JP2014191736A - Control parameter determination device, method, and program, and controller, and optimization control system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a control parameter determination device, method, and program capable of achieving feedback control having high robustness to a variation in characteristics of a control object without adopting a complicated control algorithm, and to provide a controller and an optimization control system.SOLUTION: A control parameter is determined such that a closed-loop transfer function of a control system 12 or 46 does not depend on any of coefficients {a} and b of identification for identifying a transfer function of a preliminarily identified control object 16. The control system 12 (a controller 18) or a control system 46 (a controller 48) is operated according to a determined control parameter.

Description

  The present invention relates to a control parameter determination apparatus, method, and method for determining a control parameter of a control system having a control object and a controller that executes feedback control based on a target value, an operation amount, and a control amount for the control object, and The present invention relates to a program, a controller, and an optimization control system.

  Conventionally, as one of feedback controls, PID control in which various combinations of P (proportional) operation, I (integral) operation, and D (differential) operation have been widely applied. For example, various methods for adjusting various parameters (hereinafter referred to as control parameters) used for PID control have been proposed.

  Patent Document 1 proposes a method of setting integration time and differentiation time as control parameters to smaller values as the time constant becomes smaller and approaches the dead time.

  Patent Document 2 proposes a method for adjusting each coefficient of the weighted sum of state vectors (time series of state values) so that the actual control amount in the feedback control system follows the output of the reference model.

JP 2005-284828 A JP 2012-190364 A

  By the way, depending on the characteristics of the controlled object, the shape of the transfer function (for example, the coefficient, the highest order, etc.) may not be identified completely, or the shape may vary over time.

  However, if it is attempted to adjust the control parameter using the methods described in Patent Documents 1 and 2, it is necessary to increase the tracking accuracy accordingly, and as a result, the control algorithm becomes complicated or the apparatus becomes large. There was a problem.

  The present invention has been made to solve the above-described problem, and a control parameter determination apparatus, method, and method capable of realizing feedback control with high robustness against fluctuations in characteristics of a controlled object without employing a complicated control algorithm, And a program, and a controller and an optimized control system.

  A control parameter determination device according to the present invention is a device that determines a control parameter of a control system having a control target and a controller that performs feedback control based on a target value, an operation amount, and a control amount for the control target. A parameter determining unit that determines the control parameter so that the closed-loop transfer function of the control system does not depend on any of the identification coefficients that specify the transfer function to be controlled that has been identified in advance. And

  As described above, the control parameter is determined so that the closed-loop transfer function of the control system does not depend on any of the identification coefficients that specify the transfer function to be controlled that has been identified in advance. Even when the identification coefficient fluctuates, the influence on the closed loop transfer function of the control system is reduced. Thereby, it is possible to realize feedback control with high robustness against fluctuations in the characteristics of the controlled object without employing a complicated control algorithm.

  Further, the parameter determination unit uses a transfer function calculation unit that calculates the transfer function by identifying the control object, and a gain matrix that does not depend on any of the identification coefficients, to calculate a state space model of the control system. An optimal gain matrix calculation unit that calculates and calculates an optimal gain matrix that is the gain matrix optimized based on a predetermined evaluation function, and the parameter determination unit is calculated by the transfer function calculation unit. It is preferable that the control parameter is determined using the obtained identification coefficient of the transfer function and the matrix coefficient of the optimum gain matrix obtained by calculation by the optimum gain matrix calculating unit.

  In addition, the transfer function calculation unit calculates a rational function having a denominator of a polynomial and a numerator of a constant as the transfer function, and the parameter determination unit calculates the operation that is inversely proportional to the calculated constant. Preferably, it is determined as one of the control parameters for obtaining the quantity.

  In addition, by introducing one state value depending on the target value to the controllable canonical system and giving one constraint condition based on a switching function expressing a switching operation, the state space model is Preferably constructed.

  Further, it is preferable that the controller is an I-PD controller, and the one state value is a time integration amount of a deviation between the target value and the control amount.

  Moreover, it is preferable to further include a parameter setting unit that sets the control parameter determined by the parameter determination unit to the controller.

  The control parameter determination method according to the present invention is a method for determining a control parameter of a control system having a control target and a controller that performs feedback control based on a target value, an operation amount, and a control amount for the control target. And causing the computer to execute a determination step of determining the control parameter such that the closed-loop transfer function of the control system does not depend on any of the identification coefficients that specify the transfer function to be controlled that has been identified in advance. It is characterized by.

  A control parameter determination program according to the present invention is for determining a control parameter of a control system having a control target and a controller that executes feedback control based on a target value, an operation amount, and a control amount for the control target. The computer executes a determination step for determining the control parameter such that the closed-loop transfer function of the control system does not depend on any of the identification coefficients that specify the transfer function of the control target identified in advance. It is characterized by making it.

  The controller according to the present invention is characterized in that the control parameter determined by any one of the control parameter determination devices described above is set.

  An optimization control system according to the present invention includes any one of the control parameter determination devices described above, and the control system that operates according to the control parameters determined by the control parameter determination device.

  According to the control parameter determining apparatus, method, and program, and the controller and the optimization control system according to the present invention, the closed loop transfer function of the control system has an identification coefficient that specifies the transfer function of the control target that has been identified in advance. Since the control parameters are determined so as not to depend on either, even if the identification coefficient of the shape of the transfer function varies, the influence on the closed-loop transfer function of the control system is reduced. Thereby, it is possible to realize feedback control with high robustness against fluctuations in the characteristics of the controlled object without employing a complicated control algorithm.

It is a whole block diagram of the optimization control system incorporating the control parameter determination apparatus common to 1st and 2nd embodiment. It is a flowchart with which it uses for operation | movement description of the optimization control system shown in FIG. It is a block diagram of a control system concerning a 1st embodiment. FIG. 4A is a first graph illustrating an output waveform when a square wave is input to a control system in which control parameters obtained by the determination method according to the comparative example are set. FIG. 4B is a first graph illustrating an output waveform when a square wave is input to the control system in which the control parameters obtained by the determination method according to the first embodiment are set. FIG. 5A is a second graph showing an output waveform when a square wave is input to the control system in which the control parameter obtained by the determination method according to the comparative example is set. FIG. 5B is a second graph illustrating an output waveform when a square wave is input to the control system in which the control parameter obtained by the determination method according to the first embodiment is set. It is a block diagram of a control system concerning a 2nd embodiment.

  Hereinafter, a control parameter determination method according to the present invention will be described with reference to the accompanying drawings by giving preferred embodiments in relation to a control parameter determination device and a controller.

[Overall Configuration of Optimization Control System 10]
FIG. 1 is an overall configuration diagram of an optimization control system 10 in which a control parameter determination device 14 common to the first and second embodiments is incorporated.

  The optimization control system 10 includes a control system 12 that outputs a control amount y based on an input target value r, and a control parameter determination device 14 that determines a control parameter used for a control operation of the control system 12. Is done.

  The control system 12 includes a control object 16 that outputs a control amount y based on an operation amount u, and a controller 18 that outputs an operation amount u based on a target value r input from the outside.

  The controller 18 is a feedback controller that performs feedback control on the control target 16 based on the target value r, the operation amount u, and the control amount y. For example, a control algorithm including I-PD, PI-D, etc., which is a form of PID control, may be applied.

  The control parameter determination device 14 is a computer including a CPU (Central Processing Unit), a storage device including a hard disk and a memory, and the like. A CPU (not shown) reads out and executes a program stored in a memory or the like, thereby determining a control parameter based on the characteristics of the control target 16, and the determined control parameter as a control system 12 (controller It functions as the parameter setting unit 22 set in 18).

  The parameter determination unit 20 identifies a control object 16 and calculates a transfer function F (s) by calculating a transfer function F (s), an optimal gain matrix calculation unit 26 that calculates an optimal gain matrix Go in a predetermined state space model, Is provided.

[Operation of Optimization Control System 10 (First Embodiment)]
Next, the operation of the optimization control system 10 shown in FIG. 1 will be described in detail with reference to the flowchart of FIG. In the first embodiment, it is assumed that the control system 12 is a continuous time system.

  In step S <b> 1, the transfer function calculation unit 24 identifies the control object 16 and calculates a transfer function F (s) having a Laplace operator (hereinafter referred to as “s operator”) as an argument. Here, it is assumed that the transfer function F (s) has the function form shown in the equation (1).

U (s) is a Laplace transform function of the manipulated variable u, and Y (s) is a Laplace transform function of the control amount y. Here, the transfer function F (s) is a rational function, the denominator function is an (n-1) th order polynomial of the s operator, and the numerator function is a constant (that is, the 0th order expression of the s operator). Keep in mind. Hereinafter, {a _i } (i = 1 to n−1) and b specifying the transfer function F (s) are referred to as “identification coefficients”.

  As a method for identifying the transfer function F (s), various known methods such as a boosting method, an SVM (Support Vector machine), a neural network, and an EM (Expectation Maximization) algorithm may be applied.

  In step S <b> 2, the optimum gain matrix calculation unit 26 calculates the optimum gain matrix Go included in the state equation describing the state space model of the control system 12. The equation of state handled here is expressed by equations (2) to (5).

Here, {c _i } (i = 1 to n−1) and K _I in the equation (5) are n parameters for determining the gain matrix G, and are hereinafter referred to as “matrix coefficients”. Further, z in the equation (4) corresponds to a time integration amount of a deviation between the target value r and the control amount y (hereinafter, input / output deviation (ry)), and is hereinafter referred to as “deviation integration amount”. Further, X in the equation (3) is an amount obtained by adding the deviation integral amount z to the state vector {x _i } (i = 1 to n−1), and is hereinafter referred to as “extended state vector”.

  Then, in order to solve the n simultaneous first-order differential equations shown in the equation (2), the optimum gain matrix calculation unit 26 introduces the evaluation function J shown in the equation (6), and the gain matrix G that minimizes the J is obtained. Is calculated.

  Here, the weighting factor Q is a quasi-positive definite matrix, and the weighting factor R is a positive definite matrix. In this case, the optimum gain matrix Go is determined by Equation (7) using the solution (P) of the Riccati-type differential equation.

From this optimum gain matrix Go, matrix coefficients {c _i } and K _I are determined as unknowns. By the way, as understood from the equations (2) to (5), the matrix A is a constant matrix, and the vectors B and E are constant vectors. That is, it should be noted that the optimum gain matrix Go is a constant matrix that does not depend on any of the identification coefficients {a _i } and b described above.

Thus, the parameter determination unit 20 performs the identification coefficients {a _i } and b obtained by the calculation by the transfer function calculation unit 24 (step S1) and the calculation by the optimum gain matrix calculation unit 26 (step S2). Control parameters of the control system 12 (controller 18) are determined using the obtained matrix coefficients {c _i } and K _I.

  In step S3, the parameter setting unit 22 performs parameter setting for the controller 18 by supplying the control parameter calculated in steps S1 and S2 to the control system 12 side.

  FIG. 3 is a block diagram of the control system 12 shown in FIG. 1 and shows details of the controller 18. The controller 18 is a so-called I-PD controller, and includes a subtracter 30, an integrator 32, a multiplier 34, a subtractor 36, a multiplier 38, a state observer 40, and a weighted adder 42. Prepare.

The subtracter 30 subtracts the negative input signal (current value x ₁ ) from the positive input signal (target value r), and outputs a signal corresponding to (r−x ₁ ). The integrator 32 integrates the input signal from the subtractor 30 in the time domain, and outputs a signal corresponding to ∫ (r−x ₁ ) dt. The multiplier 34 multiplies the input signal by a constant c ₁ K _I and outputs a signal corresponding to c ₁ K _I ｒ (r−x ₁ ) dt.

The subtracter 36 subtracts the negative input signal {Σ (c _i −a _i ) x _i } from the positive input signal {c ₁ K _I (r−x ₁ ) dt} and outputs a signal. . The multiplier 38 multiplies the input signal from the subtracter 36 by a constant (1 / b) and outputs a signal corresponding to the manipulated variable u.

The state observer 40 outputs a state vector {x _i } based on the input operation amount u and the current value x ₁ by applying a known estimation method including a Kalman filter. The weighting adder 42 multiplies the input state vector {x _i } by a constant {c _i −a _i } and then adds the signal to obtain a signal corresponding to Σ (c _i −a _i ) x _i. Output.

  That is, in this configuration, the operation amount u given to the control object 16 is calculated by the equation (8).

  Here, it should be noted that the manipulated variable u is inversely proportional to the numerator (that is, the constant b) of the transfer function F (s).

In step S4, the control system 12 operates according to the control parameter set in step S3. More specifically, the controller 18 performs a predetermined calculation process on the target value r input from the outside, and outputs an operation amount u. Then, the control target 16 outputs the control amount y based on the operation amount u which is input from the controller 18, and supplies to the controller 18 side output (control amount y) as the current value x _1. By sequentially repeating this operation, the controlled object 16 is appropriately feedback controlled.

[Features of this control action]
Subsequently, in order to explain the characteristics of this control operation, a closed-loop transfer function of the control system 12 is calculated. The Laplace transform function Y (s) of the control amount y is obtained by the equation (9) obtained by modifying the equation (1).

  On the other hand, the Laplace conversion function U (s) of the manipulated variable u is obtained by Expression (10) obtained by performing Laplace conversion on both sides of Expression (8).

{X _i (s)}, which is a Laplace transform function of {x _i }, is obtained by Equation (11) obtained by performing Laplace transform on both sides of Equation (2).

そうすると、Ｘ₁（ｓ）は、（９）式に（１０）、（１１）式をそれぞれ代入することで得た、（１２）式によって求められる。

Then, X ₁ (s) is obtained by Expression (12) obtained by substituting Expressions (10) and (11) into Expression (9).

以上より、閉ループ伝達関数は、（１２）式を変形して得た（１３）式によって求められる。

From the above, the closed-loop transfer function is obtained by equation (13) obtained by modifying equation (12).

As understood from the equation (13), the closed-loop transfer function is expressed only by matrix coefficients {c _i } and K _I and does not depend on any of the identification coefficients {a _i } and b. That is, even if the identification coefficients {a _i } and b in the shape of the transfer function F (s) fluctuate, theoretically, the closed loop transfer function of the control system 12 is not affected.

[result]
Next, the result of the control operation when this control parameter is used will be described with reference to FIGS. 4A to 5B. Hereinafter, an I-PD controller designed based on the state equation of the expanded system is cited as a “comparative example”, and the control operations of both are compared.

(1. Stable system)
In the first example, it is assumed that the controlled object 16 is a stable system. More specifically, in the formula (1), n = 3, a ₁ = 0, a ₂ = 10.778, and b = 339.9. Then, in order to artificially generate a change in the transfer function F (s), a ₂ is increased by three values, 50% increase (corresponding to “+ 50%”), as it is (corresponding to “nominal”), and 50% decrease ( Corresponding to “−50%”).

  FIG. 4A is a graph showing an output waveform when a square wave is input to a control system in which control parameters obtained by the determination method according to the comparative example are set. The horizontal axis of the graph is time (unit: s), and the vertical axis of the graph is an output value (control amount y). The definition of the graph is the same for FIGS. 4B, 5A, and 5B described later.

As can be understood from the figure, the output waveform changes greatly according to the setting of a ₂ . In particular, an excessive overshoot occurs in the vicinity of 1.5 (s) in the graph “+ 50%”.

FIG. 4B is a first graph illustrating an output waveform when a square wave is input to the control system in which the control parameters obtained by the determination method according to the first embodiment are set. As can be understood from this figure, an output waveform that is substantially the same regardless of the setting of a ₂ and that is desirable for control can be obtained.

(2. Unstable system)
In the second example, it is assumed that the controlled object 16 is an unstable system. More specifically, in the formula (1), n = 3, a ₁ = 0, a ₂ = −10.78, and b = 339.9. Similarly to the first example, | a ₂ | was set as three values, “+ 50%”, “nominal”, and “−50%”.

FIG. 5A is a second graph showing an output waveform when a square wave is input to the control system in which the control parameter obtained by the determination method according to the comparative example is set. As can be understood from the figure, the output waveform changes greatly according to the setting of a ₂ . In particular, an excessive overshoot occurs in the vicinity of 1.5 (s) in the graph “−50%”.

FIG. 5B is a second graph illustrating an output waveform when a square wave is input to the control system in which the control parameter obtained by the determination method according to the first embodiment is set. As can be understood from this figure, an output waveform that is substantially the same regardless of the setting of a ₂ and that is desirable for control can be obtained.

  As described above, by using the above-described determination method, it is possible to realize feedback control having high robustness with respect to fluctuations in characteristics of the controlled object 16 regardless of whether the system is stable or unstable.

[Derivation method of state equation]
Next, a method for deriving the state equations shown in the equations (2) to (5) will be described in detail.

  First, a state equation of a controllable canonical system having one input and one output is expressed by equations (14) to (16).

Here, Ac is a matrix that depends on the identification coefficient {a _i }, and Bc is a vector that depends on the identification coefficient b. In the expression (16), according to a strict definition, a case where the non-zero element (b, 1) in the vectors Bc, Cc is replaced (1, b) is called a controllable canonical system. In this specification, for convenience of explanation, the expression (16) is referred to as “controllable canonical system”.

Hereinafter, by transforming these equations based on two characteristics, a state space model independent of the identification coefficients {a _i } and b is constructed.

  As a first feature, as shown in the equation (4), one state quantity that depends on the target value r, in the example of I-PD control, a deviation integral that is a time integral quantity of the input / output deviation (ry) The quantity z is introduced.

  As a second feature, one constraint condition based on the switching function σ expressing the switching operation is given. The switching function σ is defined according to the equation (17).

Here, substituting sigma = 0 to determine the status value x _n, the state values x _n, is given by (18).

  If the equations (4) and (18) are substituted into the above equations (14) to (16), this equation of state can be rewritten according to the equation (19).

When the equation (19) is expressed in a matrix form, the state equation describing the state space model of the control system 12 is given by the above-described equations (2) to (5). In this way, a gain matrix G (optimum gain matrix Go) that does not depend on any of the identification coefficients {a _i } and b is obtained.

[Operation of Optimization Control System 10 (Second Embodiment)]
Next, the operation of the optimization control system 10 shown in FIG. 1 will be described in detail with reference to the flowchart of FIG. In the second embodiment, it is assumed that the control system 46 is a discrete time system.

  In step S <b> 1, the transfer function calculation unit 24 identifies the control object 16 and calculates a transfer function F (z) having a Z conversion operator (hereinafter referred to as a “z operator”) as an argument. For example, when the denominator function is a quadratic expression, the transfer function F (s) of the continuous time system is expressed by the expression (20).

  Further, the transfer function F (z) of the discrete time system is expressed by the equation (21) using a known sampling interval T.

It should be noted that the present invention can be applied almost directly to a discrete time system by appropriately changing the continuous time system model. For example, the time differential amount of the state value x _i (t) is approximated according to the equation (22), assuming that the sampling interval T is a primary minute amount.

  The k-th step corresponds to time t = kT. Hereinafter, for convenience, T in the parentheses is omitted as shown in the equation (23).

  In step S <b> 2, the optimum gain matrix calculation unit 26 calculates the optimum gain matrix Go included in the state equation describing the state space model of the control system 46. The equation of state handled here is expressed by equations (24) to (26).

Here, {c _i } (i = 1 to n−1) and K _I in the equation (26) are n parameters for determining the gain matrix G and correspond to “matrix coefficients”. Further, z (k) in the equation (25) corresponds to the time integration amount of the input / output deviation (r−y (k)) and corresponds to “deviation integration amount”. Further, X (k) in the equations (24) and (25) is an amount obtained by adding the deviation integral amount z (k) to the state vector {x _i (k)}, and corresponds to an “extended state vector”. .

  Then, in order to solve the n simultaneous first-order differential equations shown in the equation (24), the optimum gain matrix calculation unit 26 introduces the evaluation function J shown in the equation (27), and the gain matrix G that minimizes J is obtained. Is calculated.

  The weighting factors Q and R are defined in the same manner as in the first embodiment. The optimum gain matrix Go is determined by the equation (28) using the solution (P) of the Riccati-type differential equation.

From this optimum gain matrix Go, matrix coefficients {c _i } and K _I are determined as unknowns. As in the case of the first embodiment, the optimum gain matrix Go is a constant matrix that does not depend on any of the identification coefficients {a _i } and b described above.

Thus, the parameter determination unit 20 performs the identification coefficients {a _i } and b obtained by the calculation by the transfer function calculation unit 24 (step S1) and the calculation by the optimum gain matrix calculation unit 26 (step S2). Control parameters of the control system 46 (controller 48) are determined using the obtained matrix coefficients {c _i } and K _I.

  By the way, the above equations (26) to (28) are derived by modifying the state equation of the controllable canonical system, as in the first embodiment. At this time, the switching function σ (k) is defined according to the equation (29).

Here, substituting sigma (k) = 0 in order to obtain the state value x _n (k), the state value x _n (k) is given by equation (30). By using this, the state equations shown in the equations (26) to (28) are obtained.

  In step S3, the parameter setting unit 22 performs parameter setting for the controller 48 by supplying the control parameters calculated in steps S1 and S2 to the control system 46 side.

  FIG. 6 is a block diagram of the control system 46 (FIG. 1) shown in FIG. 1 according to the second embodiment, and shows details of the controller 48. The controller 48 is a so-called I-PD controller, and includes a subtracter 50, an integrator 52, a multiplier 54, a subtractor 56, a multiplier 58, a state observer 60, and a weighted adder 62. Prepare.

The subtracter 50 subtracts the negative input signal {current value x ₁ (k)} from the positive input signal (target value r), and outputs a signal corresponding to {r−x ₁ (k)}. . The integrator 52 integrates the input signal from the subtractor 50 in the time domain, and outputs a signal corresponding to ΣT {r−x ₁ (j)}. The multiplier 54 multiplies the input signal by a constant c ₁ K _I and outputs a signal corresponding to c ₁ K _I ΣT {r−x ₁ (j)}.

The subtractor 56 subtracts the negative input signal {Σ (c _i −a _i ) x _i (k)} from the positive input signal c ₁ K _I [ΣT {r−x ₁ (j)}]. , Output a signal. The multiplier 58 multiplies the input signal from the subtractor 56 by a constant (1 / b), and outputs a signal corresponding to the manipulated variable u (k).

The state observer 60 outputs a state vector {x _i (k)} based on the input operation amount u and the current value x ₁ (k) by applying a known estimation method including a Kalman filter. The weighting adder 62 multiplies the input state vector {x _i (k)} by a constant {c _i −a _i } and then adds them, thereby adding Σ (c _i −a _i ) x _i (k ) Is output.

  That is, in this configuration, the operation amount u (k) given to the control object 16 is calculated by the equation (31).

  Here, it should be noted that the manipulated variable u (k) is inversely proportional to the numerator (that is, the constant b) of the transfer function F (z).

In step S4, the control system 46 operates according to the control parameter set in step S3. More specifically, the controller 48 performs a predetermined calculation process on the target value r input from the outside, and outputs an operation amount u (k). Then, the control object 16 outputs a control amount y (k) based on the operation amount u (k) input from the controller 48 and outputs the output {control amount y (k)} to the current value x ₁ ( k) to the controller 48 side. By sequentially repeating this operation, the controlled object 16 is appropriately feedback controlled.

  The Z conversion function U (z) of the manipulated variable u (k) is obtained by Expression (32) obtained by performing Z conversion on both sides of Expression (31).

  Hereinafter, by performing the same calculation as in the first embodiment using Equation (32) and the like, for example, when the controlled object 16 is a quadratic system, the closed-loop transfer function is obtained by Equation (33). It is done.

As understood from the equation (33), the closed-loop transfer function is expressed only by matrix coefficients {c _i } and K _I , and does not depend on any of the identification coefficients {a _i } and b. That is, even if the identification coefficients {a _i } and b in the shape of the transfer function F (z) fluctuate, theoretically, the closed loop transfer function of the control system 46 is not affected.

[Effect of each embodiment]
As described above, the closed-loop transfer function of the control systems 12 and 46 can be applied to any of the identification coefficients {a _i } and b that specify the transfer functions F (s) and F (z) of the control target 16 identified in advance. Since the parameter determining unit 20 for determining the control parameter is provided so as not to depend on the identification coefficients {a _i } and b among the shapes of the transfer functions F (s) and F (z), The influence on the closed loop transfer function of the control systems 12 and 46 is reduced. Thereby, it is possible to realize feedback control having high robustness against fluctuations in the characteristics of the controlled object 16 without employing a complicated control algorithm.

  In addition, this invention is not limited to embodiment mentioned above, Of course, it can change freely in the range which does not deviate from the main point of this invention.

DESCRIPTION OF SYMBOLS 10 ... Optimization control system 12, 46 ... Control system 14 ... Control parameter determination apparatus 16 ... Control object 18, 48 ... Controller 20 ... Parameter determination part 22 ... Parameter setting part 24 ... Transfer function calculation part 30, 36, 50, 56 ... Subtractor 32, 52 ... Integrator 34, 38, 54, 58 ... Multiplier 40, 60 ... State observer 42, 62 ... Weighted adder

Claims (10)

An apparatus for determining a control parameter of a control system having a control object and a controller that executes feedback control based on a target value, an operation amount, and a control amount for the control object,
Control comprising a parameter determination unit that determines the control parameter so that the closed-loop transfer function of the control system does not depend on any identification coefficient that identifies the transfer function to be controlled that has been identified in advance. Parameter determination device. The apparatus of claim 1.
The parameter determination unit
A transfer function calculation unit that calculates the transfer function by identifying the control object;
An optimal gain matrix that describes a state space model of the control system using a gain matrix that does not depend on any of the identification coefficients, and calculates an optimal gain matrix that is the gain matrix optimized based on a predetermined evaluation function And a calculation unit
The parameter determination unit uses the identification coefficient of the transfer function obtained by calculation in the transfer function calculation unit and the matrix coefficient of the optimum gain matrix obtained by calculation in the optimum gain matrix calculation unit. A control parameter determination device characterized by determining a control parameter. The apparatus of claim 2.
The transfer function calculation unit calculates, as the transfer function, a rational function whose denominator is a polynomial and whose numerator is a constant,
The parameter determination unit determines the calculated constant as one of the control parameters for obtaining the operation amount inversely proportional to the constant. The apparatus according to claim 2 or 3,
The state space model is constructed by introducing one state value depending on the target value to the controllable canonical system and giving one constraint condition based on a switching function expressing the switching operation. A control parameter determination device characterized by that. The apparatus of claim 4.
The controller is an I-PD controller;
The one state value is a time integration amount of a deviation between the target value and the control amount. In the apparatus of any one of Claims 1-5,
A control parameter determination device, further comprising: a parameter setting unit that sets the control parameter determined by the parameter determination unit to the controller. A method for determining a control parameter of a control system having a control object and a controller that executes feedback control based on a target value, an operation amount, and a control amount for the control object,
The computer is caused to execute a determination step for determining the control parameter so that the closed-loop transfer function of the control system does not depend on any identification coefficient that specifies the transfer function to be controlled that has been identified in advance. Control parameter determination method. A program for determining a control parameter of a control system having a control target and a controller that executes feedback control based on a target value, an operation amount, and a control amount for the control target,
The computer is caused to execute a determination step for determining the control parameter so that the closed-loop transfer function of the control system does not depend on any identification coefficient that specifies the transfer function to be controlled that has been identified in advance. Control parameter determination program to perform.   The control parameter determined by the control parameter determination device according to claim 1 is set. The control parameter determination device according to any one of claims 1 to 6,
An optimization control system comprising: the control system that operates according to the control parameter determined by the control parameter determination device.

Images