on board the movement simulator 3 and provides in real time, inertial measurement data (that is to say real inertial data) R representative a movement applied by the movement simulator 3 and which reproduces the angular movements of the mobile in terms of travel, speed and acceleration. These inertial measurement data R come from gyrometers (or gyroscopes) and accelerometers (not shown) of the inertial unit 7 and generally allow the computer 9 on board the mobile to carry out its location, control and guidance.

This control system 1 further comprises a theoretical model (or modeling) M1, a simulation model (or modeling) M2, calculation means 11 and validation means 13. The theoretical model M1 and the simulation model M2 can be included in the computer device 5 of FIG. 1. Furthermore, the calculation means 11 comprise the computer 9 on board the mobile.

The hybrid simulation takes place in real time. Thus, the calculations and the data exchanges between the on-board computer 9 of the mobile and the theoretical model M1, the simulation model M2 and the inertial unit 7 are carried out at the real frequency, timing the operations carried out by the on-board computer 9 of the mobile, and advantageously within a period of time less than the period corresponding to this frequency, and this for each calculation loop.

The theoretical model M1 models the inertial unit 7 on board the motion simulator 3. This theoretical model M1 is supplied in real time from the motion simulator 3 by kinematic data DI (positions, speeds, and angular accelerations) actually executed by this motion simulator 3. Consequently, the theoretical model M1 supplies the calculation means 11, in real time, with theoretical inertial data T1 representative of the inertial measurement data R supposed to be measured by the inertial unit 7 on board the movement simulator 3.

The simulation model M2 comprises a modeling of the inertial unit 7 in a real navigation environment. The simulation model M2 is fed in real time by data or control commands D2 from the calculation means 11. In return, the simulation model M2 provides, in real time, inertial simulation data T2 representative of the output data of the inertial unit 7 in a real navigation environment.

Thus, the simulation model M2 takes into account the control commands D2 to unwind a trajectory of the mobile implementing the inertial unit 7.

The calculation means 11 are intended to calculate the control commands D2 as a function of the inertial measurement data R, the inertial simulation data T2 and the theoretical inertial data T1.

Then, the validation means 13 can ensure the validity of the inertial unit 7, by comparing the trajectory of the mobile obtained by implementing the real inertial unit 7, with a predetermined reference trajectory.

In a known manner, the predetermined reference trajectory can be unwound from an entirely digital simulation, for example by the simulation model M2 all by itself comprising a modeling of all the real equipment including the on-board computer and the inertial unit.

The predetermined reference trajectory is obtained in a known manner by a fully digital simulation model where all the real equipment including the computer and the inertial unit are digitized.

Thus, the validation means 13 compare in deferred time the predetermined reference trajectory produced entirely digitally, with the trajectory of the mobile, produced by the functional control system 1 implementing in real time the real inertial unit 7 and the real on-board computer 9. The comparative examination of the results determines whether the inertial unit 7 is valid, that is to say, that it has characteristics in accordance with the need or if the inertial unit is not valid. In particular, if the two trajectories are superimposed, the inertial unit 7 can be considered as perfect.

Advantageously, the inertial measurement data R comprise accelerometric information originating from the accelerometers of the inertial unit 7 in addition to the gyrometric information originating from the gyrometers of the inertial unit 7. Likewise, the theoretical inertial data Tl comprise theoretical accelerometric information and information theoretical gyrometrics.

The information coming from the accelerometers of the real inertial unit 7 participate in the location, control and guidance of the mobile, carried out by the on-board computer 9. Thus, the anomalies potentially present in this information can be detected during a comparison between the accelerometric information coming from the real inertial unit 7 and that coming from the theoretical model M1.

In addition, the inertial simulation data T2 includes simulation gyrometric information and simulation accelerometric information.

Thus, the accelerometric and gyrometric information used to calculate the piloting commands are representative of those supplied by the inertial unit on board by the mobile during the course of its trajectory around the terrestrial globe. This is due to the fact that this accelerometric and gyrometric information includes, in addition to that coming from the inertial unit located at a point of fixed coordinates, a complement of accelerometric and gyrometric information calculated in real time by the theoretical M1 and simulation models M2 .

Indeed, the information delivered by the real inertial unit 7 are measurements taken at the laboratory fixed point, and are not completely representative of those that would have been delivered by the inertial unit on board the mobile moving around the terrestrial globe.

More particularly, the theoretical inertial data T1 calculated by the theoretical modeling M1 correspond to the theoretical expression of the measurements which the real inertial unit 7 is supposed to have taken at the laboratory fixed point, and this in a very representative way, since this calculation uses the kinematic data applied to the real inertial unit 7. Furthermore, the inertial simulation data T2 are representative of the measurements made by the inertial unit 7 on board the mobile moving around the terrestrial globe.

By taking into account the inertial measurement data R and the corresponding theoretical inertial data T1, any error due to the measurements taken at the fixed point is subtracted. Consequently, by making the contribution of the inertial simulation data T2 to the pair of inertial measurement data R and theoretical T1, we obtain a trajectography representative of reality and not affected by the problem linked to the measurements made at the fixed point.

FIG. 3 schematically illustrates an embodiment of a functional control system according to FIG. 2. According to this embodiment, the calculation means 11 comprise first and second arithmetic operation means 11a and 11b performing operations arithmetic between the inertial measurement data R, the inertial simulation data T2 and the theoretical inertial data T1 to calculate inertial data I.

Thus, the first means of arithmetic operations 11a realizes the difference between the inertial simulation data T2 and the theoretical inertial data Tl. The result of this difference (T2-T1) is added to the inertial measurement data R by the second means d 'arithmetic operations 11b to express the inertial data I according to the formula 1 = T2 + R-T1.

The control commands D2 are then calculated as a function of these inertial data I defined by the sum between the inertial measurement data R and the inertial simulation data T2 minus the theoretical inertial data T1.

This embodiment is based on the use, in the hybrid simulation of the functional control system 1, of a differential obtained between the inertial measurement data R supplied by the inertial unit 7 on board the movement simulator 3 and the data theoretical inertials T1 provided by the theoretical model M1.

By virtue of the formula I = T2 -Tl + R, since R and Tl depend on the accomplishment of the motion simulator 3, they remain consistent, even if the latter does not correctly execute the command. This makes it possible to develop a precise trajectory of the mobile with a movement simulator 3 at a lower cost.

Moreover, when the validation means 13 compare in deferred time the predetermined reference trajectory with the trajectory of the mobile, produced by the functional control system 1, the validity of the inertial unit 7 is determined as a function of the differential between R and Tl .

If R = T1, then I = T2 which corresponds to the fact that the two trajectories are superimposed and then, the inertial unit 7 is perfect.

In this case, the trajectory of the mobile obtained with such an inertial unit 7 is very close to the results obtained with the only real onboard computer 9 implemented in the hybrid simulation.

If R is close to Tl, (for example, | R- Τ1 | = ε, where ε is a low value), then, we perceive a difference between the two trajectories, but we remain within acceptable bounds and the trajectory can be seen as meeting expectations. In this case, the inertial unit 7 has non-nominal characteristics, but remains within the limits or tolerances specified in its specifications.

If R is far from Tl, then the two trajectories are clearly different, the mobile does not reach the objective (true trajectory), while the on-board computer 9 believes it has reached it (estimated trajectory). In this case, the inertial unit 7 has significant defects.

In other words, the inertial unit 7 can be considered valid when a differential between the inertial data I and the inertial simulation data T2 is bounded by a predetermined threshold value ε (| I- Τ2 | <ε).

Moreover, by calculating the theoretical expression Tl of the measurements that the real inertial unit 7 at the fixed point is supposed to have carried out, by subtracting T1 from R, and by adding the contribution of the inertial simulation data T2 (representative of the measurements made by the inertial unit on board the mobile moving around the terrestrial globe), we obtain a trajectography representative of reality. Note that by subtracting Tl from R, we subtract the contribution made by the real inertial unit to within η (η = | R- Tl |), where the value of η is all the greater than the inertial unit 7 real is imperfect.

According to this example, it is possible to unwind trajectories with a simulation model M2 taking into account the different tolerances of the inertial unit 7. For example, it is possible to unwind a determined number N of trajectories by modulating the different variables in their tolerances to obtain a scalp. trajectories. If the “true” trajectory, that is to say that which is obtained with the real inertial unit 7, is included in the scalp obtained by virtue of the digital simulation, then the inertial unit 7 can be considered valid. Otherwise, it is because it has a defect.

FIG. 4 shows schematically that the functional control system 1 further comprises a model (or a modeling) of phase advance M3. This phase advance model M3 is fed in real time by data or kinematic commands of inputs C1 from the simulation model M2 and provides kinematic output commands C2 to the motion simulator 3 to compensate for an inherent execution delay. to the motion simulator 3.

In general, the navigation and piloting system of a mobile comprises a function called “pilot” which is intended to develop steering orders for the piloting means compatible with the characteristics of the mobile. This is an automatic function, judging in real time, thanks to inertial information or data and possibly of another nature (case of a complementary sensor other than inertial), of the fulfillment of the order previously ordered. and adapting the level of the order to come as needed. If the piloting means do not react quickly enough, the pilot will realize this and will give stronger orders to compensate for this delay in execution.

Advantageously, the kinematic input commands C1 have a profile synchronous with that of the kinematic data Dl. In addition, the amplitudes of the kinematic data Dl and of the kinematic input commands C1 are consistent. The stimulation of the inertial unit is then synchronous with the angular kinematics of the mobile, guaranteeing the representativeness of the implementation of the inertial unit and of any other sensors simultaneously on board the movement simulator.

FIG. 5 shows schematically that the functional control system according to FIG. 4 further comprises a masking model (or a modeling) M4 which according to this example is downstream from the simulation model M2 and upstream from the advance model phase M3. Thus, the masking model M4 is fed in real time, from the simulation model M2, by the input kinematic commands C1 and provides masked kinematic commands C3 to the phase advance model M3 to mask at least part of the inputs. phases of the movement of the mobile.

The phase advance model M3 then supplies the motion simulator 3 with output kinematic commands C4 which take into account the masked kinematic commands C3 to achieve a trajectory for which the angular deflection and / or the dynamics of the moving body is greater than that authorized by the motion simulator 3.

Thus, an angular displacement of the motion simulator 3 or a stop of this displacement, voluntarily controlled from the masking model M4, is taken into account by the inertial measurement data R and the theoretical inertial data Tl but does not affect the pilot controls D2.

Indeed, according to the example of FIG. 3 and by virtue of the formula I = T2 -Tl + R, the inertial data I are not affected by any angular displacement or stopping of the motion simulator 3 driven by the masking model M4.

Moreover, a monitoring implemented in the masking model M4 makes it possible to limit the angular movement control before reaching the hardware or software stops of the movement simulator 3.

It will be noted that the theoretical M1, simulation M2, phase advance M3, and / or masking M4 models can be included in the computer device 5 of FIG. 1.

FIG. 6 schematically shows an example of a course of a trajectory around an obstacle.

For example, by using an undersized motion simulator 3 having an angular displacement capacity around the yaw axis of +/- 100 degrees, which corresponds to an angular excursion of 200 degrees, one can unwind a trajectory exhibiting an angular excursion of 270 degrees. Thus, according to this example, the mobile 21 can travel a trajectory 23 from a starting point 25 to an ending point 27 while going around an obstacle 29 according to an angular excursion of 270 degrees.

Indeed, during the course of the trajectory 23, it is possible to modify the angular positioning of an axis of the movement simulator 3. For example, the yaw axis A3 of the movement simulator 3 can be controlled independently of the yaw. mobile 21.

Advantageously, at least part of the masked kinematic controls C3 can depend on a law internal to the masking model M3. In particular, the internal law is independent of the kinematic input commands C1. Thus, it is possible to inject into the motion simulator 3, a movement command from the masking model M4, independent of the kinematic input commands C1 from ie simulation model M2. This makes it possible to unwind a trajectory where the mobile can perform several loops with a movement simulator 3 having a limited angular movement.

In addition, the trajectory may have a greater dynamic than that authorized by the movement simulator 3.

By way of example, in the case where, for a transient event, the kinematic input commands C1 include a speed command greater than what is capable of generating the motion simulator 3, the characteristics of the kinematic data DI actually executed by the motion simulator 3 are reduced compared to the kinematic input commands C1. Thus, this altered movement is taken into account by the inertial measurement data R and the theoretical inertial data T1. However, by virtue for example of the formula I = T2-T1 + R, the inertial data I are not affected by alteration of movement. Thus, the trajectory is not affected by the undersizing of the motion simulator 3.

The invention also relates to a computer program downloadable from a communication network comprising program code instructions for the execution of the steps of the control method according to the invention when it is executed on the computer.

5. This computer program may be stored on a computer readable medium and may be executable by a microprocessor.

This program can use any programming language, and be in the form of source code, object code, or intermediate code between source code and object code, such as in a partially compiled form, or in any other. desirable shape.

The invention also relates to an information medium readable by a computer, and comprising instructions of a computer program as mentioned above.

The information carrier can be any device capable of storing the program. For example, the medium may comprise a storage means, such as a ROM, for example a CD ROM or a microelectronic circuit ROM, or else a magnetic recording means, for example a floppy disk or a disk. hard.