JP2023031903A - Information processing device - Google Patents

Abstract

To provide an information processing device capable of obtaining sufficient control performance with respect to a time change of a controlled object.SOLUTION: An information processing device optimizes control parameters of a feedforward controller and a feedback controller, respectively, the information processing device comprising: a first asymptotic unit that asymptotically evaluates an evaluation function regarding an error between a target response of a target response transfer function reference model indicating a target response to a reference signal, and an output of a controlled object; a second asymptotic unit that asymptotically evaluates an evaluation function regarding an error between an output of a sensitivity function reference model for a predetermined disturbance and a control input including a predetermined disturbance; a first parameter adjustment unit that adjusts the control parameter of the feedforward controller based on a result of minimizing the evaluation function asymptotically evaluated by the first asymptotic unit; and a second parameter adjustment unit that adjusts the control parameter of the feedback controller based on a result of minimizing the evaluation function asymptotically evaluated by the second asymptotic unit.SELECTED DRAWING: Figure 1

Description

Applied for application of Article 30, Paragraph 2 of the Patent Law ▲1▼ Publication Hokkaido University 2020 graduation thesis ▲2▼ Date of publication February 1, 2021 ▲3▼ Publisher Hokkaido University National University Corporation ▲4▼ Corresponding page Chapter 4 ▲ 5 ▼ Publisher Tatsu Sakai, Itsuro Kajiwara ▲ 6 ▼ Published title Adaptive parameter tuning of two-degree-of-freedom control system using data-driven control method

The present disclosure relates to an information processing device.

The characteristics of the controlled object may change over time due to aging and the like. Adaptive control is effective for such a time-varying system, in which control parameters are adaptively changed in accordance with changes in the controlled object. Techniques for introducing an adaptive mechanism into FRIT (Fictitious Reference Iterative Tuning) have been proposed. In this technology, online FRIT is applied to a feedback control system by using a forgetting RLS method (Recursive Least Squares with forgetting) as an adaptive mechanism. An adaptive update method for a feedforward controller in a two-degree-of-freedom control system has also been proposed. In this method, an evaluation criterion is introduced, the control performance is evaluated in a predetermined evaluation interval, and when the deterioration of the control performance exceeds a certain threshold, FRIT by the least squares method of the evaluation function is performed online for the predetermined interval. and adaptively change the parameters. Both proposals adaptively update the feedforward controller in the two-degree-of-freedom control system, but do not adaptively update the control parameters of the feedback controller. Patent Literature 1 discloses adaptive updating of control parameters of a feedforward control unit.

It is known that if the controlled object is a time-varying system, the ideal model of the feedback controller also changes with respect to the reference model of the sensitivity function.

In addition, offline FRIT in a feedback control system can automatically adjust optimal control parameters for a target response. However, a situation is also assumed in which a disturbance is applied to the controlled object in addition to the control input. In such a case, even if optimal control parameters obtained by off-line FRIT are used, responsiveness is degraded due to disturbances. Therefore, in addition to target responsiveness, it is also necessary to improve disturbance responsiveness.

JP 2020-134375 A

By the way, when the characteristics of the controlled object fluctuate with time, the control parameters of the feedback controller are not adaptively updated, so there is a problem that sufficient control performance against the time change of the controlled object may not be obtained.

The present disclosure has been made in view of these points, and an object thereof is to provide an information processing apparatus capable of obtaining sufficient control performance with respect to time change of a controlled object.

In order to achieve the above object, the information processing device in the present disclosure
a controlled object to which a control input including disturbance is input;
a feedforward controller that outputs a manipulated variable based on a reference signal as the control input to the controlled object;
a target response transfer function normative model indicating a target response to the reference signal;
a feedback controller that outputs to the controlled object as the control input the manipulated variable based on the deviation between the target response of the target response transfer function reference model and the output of the controlled object;
An information processing device that optimizes control parameters of the feedforward controller and the feedback controller in a control system comprising
a state acquisition unit that acquires the control input and the output of the controlled object;
a first asymptotic unit that asymptotically evaluates an evaluation function regarding an error between the target response of the target response transfer function reference model and the output of the controlled object;
In a sensitivity function reference model that is a transfer function from a disturbance to the control input, a second asymptotic to asymptotically an evaluation function regarding an error between an output of the sensitivity function reference model for a predetermined disturbance and a control input including the predetermined disturbance. Department and
a first parameter adjusting unit that adjusts the control parameters of the feedforward controller based on the result of minimizing the evaluation function asymptotically obtained by the first asymptotic unit;
a second parameter adjustment unit that adjusts the control parameters of the feedback controller based on the result of minimizing the evaluation function asymptotically obtained by the second asymptotic unit;
Prepare.

According to the present disclosure, it is possible to obtain sufficient control performance with respect to temporal changes in the controlled object.

FIG. 1 is a diagram showing a model-matching two-degree-of-freedom control system using the adaptive mechanism of the RLS method. FIG. 2 is a flow chart showing the procedure of a simulation using a controlled object. FIG. 3 is a diagram showing responses when adaptive control is performed by online FRIT for a predetermined disturbance (d=0.0) for a time-varying system. FIG. 4 is a diagram showing responses when adaptive control is performed by online FRIT for a predetermined disturbance (d=0.5) for a time-varying system. FIG. 5 is a diagram schematically showing a functional configuration of an information processing device according to an embodiment of the present disclosure; FIG. 6 is a flow chart showing an example of the operation of the information processing device according to the embodiment of the present disclosure.

<Technology that is a premise of the embodiment>
Embodiments of the present disclosure will be described. First, the model-matching two-degree-of-freedom control system will be described with reference to FIG. FIG. 1 is a diagram showing a model-matching two-degree-of-freedom control system using the adaptive mechanism of the RLS method. In FIG. 1, "C _r (t, ρ vector)" is a feedforward controller (FF controller), "C _e (t, ρ vector)" is a feedback controller (FB controller), and "r" is a reference signal, "d" is the unknown disturbance, "u" is the control input, "y" is the control output, "ρ vector" is the control parameter of the controller, and "T _d " is the target from the reference signal r to the output y. The response transfer function normative model, "G(t)", is the controlled object. The controlled plant G(t) is assumed to be a linear time-varying system whose dynamics are unknown. A forgetful RLS algorithm is employed as the adaptation mechanism.

The adaptation mechanism optimizes the control parameters of the FF controller _Cr with respect to the target response transfer function reference model _Td . Also, the adaptive mechanism optimizes the control parameters of the FB controller _Ce with respect to the sensitivity function reference model _Sd , which is the transfer function from the disturbance d to the control input u.

ここで、ＦＦ制御器はＦＩＲ（Finite Impulse Response）フィルタである。ＦＦ制御器は制御パラメータに対して線形である。式中の「ｎ」はフィルタ次数、「ｚ」はｚ変換における複素変数、「Ｔ」はベクトルの転置を表す。また、式中の「ρ」はＦＦ制御器の制御パラメータである。なお、以下の説明において、特に言及しない限り、ＦＦ制御器に関する式中に用いられる場合の「ρ」はＦＦ制御器の制御パラメータであり、ＦＢ制御器に関する式中に用いられる場合の「ρ」はＦＢ制御器の制御パラメータである。 The control parameters of the FF controller _Cr are given by the following equations.

Here, the FF controller is an FIR (Finite Impulse Response) filter. The FF controller is linear with respect to the control parameters. In the equation, "n" is the filter order, "z" is the complex variable in the z-transform, and "T" is the transpose of the vector. Also, "ρ" in the formula is a control parameter of the FF controller. In the following description, unless otherwise specified, "ρ" used in the formulas for the FF controller is the control parameter of the FF controller, and "ρ" used in the formulas for the FB controller is the control parameter of the FB controller.

ここで、ＦＢ制御器はＰＩＤ制御器、ＦＢ制御器は制御パラメータに対して線形である。式中の「ρ」はＦＢ制御器の制御パラメータ、「Ｋ_ｐ」は比例パラメータ、「Ｋ_ｉ」は積分パラメータ、「Ｋ_ｄ」は微分パラメータ、「τ_ｆ」はフィルタ時定数、「ｓ」はラプラス演算子、「ｚ」はｚ変換における複素変数、「Ｔ」はベクトルの転置を表す。 Also, the control parameters of the FB controller _Ce are as shown in the following equations.

Here, the FB controller is a PID controller, and the FB controller is linear with respect to the control parameters. In the formula, “ρ” is the control parameter of the FB controller, “K _p ” is the proportional parameter, “K _i ” is the integral parameter, “K _d ” is the derivative parameter, “τ _f ” is the filter time constant, and “s” is the Laplacian operator, 'z' is the complex variable in the z-transform, and 'T' is the transpose of the vector.

式中の「λ」は、ＲＬＳアルゴリズムにおける忘却係数（重み係数ともいう）であり、０＜λ＜１として設定される。定性的な意味として、忘却係数λが１に近いほど、過去の結果を強く反映し、忘却係数λが０に近いほど、時間変化に対して敏感となる。式中の「ｅハット」は、目標応答伝達関数規範モデルＴ_ｄの出力と制御出力ｙとの誤差である。 The evaluation function used for adaptive updating of the FF controller is shown in the following equation.

"λ" in the formula is a forgetting factor (also referred to as a weighting factor) in the RLS algorithm and is set as 0<λ<1. In a qualitative sense, the closer the forgetting factor λ is to 1, the stronger the past results are reflected, and the closer the forgetting factor λ is to 0, the more sensitive to time changes. "e hat" in the equation is the error between the output of the target response transfer function reference model _Td and the control output y.

式中の「ｂ」は所定値、「γ」は定数、「Ｐ（０）」は初期値、「Ｉ」は、単位行列である。また、初期値Ｐ（０）には、比較的大きい定数γ（例えば１０^２）が用いられる。 The forgetful RLS algorithm used for adaptive updating is as follows.

In the formula, "b" is a predetermined value, "γ" is a constant, "P(0)" is an initial value, and "I" is a unit matrix. A relatively large constant γ (eg, 10 ² ) is used for the initial value P(0).

[Adaptive update of FB controller when offline]
Next, adaptive updating of the FB controller will be described. As shown in FIG. 1, in offline FRIT for a two-degree-of-freedom control system, input/output data u _{ν_ini} and y _{ν_ini} when the reference signal r=0 and the disturbance signal ν≠0 are used to optimize the FB controller. was necessary. In online FRIT with adaptive updating, it is assumed that reference signal r≠0 and disturbance signal ν=0 during control execution. cannot be obtained.

Therefore, a response prediction technique called ERIT (Estimated Response Iterative Tuning) is used. ERIT is based on the concept of "data-driven prediction" in a two-degree-of-freedom control system. It is possible to predict the response when updating the FF controller using known input/output data before updating parameters. It is a method that can evaluate its stability and responsiveness before doing so. In this method, for example, the input/output data u _{ini and} y _ini obtained under the condition that the reference signal r=r ₁ ≠0 and the disturbance signal ν=ν ₁ =0 are used to obtain the reference signal r=r ₂ =0 and the disturbance signal ν = ν ₂ ≠0, the input/output data u _{ν_ini} and y _{ν_ini} to be acquired are predicted. Note that the disturbance signal ν=ν ₂ ≠0 when the reference signal r=r ₂ =0 corresponds to the “predetermined disturbance” of the present disclosure.

First, under the condition that r=r ₂ =0 and ν=ν ₂ ≠0, the input/output data u _{ν_ini} and y _{ν_ini} to be obtained are expressed as follows.

Further, the input/output data u _ini and y _ini obtained under r=r ₁ ≠0 and ν=ν ₁ =0 are represented by the following equations.

The following equation can be obtained by writing equations (9) to (12) in the form of z-transform and transforming the equation so as to eliminate the unknown controlled object G(z).

At this time, the reference signal r=r ₁ ≠0 used in the first experiment is assumed to be a constant value, and the artificially applied disturbance ν=ν ₂ ≠0 emphasized at a low frequency, which is supposed to be used in the second experiment. is a fixed value, the equations (13) and (14) become the following equations.

Here, when the evaluation function is represented by equation (17) and fixed as C _e (z, ρ vector)=C _e (z, ρ _ini vector), the prefilter F that satisfies equation (17) is expressed by equation ( 18).

In particular, if we set r ₁ =1 and ν ₂ =1, it has the same form as the prefilter F(z, ρ _ini vector) shown in equation (18), so we express it as F(z, ρ _ini vector) Then, equation (15) becomes equation (19). Also, equation (16) becomes equation (20).

As described above, during execution of control, the input/output data u and y at r=r ₁ ≠0, ν=ν ₁ =0 are changed to the input/output data u at r=r ₂ =0, ν=ν ₂ ≠0. _ν and y _ν are obtained by calculation. The obtained input/output data u _ν and y _ν are expressed by the following equations. For the purpose of constant value control, the reference signal r is assumed to be constant, and the disturbance signal emphasized in the low frequency band ν is also assumed to be constant.

The evaluation function for optimization of the FB controller in offline FRIT is shown in the following equation.

[Adaptive update of FB controller when online]
Next, the adaptive updating of the FB controller when online will be described.
According to the equation (23), the evaluation function J tilde _Sd ^k (ρ vector) in the online case is given by the following equation.

At this time, the characteristics of the prefilter F(z, ρ(k) vector) change with the adaptively updated control parameter ρ vector. For a time-varying prefilter F(z, ρ(k) vector), it is difficult to compute its response F(z, ρ(k) vector) y(k). Therefore, when minimizing, the following equation, which is asymptotically equivalent to equation (24), is derived.

式中の「λ」は、ＲＬＳアルゴリズムにおける忘却係数（重み係数ともいう）である。 Minimizing equation (25) is asymptotically equivalent to minimizing equation (24). In addition, although the reference signal r is assumed to be a constant value, even when r=r(t) where the reference signal r changes with time, the minimization of Equation (25) is asymptotically equivalent to Equation (24). becomes. Based on the above, the evaluation function used for adaptive updating of the FB controller is shown in the following equation.

"λ" in the formula is a forgetting factor (also called a weighting factor) in the RLS algorithm.

式中の「ａ」は所定値、「γ」は定数である。「Ｐ（０）」は初期値、「Ｉ」は、単位行列である。初期値Ｐ（０）には、比較的大きい定数γ（例えば１０^２）が用いられる。 The forgetful RLS algorithm used for adaptive updating is as follows.

In the formula, "a" is a predetermined value and "γ" is a constant. "P(0)" is an initial value, and "I" is a unit matrix. A relatively large constant γ (eg, 10 ² ) is used for the initial value P(0).

Note that when r=0, the input/output data necessary for optimizing the FF controller cannot be obtained. Input/output data necessary for the FB controller are calculated by equations (21) and (22), but when r=0, both cannot be calculated. Since the present disclosure aims at constant value control when r≠0 and not constant value control when r=0, the adaptive algorithm when r=0 does not work properly.

[Optimization of FF controller]
Next, optimization of the FF controller will be described. An FIR filter is used for the FF controller. The reasons for using the FIR filter are that it is always stable, that it can have a linear phase, and that the structure is easy to understand. However, since the FIR filter generally needs to have a high order, the nonlinear optimization requires a large amount of calculation, time and cost. Therefore, since the FIR filter and the PID controller are linear with respect to the parameters, it is ideal to be able to solve them by the least squares method. The least squares method is introduced in FRIT in the optimization of the FF controller of the two-degree-of-freedom control system.

Consider the optimization of the FF controller. Substituting equation (32) into equation (33) defines the error signal e _M as shown in equation (34).

The input/output data u _ini and y _ini obtained under r=r ₁ ≠0 and ν=ν ₁ =0 are shaped by the pre-filter F, and the evaluation function J tilde _Td ^F (ρ vector) shown in the following equation is obtained. think of.

Also, the evaluation function shown in Equation (36) is rewritten as Equation (37).

Similarly, equation (35) is also rewritten as equation (38).

Equations (37) and (38) are equivalent when the following equations are satisfied.

Since they can be designed independently, we fix C _e (z, ρ vector)=C _e (z, ρ _ini vector) in order to optimize them separately.
A pre-filter F that satisfies the equation (39) becomes the equation (40).

Let the FF controller C _r be linear with respect to the control parameter ρ, as in equation (41).

At this time, equation (35) becomes equation (42).

以上、ＦＦ制御器の最適化について説明した。 The optimum parameters are solved by the method of least squares as shown in Equation (43).

The optimization of the FF controller has been described above.

Next, a simulation for a time-varying system will be described. FIG. 2 is a flow chart showing the procedure of a simulation using a controlled object.

First, in step S1, the controlled object G(t) is assumed to be time-invariant, and input/output data u _ini and y _ini with initial parameters are obtained. At this time, the reference signal r=r ₁ =1.0.

Next, in step S2, the obtained input/output data u _ini and y _ini are used to optimize the control parameters of the FF controller C _r and the control parameters of the FB controller C _e by offline FRIT.

Next, in step S3, the optimal parameters obtained by the offline FRIT are used as initial values to start controlling the time-varying controlled object G(t).

Next, in step S4, the control parameters of the FF controller _Cr are adaptively updated, and the control parameters of the FB controller _Ce are adaptively updated by online FRIT based on the RLS algorithm.

[inspection result]

Using the optimum parameters obtained by offline FRIT as initial values, adaptive control by online FRIT was performed on the time-varying system. FIG. 3 shows the time history of input/output data of online FRIT under disturbance (d=0.0). FIG. 4 shows the time history of input/output data of online FRIT under disturbance (d=0.5). FIGS. 3 and 4 respectively show input/output data for no adaptive update (dashed-dotted line), adaptive update only for the FF controller (dashed-dotted line), and adaptive update for both the FF controller and the FB controller (dotted line).

As shown in FIGS. 3 and 4, it can be seen that the response with the adaptive update shown by the dashed and dotted lines is improved compared to the response without the adaptive update shown by the dashed two-dotted line. In particular, the response when both the FF controller and the FB controller shown by the dashed line are adaptively updated has better followability to the target response as the gain of the controlled object decreases than when only the FF controller is adaptively updated. is high.

<Functional Configuration of Information Processing Apparatus 1 According to Embodiment>
Based on the above technology, the information processing apparatus 1 according to the embodiment will be described. The information processing apparatus 1 has an adaptive mechanism denoted by "RLS" in FIG.

FIG. 5 is a diagram schematically showing the functional configuration of the information processing device 1 according to the embodiment of the present disclosure. The information processing device 1 includes a storage section 2 and a control section 3 . In FIG. 5, arrows indicate main data flows, and data flows not shown in FIG. 5 may exist. In FIG. 5, each functional block does not show the configuration in units of hardware (apparatus), but the configuration in units of functions. Therefore, the functional blocks shown in FIG. 5 may be implemented within a single device, or may be implemented separately within a plurality of devices. Data exchange between functional blocks may be performed via any means such as a data bus, network, or portable storage medium.

The storage unit 2 includes a ROM (Read Only Memory) that stores a BIOS (Basic Input Output System) of a computer that implements the information processing apparatus 1, a RAM (Random Access Memory) that serves as a work area of the information processing apparatus 1, an OS ( Operating System), application programs, various information referenced when the application programs are executed, the state of the controlled object G and the control parameters of the controller C are linked and stored in an HDD (Hard Disk Drive) or SSD (Solid State Drive) or other large-capacity storage devices.

The control unit 3 is a processor such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) of the information processing device 1, and executes a program stored in the storage unit 2 to obtain the state acquisition unit 30, the first asymptotic It functions as a unit 31 , a second asymptotic unit 32 , a first parameter adjustment unit 33 , a second parameter adjustment unit 34 and an input/output data estimation unit 35 .

Note that FIG. 5 shows an example in which the information processing apparatus 1 is composed of a single apparatus. However, the information processing apparatus 1 may be realized by computational resources such as a plurality of processors and memories, for example, like a cloud computing system. In this case, each unit that configures the control unit 3 is implemented by executing a program by at least one of a plurality of different processors.

The state acquisition unit 30 acquires input/output data of a controlled object. The input/output data acquired by the state acquisition unit 30 is stored in the storage unit 2 . Specifically, the state acquisition unit 30 acquires the input/output data u ini _{and y ini when} the reference signal r=r ₁ ≠0 and the disturbance signal ν=ν ₁ =0.

The input/output data estimator 35 estimates the input/output data when the reference signal r=r ₂ =0 and the disturbance signal ν=ν ₂ ≠0 based on the input/output data obtained by the state obtaining unit 30 . . Specifically, based on the input/output data u _ini and y _ini , the input/output data u _{ν_ini} and y _{ν_ini} are estimated when the reference signal r=r ₂ =0 and the disturbance signal ν=ν ₂ ≠0.

The first asymptotic unit 31 asymptoticizes an evaluation function regarding an error between the target response of the target response transfer function reference model _Td and the output of the controlled object, based on the input/output data acquired by the state acquisition unit 30 . Specifically, the first asymptotic unit 31 asymptoticizes the evaluation function shown in Equation (42). Although the evaluation function shown in Equation (42) includes the prefilter F, the evaluation function shown in Equation (44) is asymptotically similar to the evaluation function shown in Equation (42) even without using the prefilter F. can be said to be equivalent to This eliminates the need to calculate the pre-filter F when executing FRIT by the least squares method on-line, so that the calculation cost can be reduced.

The second asymptotic unit 32 asymptoticizes an evaluation function regarding an error between the output of the sensitivity function reference model and the control input including the predetermined disturbance based on the input/output data estimated by the input/output data estimation unit 35 . Specifically, the second asymptotic unit 32 asymptoticizes the evaluation function shown in Equation (24). Although the evaluation function shown in Equation (24) includes the prefilter F, the evaluation function shown in Equation (24) is asymptotically similar to the evaluation function shown in Equation (26) without using the prefilter F. can be said to be equivalent to This eliminates the need to calculate the pre-filter F when executing FRIT by the least squares method on-line, so that the calculation cost can be reduced.

The first parameter adjusting section 33 adjusts the control parameters of the FF controller based on the result of minimizing the evaluation function asymptotically obtained by the first asymptotic section 31 .

A second parameter adjustment unit 34 adjusts the control parameters of the FB controller based on the result of minimizing the evaluation function asymptotically obtained by the second asymptotic unit 32 . Specifically, the control parameters of the FB controller are adjusted based on the result of minimizing the evaluation function shown in Equation (25).

<Processing Flow of Information Processing Executed by Information Processing Apparatus 1>
FIG. 6 is a flowchart for explaining the flow of information processing of online FRIT executed by the information processing apparatus 1 according to the embodiment. The processing in this flowchart is started, for example, when the information processing apparatus 1 is activated, and is repeatedly executed at predetermined time intervals after the start.

The state acquisition unit 30 acquires the input/output data of the controlled object when the reference signal r=r ₁ ≠0 and the disturbance signal ν=ν ₁ =0 (step S100).

The input/output data estimator 35 estimates the input/output data of the controlled object when the reference signal r=r ₂ =0 and the disturbance signal ν=ν ₂ ≠0 based on the acquired input/output data (step S110).

The first asymptotic unit 31 asymptoticizes the evaluation function regarding the error between the target response of the target response transfer function reference model and the output of the controlled object (step S120).

The second asymptotic unit 32 asymptoticizes the evaluation function regarding the error between the output of the sensitivity function reference model for a given disturbance and the control input including the given disturbance (step S130).

The first parameter adjuster 33 adjusts the control parameters of the FF controller (step S140).

The second parameter adjuster 34 adjusts the control parameters of the FB controller (step S150). After that, the processing in this flowchart ends.

<Effects of Information Processing Apparatus 1 According to Embodiment>
As described above, according to the information processing apparatus 1 according to the embodiment, the evaluation function regarding the error between the target response of the target response transfer function reference model and the output of the controlled object, and the output of the sensitivity function reference model and the predetermined Each of the evaluation functions is minimized without using the pre-filter F included in each of the evaluation functions regarding the error with the control input including the disturbance of . Accordingly, when executing online FRIT for each of the feedforward control system and the feedback control system, there is no need to calculate the prefilter F, so the calculation cost can be reduced. As a result, it is possible to obtain sufficient control performance with respect to the time change of the controlled object.

In addition, in the present embodiment, the application of online FRIT to each of the feedforward control system and the feedback control system has been described, but in the present disclosure, the control parameters of the controller are adaptively changed according to changes in the controlled object It is not limited to FRIT, as long as it is controlled while allowing it to move. Moreover, although the application of the FIR filter to the FF controller has been described, the present disclosure is not limited to the FIR filter.

As described above, the present disclosure has been described using the embodiments, but the technical scope of the present disclosure is not limited to the range described in the above embodiments, and various modifications and changes are possible within the scope of the gist thereof. be. For example, specific embodiments of device distribution/integration are not limited to the above-described embodiments. can be done. In addition, new embodiments resulting from arbitrary combinations of multiple embodiments are also included in the embodiments of the present disclosure. The effect of the new embodiment caused by the combination has the effect of the original embodiment.

In addition, the above-described embodiments are merely examples of specific implementations of the present disclosure, and the technical scope of the present disclosure should not be construed to be limited by these. . That is, the present disclosure can be embodied in various forms without departing from its spirit or key features.

INDUSTRIAL APPLICABILITY The present disclosure is suitably used for industrial systems equipped with information processing devices that are required to obtain sufficient control performance with respect to changes in controlled objects over time, particularly for industrial systems with strong nonlinearity such as engines and automatic transmissions. be.

1 information processing device 2 storage unit 3 control unit 30 state acquisition unit 31 first asymptotic unit 32 second asymptotic unit 33 first parameter adjustment unit 34 second parameter adjustment unit C _r feedforward controller (FF controller)
_Ce feedback controller (FB controller)
G controlled object T _d target response transfer function reference model

Claims (2)

a controlled object to which a control input including disturbance is input;
a feedforward controller that outputs a manipulated variable based on a reference signal as the control input to the controlled object;
a target response transfer function normative model indicating a target response to the reference signal;
a feedback controller that outputs to the controlled object as the control input the manipulated variable based on the deviation between the target response of the target response transfer function reference model and the output of the controlled object;
An information processing device that optimizes control parameters of the feedforward controller and the feedback controller in a control system comprising
a state acquisition unit that acquires the control input and the output of the controlled object;
a first asymptotic unit that asymptotically evaluates an evaluation function regarding an error between the target response of the target response transfer function reference model and the output of the controlled object;
In a sensitivity function reference model that is a transfer function from a disturbance to the control input, a second asymptotic to asymptotically an evaluation function regarding an error between an output of the sensitivity function reference model for a predetermined disturbance and a control input including the predetermined disturbance. Department and
a first parameter adjusting unit that adjusts the control parameters of the feedforward controller based on the result of minimizing the evaluation function asymptotically obtained by the first asymptotic unit;
a second parameter adjustment unit that adjusts the control parameters of the feedback controller based on the result of minimizing the evaluation function asymptotically obtained by the second asymptotic unit;
comprising
Information processing equipment. The first parameter adjustment unit adjusts control parameters of the feedforward controller online,
The second parameter adjustment unit adjusts control parameters of the feedback controller online.
The information processing device according to claim 1 .

Images