Parameter determination device and parameter determination method

Abstract

To easily determine a design parameter suitable for a model free type controller.SOLUTION: The parameter determination device 20 includes the acquisition unit 221 that acquires a plurality of pieces of input data u to the control target P and a plurality of pieces of output data y output by the control target P in a predetermined period, and the parameter determination device 20 obtains a minimum value of a cost function by inputting the input data u and the output data y to the cost function indicating a difference between the input data u acquired by the acquisition unit 221 and a virtual input of the control system 10. The control system 10 includes the calculator 222 that calculates the control parameter of the feedback controller 13, and the determiner 223 that determines, as the design parameter used as the constant in the control system 10, one design parameter in which the difference between the signal levels of the first signal uest, the second signal uff, and the third signal ufb output by inputting the control parameter and the one design parameter to the control system 10 is included in a predetermined range.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present invention relates to a parameter determination device and a parameter determination method. [Background technology]

[0002] The parameter adjustment device of Patent Document 1 adjusts the control parameters of a feedback controller and a feedforward controller that control the input and output of a controlled object based on the output value of an evaluation function that uses a pseudo reference signal and a reference model. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-112300 Summary of the Invention [Problem to be solved by the invention]

[0004] However, a model-free controller that controls a control target based on output data of the control target and a target value of the output data requires adjustment of design parameters determined by a user in addition to control parameters of a feedback controller included in the controller. Adjusting the design parameters requires repeated testing using the model-free controller and the control target.

[0005] The present invention has been made in view of these points, and has as its object to easily determine design parameters suitable for a model-free type controller. [Means for solving the problem]

[0006] A parameter determination device according to a first aspect of the present invention includes an acquisition unit that acquires a plurality of input data for a control object and a plurality of output data output by the control object over a predetermined period of time; an estimator that estimates and outputs a first signal including a disturbance acting on the control object based on the input data and the output data; a feedforward controller that outputs a second signal for calculating a virtual input corresponding to the input data based on a target value of the output data; a feedback controller that outputs a third signal for calculating the virtual input based on a difference between the output data acquired by the acquisition unit and the target value; and a controller that calculates a sum of the third signal and a subtraction value obtained by subtracting the first signal from the second signal, the subtraction value being multiplied by a reciprocal of a predetermined design parameter, and the third signal as the virtual input. a calculation unit that calculates a control parameter of the feedback controller by inputting the input data and the output data acquired by the acquisition unit and one design parameter among the plurality of design parameters into a cost function indicating a difference between the input data acquired by the acquisition unit and the virtual input, and finding a minimum value of the cost function; and a determination unit that determines, as the design parameter to be used as a constant in the control system, the one design parameter such that a difference in signal levels of the first signal, the second signal, and the third signal output by inputting the control parameter calculated by the calculation unit and the one design parameter into the control system falls within a predetermined range.

[0007] The calculation unit may calculate, as the control parameters, a proportional gain, an integral gain, and a differential gain of a proportional unit, an integrator, and a differentiator, respectively, included in the feedback controller.

[0008] The determination unit may determine, as the design parameter to be used as the constant, the one design parameter such that a first divided value obtained by dividing the norm of the third signal by the norm of the second signal and a second divided value obtained by dividing the norm of the first signal by the norm of the second signal are within a predetermined range.

[0009] The determination unit may determine the one design parameter that has the smallest value among the plurality of one design parameters for which the first division value and the second division value are included in the predetermined range as the design parameter to be used as the constant.

[0010] The calculation unit may calculate the control parameter by finding a minimum value of the cost function of the control system including the feedforward controller including a reference model of the controlled object, and the determination unit may determine the design parameter to be used as the constant based on differences in signal levels of the first signal, the second signal, and the third signal output from the control system to which the control parameter is input.

[0011] A parameter determination method according to a second aspect of the present invention is executed by a computer and includes an acquisition step of acquiring a plurality of input data to a controlled object and a plurality of output data output by the controlled object over a predetermined period of time, an estimator that estimates and outputs a first signal including a disturbance acting on the controlled object based on the input data and the output data, a feedforward controller that outputs a second signal for calculating a virtual input corresponding to the input data based on a target value of the output data, a feedback controller that outputs a third signal for calculating the virtual input based on a difference between the output data acquired in the acquisition step and the target value, and a summation step of calculating a sum obtained by subtracting the first signal from the second signal and multiplying the result by a reciprocal of a predetermined design parameter and then adding the third signal to the virtual input. a calculation step of calculating a control parameter of the feedback controller by inputting the input data and the output data acquired in the acquisition step and one design parameter among the plurality of design parameters into a cost function indicating a difference between the input data acquired in the acquisition step and the virtual input, and finding a minimum value of the cost function; and a determination step of determining, as the design parameter to be used as a constant in the control system, the one design parameter such that a difference in signal levels of the first signal, the second signal, and the third signal output by inputting the control parameter calculated in the calculation step and the one design parameter into the control system falls within a predetermined range. [Effects of the Invention]

[0012] The present invention provides the advantage of easily determining design parameters suitable for a model-free controller. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a diagram showing an overview of a parameter determination system 1 according to the present embodiment. [Figure 2] 1 is a diagram illustrating an example of the configuration of a control system 10. FIG. [Figure 3]FIG. 2 is a diagram for explaining a reference model Md. [Figure 4] FIG. 1 is a diagram showing the configuration of a control system 10 to which a reference model Md is applied. [Figure 5] FIG. 2 is a diagram showing the configuration of a parameter determination device 20. [Figure 6] 10 is a diagram showing an example of a processing sequence in the parameter determination device 20. FIG. DETAILED DESCRIPTION OF THE INVENTION

[0014] <Overview of Parameter Determination System 1> FIG. 1 is a diagram showing an overview of a parameter determination system 1 according to this embodiment. The parameter determination system 1 shown in FIG. 1 includes a control system 10 and a parameter determination device 20. The parameter determination system 1 is a system for determining control parameters and design parameters used when calculating an input signal u for controlling a control object P. The control object P is, for example, a vehicle, and the control parameters are proportional gains K of a proportional unit, an integrator, and a differentiator included in the control system 10. p , integral gain K i and differential gain K d and the design parameter is the coefficient α used by the control system 10.

[0015] The control system 10 is a system that calculates an input signal u based on an output signal y output by a controlled object P and a target value r of the output signal y. As an example, the control system 10 calculates a steering angle (input signal u) for controlling the yaw rate (output signal y) of a vehicle. The control system 10 calculates a proportional gain K obtained from the parameter determination device 20. p , integral gain K i and differential gain K d and the coefficient α are applied to the control system 10 to perform model free control (MFC) to calculate the input signal u.

[0016] Fig. 2 is a diagram showing an example of the configuration of a control system 10. The control system 10 shown in Fig. 2 includes an estimator 11, a feedforward controller 12, a feedback controller 13, and a calculator 14. In the following description, an output signal y output by a control object P may be referred to as "output data y," an input signal u to the control object P may be referred to as "input data u," and an input signal u calculated by the control system 10 may be referred to as "virtual input."

[0017] The estimator 11 derives a first signal u including a disturbance acting on the control target P based on input data u and output data y. est The feedforward controller 12 estimates and outputs a second signal u for calculating a virtual input corresponding to the input data u based on the target value r of the output data y. ff The feedback controller 13 is a PID (Proportional Integral Differential) controller having a proportional, an integrator, and a differentiator, and outputs a third signal u for calculating a virtual input based on the difference e between the output data y and the target value r. fb The calculator 14 outputs the second signal u ff From the first signal u est The subtracted value is multiplied by the reciprocal α of the specified design parameter (coefficient α). -1 The product of multiplication and the third signal u fb The sum of these is calculated as a virtual input.

[0018] 1, the parameter determination device 20 determines control parameters by minimizing a cost function used in data-driven control such as VRFT (Virtual Reference Feedback Tuning) based on input data u and output data y over a predetermined period. The predetermined period is, for example, 100 seconds. For example, the parameter determination device 20 inputs the input data u, output data y, and a coefficient α into a cost function that indicates the difference between an input signal u (virtual input) calculated by the control system 10 and input data u actually input to the control target P, and determines control parameters that minimize the output of the cost function.

[0019] The parameter determination device 20 can acquire input data u and output data y by operating the controlled object P for a predetermined period of time. However, in order for the parameter determination device 20 to acquire the optimal coefficient α, it is necessary to conduct tests using the control system 10 and the controlled object P for each possible value of the coefficient α. Therefore, the parameter determination device 20 determines the coefficient α when the levels of the output signals of the estimator 11, the feedforward controller 12, and the feedback controller 13 are within predetermined ranges as the optimal coefficient α. By operating in this manner, the parameter determination device 20 can easily determine the coefficient α without repeating tests using the control system 10 and the controlled object P. The configurations and operations of the control system 10 and the parameter determination device 20 will be described in detail below.

[0020] <Configuration of control system 10> The configuration of the control system 10 shown in FIG. 2 will be explained using a mathematical model based on a microlocal model. A single-input, single-output nonlinear discrete system can be given as equation (1).

number

[0021] Then, the microlocal model can be given as equations (2) and (3).

number

[0022] Based on the above-described microlocal model, the control system 10 shown in FIG. 2 is derived and expressed as equations (4) and (5).

number

[0023] In the estimator 11, the equation (5) is algebraically identified and expressed as the equations (6) to (9).

number

[0024] Furthermore, the third signal u output by the feedback controller 13 fb is the proportional gain K p , integral gain K i and differential gain K d Using the above, it is expressed as in equations (10) to (12).

number

[0025] The control system 10 is based on a reference model M d The reference model M d FIG. 4 is a diagram for explaining the reference model M d 4 is a diagram showing the configuration of a control system 10 to which the reference model M shown in FIG. d is applied to the feedforward controller 12 and the feedback controller 13 is an I-PD (Integral-Proportional Differential) controller, but is the same in other respects.

[0026] First, the reference model M shown in Figure 3 dExplain. Reference model M d is a model corresponding to a system M including the control system 10 and the controlled object P shown in FIG. 2. d is the signal y corresponding to the output signal y when the target value r of the output signal y is input. d The reference model M d can be given as equation (13).

number

[0027] Next, the reference model M shown in Figure 4 d A control system 10 to which the above is applied will be described. The mathematical model of the control system 10 can be given as in equations (14) and (15).

number

[0028] The estimator 11 can be given as in equation (16).

number

[0029] By using equations (14) to (16), the control system 10 can be expressed as in equation (17).

number

[0030] The feedback controller 13 can be given as in equation (18).

number

[0031] Therefore, when m=1, equation (18) can be expressed as the I-PD controller shown in equation (19).

number

[0032] Since the I-PD controller has two degrees of freedom, in the feedback controller 13, equation (18) can be replaced by equations (20) to (22).

number

number

[0033] Furthermore, when v and m shown in equations (2), (14), and (18) are defined as v=n-1 and m=v-1, respectively, the control system 10 can calculate the estimated error formula F dltcan be given as equations (25) to (29) using

number

[0034] Here, the feedforward controller 12 and the feedback controller 13 are independent controllers, so F dlt is approximated to 0, the transfer characteristic of the feedback controller 13 is G1r+G3y d Then, the reference model M d Applying the above, the control parameters for m+1=n are given by equation (30).

number

[0035] Then, by substituting 3 for n in equation (30), the control parameters are given as in equation (31).

number

[0036] <Configuration of parameter determination device 20> 5 is a diagram showing the configuration of the parameter determining device 20. The parameter determining device 20 shown in FIG. 5 includes a storage unit 21 and a control unit 22. The control unit 22 includes an acquisition unit 221, a calculation unit 222, and a determination unit 223.

[0037] The storage unit 21 has a storage medium such as a read-only memory (ROM), a random access memory (RAM), a hard disk drive (HDD), or a solid state drive (SSD), etc. The storage unit 21 stores programs executed by the control unit 22 and various information for determining control parameters and design parameters of the control system 10.

[0038] The control unit 22 is a processor such as a CPU (Central Processing Unit) or an ECU (Electronic Control Unit). The control unit 22 functions as an acquisition unit 221, a calculation unit 222, and a determination unit 223 by executing a program stored in the storage unit 21. The control unit 22 may be configured with one processor, or may be configured with multiple processors or a combination of one or more processors and an electronic circuit.

[0039] The acquisition unit 221 acquires a plurality of pieces of input data u to the control object P and a plurality of pieces of output data y output by the control object P during a predetermined period. The predetermined period is, for example, 100 seconds. The acquisition unit 221 may acquire, from the control system 10, the plurality of pieces of input data u that were actually input to the control object P during the predetermined period and the plurality of pieces of output data that were actually output from the control object P during the predetermined period. The acquisition unit 221 acquires a plurality of pieces of input data u that were actually input to the control object P during the predetermined period and a plurality of pieces of output data y ... from the control system 10. d and pre-filter L p Obtain the prefilter L p More on this later.

[0040] The calculation unit 222 calculates the proportional gain K p , integral gain K i and differential gain K dThe calculation unit 222 calculates the control parameters of the feedback controller 13 shown in equations (10) to (12) in the control system 10 shown in Fig. 2, and determines the control parameters of the feedback controller 13 shown in equation (19) in the control system 10 shown in Fig. 4, for example.

[0041] The calculation unit 222 has a cost function that indicates the difference between the input data u of the control system 10 acquired by the acquisition unit 221 and a virtual input (i.e., input signal u) calculated by the control system 10. The calculation unit 222 has the cost function by, for example, referring to the storage unit 21. The calculation unit 222 then inputs the input data u and output data y acquired by the acquisition unit 221 into the cost function and determines the minimum value of the cost function, thereby determining the control parameters of the feedback controller 13.

[0042] For example, in the control system 10 shown in FIG. 2, the calculation unit 222 calculates a reference model M d 4, the calculation unit 222 calculates the control parameters by finding the minimum value of the cost function of the control system 10 including the feedforward controller 12 that does not include the reference model M of the control target P. d The control parameters are calculated by finding the minimum of a cost function of the control system 10 that includes a feedforward controller 12 that includes: As an example, the cost function of the control system 10 shown in FIG. 4 will be explained below using VRFT, which is one of the data-driven controls.

[0043] When time series data D obtained when an open / closed experiment including a control target P is conducted at a predetermined time is given as in equation (32), the virtual value of the target value r is expressed as in equation (33), where N is the data length and t is the discrete time.

number

[0044] In VRFT, processing is performed based on the difference between the input data u actually input to the control target P and the virtual input. Therefore, if a feedback controller C using an adjustable parameter ρ is given as in equation (34), the cost function J VR (ρ) can be given by equation (35).

number

[0045] The reference model M shown in Figure 4 d In a control system including the above, the cost function corresponding to equation (35) is given as equation (38), so the pre-filter L p By applying the pre-filter L, we replace equation (35) with equation (39), which is similar to equation (38). p For example, |L p |=|M d | / φ uu It is expressed as φ uu is, for example, 1.

number

[0046] Furthermore, the virtual inputs calculated by the control system 10 shown in Fig. 4 are given by equations (42) to (46), where I is a unit matrix.

number

[0047] Therefore, equation (39) is given as equations (47) to (50).

number

[0048] The calculation unit 222 calculates the reference model M d and pre-filter L pand the design parameters determined by the determination unit 223, and inputs the input data and output data acquired by the acquisition unit 221. Then, the calculation unit 222 minimizes the cost function (i.e., identifies w corresponding to the minimum value of J(w)), thereby determining the control parameters shown in equation (31). The cost function of the control system 10 shown in FIG. 4 has been described above.

[0049] The determination unit 223 identifies an optimal value of the coefficient α used by the control system 10 and determines the optimal value as a design parameter to be used as a constant in the control system 10. The determination unit 223 determines, for example, the first signal u est , second signal u ff and the third signal u fb The design parameters to be used as constants are determined based on the difference in the signal levels. For example, in the control system 10 shown in FIG. 4, the first signal u est is expressed by equation (44), and the second signal u ff is expressed by equation (43), and the third signal u fb are expressed by equation (45), respectively.

[0050] The determination unit 223 inputs the control parameters calculated by the calculation unit 222 and one of the plurality of design parameters to the control system 10. Then, the determination unit 223 determines the first signal u est , second signal u ff and the third signal u fb A design parameter for which the difference between the signal levels falls within a predetermined range is determined as the design parameter to be used as a constant. The predetermined range is est , second signal u ff and the third signal u fb This is the range in which the respective signal levels match or are close to each other, and is stored in the storage unit 21.

[0051] The decision unit 223 determines, for example, the third signal u fb The third norm of ||u fb || to the second signal u ffThe second norm of ||u ff The first division value r1 divided by || and the first signal u est The first norm of ||u est || is the second norm ||u ff Then, the determination unit 223 determines, from among the plurality of design parameters, one design parameter for which the calculated first division value r1 and second division value r2 are included in a predetermined range, as the design parameter to be used as a constant.

[0052] For example, the determination unit 223 determines the smallest design parameter among a plurality of design parameters for which the first division value r1 and the second division value r2 are included in a predetermined range as the design parameter to be used as the constant. By determining the design parameter (coefficient α) as described above, the determination unit 223 can easily determine the coefficient α without repeating tests using the control system 10 and the controlled object P.

[0053] <Processing Sequence in Parameter Determination Device 20> Fig. 6 is a diagram showing an example of a processing sequence in the parameter determining device 20. The processing sequence shown in Fig. 6 is a processing sequence showing an operation for determining a design parameter (coefficient α) used as a constant by the control system 10. Note that in the following description, an operation for determining a design parameter in the control system 10 shown in Fig. 4 is shown as an example.

[0054] The acquisition unit 221 acquires a plurality of pieces of input data u to the control object P and a plurality of pieces of output data y output by the control object P during a predetermined period (step S11). The acquisition unit 221 acquires a reference model M set by a user operating the operation unit. d and pre-filter L p The acquisition unit 221 acquires the reference model M by referring to the storage unit 21. d and pre-filter L p may be obtained.

[0055] The determination unit 223 sets the maximum value that the coefficient α can take as the coefficient α (step S12). The maximum value is stored in the storage unit 21. The calculation unit 222 calculates the reference model M d and then add the pre-filter L p The calculation unit 222 sets a coefficient α in the cost function shown in equation (47), and inputs the input data u and output data y acquired by the acquisition unit 221 to the cost function (step S13).

[0056] The calculation unit 222 calculates w that minimizes the cost function shown in equation (47) to obtain the proportional gain K p , integral gain K i , differential gain K d (Step S14). The calculation unit 222 identifies the control parameter that indicates the identified proportional gain K p , integral gain K i , differential gain K d is set in the control system 10 shown in equation (17), and the determining unit 223 sets the coefficient α in the control system 10 (step S15).

[0057] JPEG2026018983000024.jpg70170

[0058] If the first division value r1 and the second division value r2 are within a predetermined range (YES in step S18), the determination unit 223 updates the coefficient α (α* shown in FIG. 6) stored in the storage unit 21 to the coefficient α set in step S13 (step S19). If the first division value r1 or the second division value r2 is not within the predetermined range (NO in step S18), the determination unit 223 does not update the coefficient α stored in the storage unit 21.

[0059] The determination unit 223 specifies the minimum value of the coefficient α by referring to the storage unit 21. If the coefficient α set in step S13 is the same as the minimum value of the coefficient α (YES in step S20), the determination unit 223 outputs the coefficient α (α* shown in FIG. 6) stored in the storage unit 21 to the control system 10 (step S22). If the coefficient α set in step S13 is different from the minimum value of the coefficient α (NO in step S20), the determination unit 223 subtracts a predetermined value (Δα) from the coefficient α and sets the result as a new coefficient α (step S21). The predetermined value is, for example, a fixed value stored in the storage unit 21. Then, the parameter determination device 20 returns to the processing of step S13.

[0060] <Effects of the parameter determination device 20> As described above, the parameter determination device 20 includes an acquisition unit 221 that acquires a plurality of pieces of input data u to the control object P and a plurality of pieces of output data y output by the control object P during a predetermined period, a calculation unit 222 that calculates control parameters of the feedback controller 13 by inputting the input data u and output data y acquired by the acquisition unit 221 into a cost function of the control system 10 that indicates the difference between the input data u acquired by the acquisition unit 221 and a virtual input, and finding the minimum value of the cost function, and a first signal u that is output by inputting the control parameters calculated by the calculation unit 222 and one design parameter into the control system 10. est , second signal u ff and the third signal u fb and a determination unit 223 that determines one design parameter for which the difference between the signal levels falls within a predetermined range as the design parameter to be used as a constant in the control system 10.

[0061] By configuring the parameter determination device 20 in this way, the first signal u outputted by the control system 10 est , second signal u ff and the third signal u fbBased on each signal level, it is possible to determine optimal design parameters for the control system 10. As a result, it is possible to determine optimal design parameters without performing tests using the control system 10 and the controlled object P for each possible value of the design parameters, and therefore it is possible to easily determine the design parameters without repeating the tests.

[0062] The present invention has been described above using embodiments, but the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes are possible within the scope of the gist of the present invention. For example, all or part of the device can be configured by functionally or physically distributing or integrating any unit. Furthermore, new embodiments resulting from any combination of multiple embodiments are also included in the embodiments of the present invention. The effects of the new embodiments resulting from the combination also have the effects of the original embodiments. [Explanation of symbols]

[0063] 1. Parameter determination system 10. Control System 11 Estimator 12 Feedforward Controller 13 Feedback Controller 14 Calculator 20 Parameter determination device 21 Memory section 22 Control Unit 221 Acquisition Department 222 Calculation Unit 223 Decision Section

Claims

1. an acquisition unit that acquires a plurality of input data to a control target and a plurality of output data output by the control target during a predetermined period; a control system including: an estimator that estimates and outputs a first signal including a disturbance acting on the controlled object based on the input data and the output data; a feedforward controller that outputs a second signal for calculating a virtual input corresponding to the input data based on a target value of the output data; a feedback controller that outputs a third signal for calculating the virtual input based on a difference between the output data acquired by the acquisition unit and the target value; and a calculator that calculates, as the virtual input, an added value obtained by subtracting the first signal from the second signal and multiplying the result by a reciprocal of a predetermined design parameter, and then adding the third signal to the multiplied value; a determination unit that determines, as the design parameter to be used as a constant in the control system, the one design parameter such that a difference in signal level between the first signal, the second signal, and the third signal output by inputting the control parameter calculated by the calculation unit and the one design parameter falls within a predetermined range; A parameter determination device having:

2. the calculation unit calculates, as the control parameters, a proportional gain, an integral gain, and a differential gain of a proportional unit, an integrator, and a differentiator, respectively, included in the feedback controller; The parameter determination device according to claim 1 .

3. the determiner determines, as the design parameter to be used as the constant, the one design parameter such that a first divided value obtained by dividing the norm of the third signal by the norm of the second signal and a second divided value obtained by dividing the norm of the first signal by the norm of the second signal are included in a predetermined range. The parameter determination device according to claim 1 .

4. the determination unit determines, as the design parameter to be used as the constant, the one design parameter that exhibits the smallest value among the plurality of one design parameters for which the first division value and the second division value are included in the predetermined range. The parameter determination device according to claim 3 .

5. the calculation unit calculates the control parameters by finding a minimum value of the cost function of the control system including the feedforward controller that includes the reference model of the controlled object; the determination unit determines the design parameter to be used as the constant based on differences in signal levels of the first signal, the second signal, and the third signal output from the control system to which the control parameter has been input. The parameter determination device according to claim 1 .

6. The computer executes an acquisition step of acquiring a plurality of pieces of input data to a control target and a plurality of pieces of output data output by the control target during a predetermined period; a control system including: an estimator that estimates and outputs a first signal including a disturbance acting on the controlled object based on the input data and the output data; a feedforward controller that outputs a second signal for calculating a virtual input corresponding to the input data based on a target value of the output data; a feedback controller that outputs a third signal for calculating the virtual input based on a difference between the output data acquired in the acquisition step and the target value; and a calculator that calculates, as the virtual input, an added value obtained by adding a multiplied value obtained by subtracting the first signal from the second signal and multiplying the multiplied value by a reciprocal of a predetermined design parameter to the third signal; a determination step of inputting the control parameter calculated in the calculation step and the one design parameter into the control system, and determining the one design parameter such that a difference in signal level between the first signal, the second signal, and the third signal outputted falls within a predetermined range as the design parameter to be used as a constant in the control system; A parameter determination method having the following:

Images