Self-learning control method and self-learning control system for the control of a tempering device

The invention relates to a self-learning control method for controlling a tempering device for supplying or removing heat to or from a room, object, gas or a liquid.

Such self-learning control methods are not known. Self-optimizing control systems are known for the control of space heating systems, in which, however, the user a number of details about the type of heating, the flow temperature, the heat losses at different indoor and outdoor temperature, etc. must be entered. Especially in very sluggish systems, as they are usually the usual used in the private sector house and apartment heaters, as well as when approaching changed set temperatures, these self-optimizing controller can not prevent a relatively strong overshoot of the temperature above and below the set temperature. In addition, the presetting of the control parameters in individual cases is often very complicated, and it requires many attempts until a reasonably satisfactory operation is found.

In addition, thermostats are known for the regulation of radiators for living spaces, which the heating z. B. via a valve or a circulation pump on and off, as soon as a certain setpoint temperature is exceeded or exceeded. The hysteresis between the two switching points is usually 0.5-1 ° C, with simple thermostatic valves for radiators also up to 2 ° C. The switching behavior of this thermostat controller causes the room temperature fluctuates greatly, because the heating system reacts only a long time delayed to the control signals of the controller. For example, if the thermostat switches off the heating, the still hot radiator continues to give off heat to the room for many minutes, so that the room temperature rises significantly above the setpoint temperature. If, in the opposite case, the controller switches on the heating, it takes some time for the radiators to warm up, during which the room temperature drops even further below the setpoint temperature. The design-related necessary relatively large hysteresis of the thermostat controller amplifies this overshoot and undershoot yet.

On this basis, the present invention seeks to provide a self-learning control method in which information about the type of heating or cooling, environmental conditions, in particular about temperature losses between the space to be heated, subject, gas or liquid and the colder environment or about the heat transfer from the warmer environment to the room to be cooled, object, gas or liquid are not required, and which not only ensure a very accurate maintenance of a user desired temperature desired, but also a rapid achievement of this target temperature without strong over- or Allow under-swing above or below the setpoint temperature. In addition, the method is also applicable to existing heating or cooling systems or be retrofitted without much effort.

(to 1.)

The object is achieved on the one hand by a self-learning control method for controlling a temperature control device for supplying or removing heat to or from a room, object, gas or liquid, wherein a maximum temperature setpoint (setpoint max) and a minimum setpoint temperature (setpoint) Min) and the time course of the actual temperature of the room, object, gas or liquid is measured regularly, with the following steps:<ul id="ul0001" list-style="none" compact="compact"><li>a) determining an n-th switch-on time (t-on (n)), to which the tempering device (10) is turned on,</li><li>b) determining an n-th switch-off time (t-off (n)) to which the tempering device (10) is switched off,</li><li>c) detecting the two n-th extreme values ​​(actual max (n) and actual min (n)) resulting after switching on the tempering device (10) in the actual temperature profile, in which the actual temperature is a local minimum and a has local maximum,</li><li>d) determining the first derivative of the temperature profile after the time in the n-th switch-on time (t-a (n)),</li><li>e) determining the optimum n-th switch-on and switch-off times (t-optein (n) and t-optaus (n)) from the values ​​of the n-th switch-off time (t-off (n)) and the n-th switch-on time ( t-in (n)), the two nth extreme values ​​in the course of temperature (actual max (n), actual min (n)) and the 1. Derivation of the temperature curve after the time in the n-th switch-on time (t-in (n)), where for the n-th extreme value (actual max (n)) a local maximum in the temperature profile equal to the specified maximum temperature setpoint (set max (n)) and for the n-th extreme value (actual min (n)), a local minimum in the temperature profile equal to the predetermined minimum temperature setpoint (setpoint Min (n)) is reached.</li><li>f) incrementing n by 1 and repeating steps a) to f), wherein the determined n-th switching on and off times (t-optein (n) and t-optaus (n)) as well as the 1st derivative of the temperature curve after the time in the n-th switch-on time (t-on (n)) in determining the n + 1 switch-on and switch-off (t-on (n + 1) and t-off (n + 1)).</li></ul>

The invention is thus based on the idea that the temperature of a space to be heated or cooled, object, gas or liquid will always perform oscillations around the actual desired value due to the inertia of the heating or cooling system and the environmental losses. While only one setpoint is specified for simple thermostats, if it is exceeded, the heating is switched off and if it falls below the heating is switched on, In the control method according to the invention, two setpoints are specified, which mark the maximum amplitude of the oscillations around the finally desired temperature value, the reaction of the room, Article Gas or liquid are automatically detected on a heating or cooling and taken into account at the next switching on the heating or cooling system. The method thus learns the parameters necessary for optimum regulation of the respective tempering device itself, without the user having to prescribe them. If the procedure z. B. determined that a radiator after switching off still gives off a lot of heat to the room to be heated, so that the actual temperature exceeds the predetermined maximum target temperature, the radiator is turned off earlier in the next heating cycle.

A great advantage of the method is that it can be used to control all types of tempering, such. B. Oil, gas, electric heaters, fan heaters, heating coils, heating or cooling medium flowed through by heating or cooling means, in particular also heat sinks of refrigerators, cooling and air conditioning systems, etc. Also, it does not matter to the process, whether a room or an object heat to or be removed from this: The method can be used to control tempering equipment for houses, halls, rooms, tents, interiors of vehicles, ovens, interiors of refrigerators , Refrigerated warehouses, refrigerated trailers for trucks and trains etc. can be used because it automatically determines the environmental parameters necessary for optimal control.

(to 2.)

In a particularly simple implementation of the control method, with unchanged desired values ​​Soll-Max and Soll-Min the (with respect to the temperature profile) relative switch-on time t-a (n + 1) = t-optein (n) and / or the relative switch-off time t-off (n + 1) = t-optout (n) set. Of course, it is clear that instead of the relative switch-on time, it is also possible to work with the absolute switch-on time, in which case, of course, a constant time factor is added to the values ​​t-optein (n) or t-optaus (n) must be added; In other words, the method can either be carried out so that each relative optimum on and off times are determined for the tempering, wherein it has proved to be useful to determine the switch-on of the temperature control relative to the actual temperature (ie z. B. Switch on at 20 ° C) and the switch-off time relative to the switch-on (ie z. B. Switch off after 5 min.), Or to work with absolute times (ie Z. B. Switching on at 12:05, switching off at 12:10, etc.).

(to 3.)

The stability of the control can be further increased if, in addition, the time interval between the switch-on time (t-on (n)) and the local minimum (actual -min (n)) is determined when the temperature control device (10) is used for heating , or. between the switch-on time (t-on (n)) and the local maximum (actual max (n)) when the tempering device (10) is used for cooling. This time interval is additionally taken into account when determining the n + 1th switch-off time (t-off (n + 1)). Appropriately, for this purpose, the n + 1te off time (t-off (n + 1)) is set so that a duty cycle of the temperature of at least the said temporal decency multiplied by a correction factor results.

(to 4.)

An expedient variant of the control method provides, ß after switching on the tempering at the switch-on (t-a (n)), the 1st derivative of the temperature profile is continuously determined after the time and the tempering is turned off only when<ul id="ul0002" list-style="dash" compact="compact"><li>the specified switch-off time (t-off (n)) has elapsed and</li><li>in the case of heating, a local minimum in the temperature profile (actual min (n)), or in the case of cooling, a local maximum in the temperature profile (actual max (n)) has occurred and</li><li>the 1st derivation of the temperature curve after time in the case of heating is greater or in the case of cooling is less than a specified value, and</li><li>the current actual temperature in the case of heating is greater than the minimum temperature setpoint (setpoint min.), or in the case of cooling is less than the maximum temperature setpoint (setpoint max.).</li></ul>

(to 5.)

The control method can also be advantageously simplified by omitting the determination of a switch-off time (t-off (n)), and switching off the tempering device only depends on passing a local maximum or minimum, a specific minimum slope of the temperature curve and the achievement of the minimum temperature setpoint is done.

(to 6. + 7.)

In a preferred embodiment of the method, to determine the turn-on instant t-ein (n + 1), at least the three values ​​t-optein (n), t-optein (n-1) and t-optein (n-2) are determined in magnitude arranged and <img file="EP0935181A2_D0001.tif" />median-filtered<img file="EP0935181A2_D0002.tif" />that is, the average value is selected as the value for t-in (n + 1). Such median filtering has the advantage of so-called<img file="EP0935181A2_D0003.tif" />Outliers ", but at the same time to record actual changes in environmental conditions and adjust the regulation accordingly. B. If a window is opened for a short time when the outside temperature is very cold, it may happen that the temperature maximum value Ist-Max resulting after the end of a heating cycle does not approach the desired value setpoint Max or that the actual actual temperature is approaching. Value approaches the target value, but that a temperature sensor detecting the actual temperature is so unhappy in the vicinity of the opened window that he erroneously detects too low a room temperature.

So that the short-term opening of the window does not lead to an overheating of the room in the next heating cycle, the last at least three calculated optimal switch-on and switch-off are each arranged according to their size and it will take the average of these values ​​as the next switch-on and switch-off, so that so <img file="EP0935181A2_D0004.tif" />Runaway<img file="EP0935181A2_D0005.tif" /> do not affect the regulation. On the other hand change the environmental conditions, it cools z. B. strong, so that it actually takes longer to heat to reach the desired set temperatures, these altered environmental conditions will be effective after only a few cycles, depending on the number of calculated optimal turn-on and turn-off times used for median filtering. Are z. B. in each case three switch-on and switch-off times are arranged according to size and in each case the average value is selected, changed environmental conditions already from the second cycle lead to a change to the correspondingly changed new optimum switch-on and switch-off times.

(to 8.)

Alternatively or additionally, the method can be carried out such that a limit value for the 1. Derivation of the temperature profile is given by the time, when its exceeding or falling below (ie what a sudden particularly steep rise or fall in temperature eg due to the opening of a window) the subsequently measured values ​​for determining the optimum switch-on and switch-off times until the occurrence of a predefinable condition, in particular falling below an absolute minimum temperature or the change in the temperature profile, are no longer taken into account. If the method used to control a space heater, it can be provided that the heater is turned off in case of sudden drop in temperature until a minimum in the temperature profile is reached. This way, one,<img file="EP0935181A2_D0006.tif" />To-window-out-heating "reliably prevented.

(to 9.)

The method can be advantageously applied so that the user must specify only a temperature setpoint from which the values ​​of setpoint max and setpoint min are automatically determined. If the method is used to control a heater, the user specifies, as in the known control method only his desired average temperature from which then depending on the sensitivity of the devices used to determine the actual temperature, the values ​​of target Min and setpoint Max be determined. If z. B. a temperature sensor used, which has an accuracy of 0.05 ° C, it has proved expedient to choose the distance between the setpoint Min and setpoint Max between 0.3 and 0.4 ° c, so what a control accuracy of Room temperature of +/- 0.15 ° c or +/- 0.2 ° C around the user's desired midpoint. The more sensitive the temperature measuring instruments used, the greater the control accuracy.

(to 10th)

In an expedient embodiment of the method is provided in that the setpoint values ​​Soll-Max and Isoll-Min as well as the respectively last-determined switch-on and switch-off times t-in (n) and t-out (n) are stored in a memory unit and that upon a change of the setpoint values ​​the memory unit then is checked whether on and off times have already been stored for the changed setpoint values, which are then taken over as the switch-on and switch-off times for the first passage through the control method with the new setpoint values. This embodiment allows a particularly accurate achievement of changed set temperatures already at the first run through a control cycle, without the process must first gain information about possible optimal on and off times, since they are already stored.

(to 11. +12. +13.)

An advantageous embodiment of the method provides that the determination of t-an (n) and t-off (n) by means of a central evaluation and control unit, by means of which also corresponding control commands for switching on and off of the tempering are generated. The control commands can be transmitted from the central evaluation and control unit wirelessly or wired to tempering. In the case of wireless transmission, consideration must be given, in particular, to the transmission via commercially available and now very inexpensive radio transmitter-receiver units, which preferably operate at the frequency of 433 MHz released for general applications. In the case of wired transmission, it makes sense to resort to a power grid which is generally present, as it is known, for example. B. from the so-called <img file="EP0935181A2_D0007.tif" />Baby Phones "is known.

(to 14.)

If the control method is used to control a tempering device through which a heating or cooling medium flows, wherein the temperature of the heating or cooling means is adjustable, so the process can be advantageously carried out so ß from the measured time interval between t-in (n) and t-out (n) (hereinafter called duty cycle (D (n)), and the measured time interval between the local extreme values ​​Ist-Max (n-1) and Ist-Min (n), if the tempering device is used for heating, or. the measured time interval between the local extreme values ​​actual min (n) and actual max (n) when the tempering device is used for cooling (in the following half cycle (HZ (n)) called), the required difference between heating or coolant temperature and the actual temperature (referred to below as the flow temperature (VT)) is determined and set with the following steps:<ul id="ul0003" list-style="none" compact="compact"><li>a) If the duty cycle (D (n)) is less than the half-cycle (HZ (n)) multiplied by a factor (F), the pre-charge temperature (VT) is decreased by a fraction (B) and the heating or coolant temperature accordingly changed.</li><li>b) If the duty cycle (D (n)) is greater than the half-cycle (HZ (n)) multiplied by a factor (F), the flow temperature (VT) is increased by a fraction (B) and the heating or coolant temperature accordingly changed.</li><li>c) The specified according to claims 1 and 2 off time (t-off (n + 1)) is redefined so that the duty cycle (D (n + 1)) in the case of a) extended by the fraction (B) and in Case b) is shortened by the fraction (B).</li></ul>

This embodiment of the method allows a particularly energetically favorable cooling or heating, since the coolant is not cooled unnecessarily far below the desired mean target temperature or the heating must not be heated unnecessarily far above the desired mean target temperature, because it is all the more energy unnecessary Expended, the more cooling or heating medium temperature and desired setpoint temperature differ from each other.

(to 15.)

In a suitable alternative of the control method, a half cycle (HZ (n)) is measured, the time interval between the local maxima (actual max (n-1)) and (actual max (n)) (in the following whole cycle (GZ (n) ) and, in the method described in the preceding paragraph, this whole cycle (GZ (n)) is taken into account instead of the half cycle (HZ (n)).

(To 16.)

A further expedient alternative is to determine the required difference between heating or coolant temperature and the actual temperature (in the following flow temperature (in the following flow temperature) from the continuously determined 1st derivative of the temperature profile over time in the heated or cooled space, object, gas or liquid ( VT) is determined and set, with the following steps:<ul id="ul0004" list-style="none" compact="compact"><li>a) Forming the arithmetic mean of the amounts of the determined 1st derivatives of the temperature profile over time in the period between the temperature extremes Ist-Max (n-1) and actual Min (n) (hereinafter referred to as (1)) and in the Period between the temperature extremes Actual Min (n) and Actual Max (n) (hereinafter referred to as (2)).</li><li>b) If means (1) is smaller than means (2) multiplied by a factor (F), the flow temperature (VT) is reduced by a fraction (B) and the heating or coolant temperature is changed accordingly.</li><li>c) If means (1) is greater than means (2) multiplied by a factor (F), the preset temperature (VT) is increased by a fraction (B) and the heating or coolant temperature is changed accordingly.</li><li>d) The set according to claims 1 and 2 off time (t-off (n + 1)) is redefined so that the duty cycle (D (n + 1)) in the case of b) extended by the fraction (B) and in Case c) is shortened by the fraction (B).</li></ul>

(to 17.)

Alternatively, instead of the arithmetic mean from the first derivatives of the temperature profile over time (average (1)), a fixed value can be specified.

(to 18.)

Finally, the control method can advantageously be used even if not the heating or cooling means<u>temperature</u>but instead the heating or cooling<u>power</u> can be regulated. This will be especially the case when the tempering is operated with electric current. Dedicated control method is used with the modification that instead of the difference between the heating or coolant temperature and the actual temperature (ie, the flow temperature (VT)), respectively, the heating power is increased or decreased.

(to 19.)

The determination of the heating or coolant temperature or the heating or cooling power is advantageously carried out by means of a central evaluation and control unit, by means of which also generates corresponding control commands for adjusting the heating or coolant temperature or the heating or cooling power and transmits wired or wireless to tempering can be.

Further details and advantages of the invention will become apparent from the following description taken in conjunction with the drawings. Show it:<ul id="ul0005" list-style="none"><li>Fig. 1 the basic arrangement of serving for heating a room temperature control, which is regulated by means of a control method according to the invention,</li><li>Fig. 2 the basic structure of a control device with a central evaluation and control unit,</li><li>Fig. 3 a schematic diagram of the control device of FIG. 2, but for the transmission of corresponding control signals between tempering and evaluation and control unit two transmitter-receiver units are provided and</li><li>Fig. 4 a diagram in which the course of the actual temperature against time in inventive control of a heating system, and the duration for which the heating system is turned on, are registered.</li></ul>

In the Fig. 1 is shown in its entirety with 10 designated tempering, which consists essentially of a heat source 12, for. B. an oil burner, and arranged in the room to be heated 14 radiator 16. Heat source 12 and radiator 16 are connected via lines 18 and 20 for the heating means, for. B. Water, connected, wherein the heating means via the line 18 (the so-called flow) is supplied to the radiator 16, from which it flows off again via the so-called return 20 and reaches the heat source 12 for renewed heating. In the flow 18, a valve 22 is arranged, which has only two switching states in the illustrated embodiment, namely either fully open or fully closed. The valve is controlled by means of a central evaluation and control unit 24 which generates corresponding control signals and - as indicated by the dashed line 26 - transmits to the valve 22. To detect the actual temperature, a temperature sensor 28 is arranged in the space 14, which is coupled to the central evaluation and control unit 24 and the last-mentioned transmits the measured temperature values ​​continuously.

In practice, the central evaluation and control unit 24 not only with a temperature sensor 28, but - as shown in FIG. 2 shown - even with a corresponding setpoint generator 30, z. B. a keyboard or a knob, by means of which the user of the method or the user of the control device can set a desired temperature of him. Furthermore, a memory 32 is expediently provided, to which the central evaluation and control unit can access both for storing data and for retrieving data. In such a memory z. B. Optimum switch-on and switch-off times may be stored for different, not yet set temperature setpoints. Depending on the type and configuration of the valve 22, a switching stage 34 is still connected in practice between the central evaluation and control unit 24 and valve 22, which converts the control signals generated by the central evaluation and control unit into a mechanical movement of the valve 22. It is also possible - as shown in FIG. 3 shown - between the valve-operating switching stage 34 and the central evaluation and control unit 24 to provide two transmitter-receiver units 36 and 38, by means of which generated by the central evaluation and control unit 24 and converted by the switching stage 34 control signals wireless or wired be transmitted.

In the in Fig. 3 arrangement shown is provided that the correct receipt of the respective control signals from the switching stage 34 associated transmitter-receiver unit 38 is confirmed by generating a corresponding confirmation signal, which in turn received by the transceiver unit 36 ​​and the central evaluation and Control unit 24 is checked. If this check reveals that the control signal was not received or was received only incompletely, a corresponding error message is generated and emitted visually and / or acoustically via display means known per se and therefore not further shown here. Alternatively or additionally, it is also possible to first resend the control signals in the event of a fault, since in particular when the transmission is wireless, short-term disturbances in the transmission link may occur, which does not involve intervention by the user of the method or require the user of the control device.

The self-learning control method according to the invention will be described below with reference to the diagram of FIG. 4 described in detail by the example of the control of a heater with a radiator. The method is based on the basic idea that heating and space to be heated form a system that can be stimulated to temperature oscillations. From the type of excitation and the temperature profile, which responds to a specific suggestion, information about the system and its current state can be continuously obtained. This information is sufficient to produce controlled low amplitude oscillations of temperature around a set point that is close to this setpoint. According to the invention, the radiator is periodically and in the correct phase in each case turned on so long that a temperature oscillation with defined minima and maxima arises. Minima and maxima are under very small amounts below or above a setpoint, which is thereby approximated very accurately. Deviations between the expected and the actual vibrations use the self-learning control method to correct the excitation, i. H. of the switch-on time and the switch-off time or the duty cycle of the radiator. The method thus continuously tests the behavior of the system and determines from its reaction the optimum on and off times for the radiator.

In this way, a very high control accuracy is achieved, despite the large inertia commonly encountered in heating systems. The method does not require any presettings, as it extracts all parameters from the system itself and constantly adapts them. It responds very quickly to changing environmental conditions, because the parameters are constantly checked in the steady state. Significant changes are recognized and corrected right from the start. To measure the temperature profile, a single temperature sensor is sufficient. From the tempering, so in the present example, the heater is only required that the heating circuit via a circulation pump or a valve can be turned on and off. Proportional actuators are not required. Further data of the tempering must not be known.

To realize the control principle according to the invention, the course of the room temperature must be set in relation to the switch-on time and the switch-on duration D of the radiator. Are the switch-on time and the switch-on time or Switching time known, can be determined from the measurement of the actual temperature profile in the heated room, the parameters for the solution of the relationship equations. Conversely, the solution of the equations then allows a prognosis for the future temperature profile. From the comparison between the expected and the actually measured temperature profile, changes in the ambient conditions can then be read and used to correct the parameters. These steps must be repeated cyclically (heating cycle of duration Z = distance between actual max. (N + 1) and actual max. (N)) at certain time intervals in order to allow continuous adjustment of the parameters.

The following relationships apply to small temperature fluctuations (distance A between setpoint max and setpoint Min &lt;1 ° C): The position of actual min (n) depends linearly on t-in (n) and the slope of the temperature curve of the actual temperature profile T-is off at time t-a (n). For a given slope, t-a (n) can be chosen so that a certain actual min (n) is exactly reached. The position of actual max (n) depends linearly on t-in (n), the slope of the temperature curve of the temperature profile T-actual and the duty cycle D, ie the difference between t-out (n) and t-in (n) from. For a given switch-on time and given slope, it is thus possible to select t-off (n) or the switch-on duration D in such a way that a specific actual max is exactly reached.

The actual temperature T-actual can now be regulated as follows: The setpoint values ​​given are a maximum setpoint temperature setpoint max and a minimum setpoint setpoint setpoint min. At the beginning of each heating cycle Z, the equation parameters are determined from the data of the preceding cycle. t-on (s) and t-off (s) or D are now calculated so that actual min (n) and actual max (n) coincide with setpoint min and setpoint max. The course of the temperature curve of the actual actual temperature T-Ist is then monitored. If it does not agree with the predicted course, the switch-on and switch-off times are readjusted accordingly. The parameters of the current cycle are equal to the parameters of the previous cycle, when the energy flow into the radiator, the heat dissipation of the radiator to the room and the heat emission of the room to the environment have remained unchanged. The energy flow into the radiator is essentially determined by the flow temperature of the transport medium and is either constant or changes only relatively slowly. The heat output of the radiator in the room can be assumed to be constant with small temperature changes. The heat output of the room to the environment depends essentially on the ambient temperature, which is also constant, or changes only relatively slowly. Changes in the flow temperature or ambient temperature are thus reflected in parameters that have changed slightly from cycle to cycle. As a result, changes in the environmental conditions are detected and displayed.

It is essential that setpoint Min and setpoint Max have a minimum distance A, which depends on the sensitivity of the temperature sensor used. At a still relatively easy to achieve resolution of 0.05 ° C are useful values ​​for A z. B. 0.3 or 0.4 ° C, which is a principle control accuracy of T actual +/- 0.15 ° C or +/- 0.2 ° c. With more sensitive temperature sensors, the accuracy can be further increased. The control method according to the invention thus enforces controlled temperature oscillations around the mean setpoint temperature with maxima in setpoint min and setpoint max.

If now a lower target temperature to be approached, the switching on the heater is delayed correspondingly long until the currently calculated actual-Min coincides with the new target Min.

If a higher target temperature is to be approached, a somewhat different relationship arises. For temperature increases&gt; 1 ° C namely the following applies: As long as the radiator is still in the heating phase, its energy output increases exponentially and T-Ist accordingly increases also exponentially. When the radiator has reached its final temperature, its energy output is constant and T-Ist increases linearly. When the radiator is finally switched off, its heat output decreases exponentially, and T-Ist now increases according to a negative e-function. These non-linear relationships can be neglected for small temperature increases (&lt;1 ° C), but not for larger ones. The temperature increase can therefore no longer be regarded as linearly dependent on the slope of the temperature curve T-actual and the duty cycle D. However, a good approximation is obtained by assuming the dependence as parabolic. In the relationship equation, therefore, an additional square member must be introduced. The corresponding additional parameter is determined from the first measured larger temperature jumps and then continuously leads nachge.

Only when the temperature rises by more than 6-8 ° C are this approximation calculated too short turn-on, so that the new target temperature is not achieved with only one heating cycle. Then a second cycle is necessary. As a result, the heating phase lasts a bit longer, but an overshoot of the actual temperature beyond the setpoint Max is certainly prevented.

With the method according to the invention can be advantageously determine the minimum required flow temperature (difference between heating or coolant temperature and actual temperature) from the following considerations: In the steady state, the heat output of the radiator during the duty cycle D just reached by the required temperature oscillations to the setpoint force. Since the heat output is proportional to the flow temperature and the time, the same heat is z. B. delivered at twice the duty cycle and halved flow temperature. So that the control does not become too slow, the duty cycle D should not be longer than about twice the time interval between actual max (n) and actual min (n + 1). The currently occurring duty cycle D can therefore be extended or shortened by a factor, depending on the case, and the feed flow temperature must then be increased or decreased by the same factor.

The control method described leads already after two to three heating cycles to a very accurate compliance with the target temperature. Especially with frequently changing target temperatures or strongly fluctuating environmental conditions, the accuracy can be further increased if the once found parameters are stored in a table. By comparing the currently found with the stored parameters, the probable parameters can be predicted even with larger changes and preset accordingly. If the control procedure has been operated for some time with changing setpoint temperatures and ambient conditions, then the table contains virtually all required operating parameters. Even after a long standstill, the control is then immediately set optimally.

The control method described on the example of a heater can also be used with small modifications for the control of cooling or air conditioning systems, the equations need only be adapted to control a cooling system in that the signs of some parameters are reversed. Otherwise, the same procedure applies.

In the context of the inventive concept, numerous modifications and developments are possible, the z. B. on the type of measured or relate determined parameters. Thus, instead of the said optimal switch-off of the tempering relative to the switch-on also an optimal switch-off or else - with respect to the time - absolute switch-off can be determined. In any case, it is essential to the invention that the actual temperature of a room to be tempered, Article Gas or a liquid to controlled vibrations around a mean target temperature is excited by periodically switching on and off of the tempering, wherein the actual temperature resulting in response to the switching on and off of the tempering device in the room to be tempered, Object, Gas or liquid continuously monitored and used to determine the next on and off times of the tempering.