Control device, control method, and computer program

Abstract

To improve a reliability of a control device in the control device which controls a controlled object having a constraint.SOLUTION: A control device for controlling a controlled object having a predetermined constraint, comprises: a reference filter for acquiring a target value of a state of the controlled object and for calculating an equilibrium point close to the acquired target value within a constraint while the controlled object outputs the state of the controlled object in response to an input; an input filter for determining an input that satisfies the constraint and follows the calculated equilibrium point; and an output unit for outputting the determined input to the controlled object.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a control device, a control method, and a computer program. [Background technology]

[0002] Control devices (controllers) that control the state of a control target are known. For example, Patent Document 1 describes a design method for a nonlinear controller that controls a nonlinear process. For example, Patent Document 2 describes a design method for an LQ controller that controls the state of each individual for generating crowd animation. In Patent Document 1, in order to find a parameter set that can ensure the stability of a feedback loop, and in Patent Document 2, in order to ensure the stability of a feedback loop, both use as a determination criterion whether or not a Lyapunov function can be constructed. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6285466 [Patent Document 2] Japanese Patent Application Laid-Open No. 2001-344615 Summary of the Invention [Problem to be solved by the invention]

[0004] Power plants such as engines and motors have constraints, such as upper and lower limits on turbine speed, that are necessary for safe operation of the plants. When designing a control device for controlling the state of a control plant having such constraints, it is not always possible to find an input that can construct a Lyapunov function within the range of the imposed constraints. Therefore, when the methods described in Patent Documents 1 and 2 are directly applied to a control plant having constraints, the stability of the control plant cannot be guaranteed, and the state of the control plant may diverge or oscillate continuously, resulting in a problem of low reliability. Note that this problem is not limited to control plants such as engines and motors, but is common to all control plants having some kind of constraints.

[0005] The present invention has been made to solve at least part of the above-mentioned problems, and aims to improve the reliability of a control device that controls a control object having constraints. [Means for solving the problem]

[0006] The present invention has been made to solve at least part of the above-mentioned problems, and can be realized in the following aspects. A control device for controlling a control object having a predetermined constraint, the control object outputs a state of the control object in response to an input, the control device comprising: a reference filter that acquires a target value of the state of the control object and calculates an equilibrium point close to the acquired target value within the range of the constraint; an input filter that determines the input that satisfies the constraint and follows the calculated equilibrium point; and an output unit that outputs the determined input to the control object, the input filter determining the input using an inequality that includes a variable representing the equilibrium point and that can ensure stability of a closed-loop system, the inequality being defined by the following equation: TIFF0007804518000001.tif12170 In the above inequality, x is a vector representing the state of the controlled object, e is a vector representing the equilibrium point, t is time, T is the transpose of a matrix or vector, γ is a positive scalar, and P is a positive definite symmetric matrix.

[0007] (1) According to one aspect of the present invention, there is provided a control device for controlling a control object having a predetermined constraint, wherein the control object outputs a state of the control object in response to an input, and the control device includes: a reference filter that acquires a target value of the state of the control object and calculates an equilibrium point close to the acquired target value within the range of the constraint; an input filter that determines the input that satisfies the constraint and follows the calculated equilibrium point; and an output unit that outputs the determined input to the control object.

[0008] According to this configuration, the reference filter acquires a target value for the state of the controlled object, and calculates an equilibrium point (a point to which the state of the controlled object converges over time) that is close to the acquired target value within the range of the constraints. The input filter then selects a point that satisfies the constraints and is close to the calculated equilibrium point. The input that tracks the target point is determined. In this way, by using the reference filter to calculate in advance an equilibrium point close to the target value, it is possible to guarantee that a solution for the input filter (i.e., the input determined by the input filter) always exists within the constraint range. The output unit outputs the input determined in this way to the control object, so that the state of the control object based on the input converges from its initial state to the equilibrium point. As a result, this configuration makes it possible to suppress the occurrence of events such as the state of the control object diverging or the state of the control object continuing to oscillate in a control device that controls a control object with constraints, thereby improving the reliability of the control device.

[0009] (2) In the control device of the above form, the input filter may determine the input using an inequality that includes a variable representing the equilibrium point and that can ensure stability of the closed-loop system. With this configuration, the input filter determines the input using an inequality that can guarantee the stability of the closed-loop system, so the closed-loop system including the controlled object and the input filter can be made globally asymptotically stable, and it becomes possible to converge the state of the controlled object to an equilibrium point over time.

[0010] (3) In the control device of the above aspect, the inequality may be defined by the following equation: TIFF0007804518000002.tif13169In the above inequality, x is a vector representing the state of the controlled object, e is a vector representing the equilibrium point, t is time, T represents the transpose of a matrix or vector, γ is a positive scalar, and P may be a positive definite symmetric matrix. With this configuration, the input filter determines the input using the above inequality. V, which satisfies the above inequality, is called a Lyapunov function, and when the above inequality is always satisfied, the closed-loop system including the controlled object and the input filter can be made globally asymptotically stable, and it becomes possible to converge the state x of the controlled object to the equilibrium point e over time.

[0011] (4) In the control device of the above form, the input filter may determine the input by solving an optimization problem that includes variables representing the input and a prior input that is a candidate for the input, and that finds the input that can minimize the difference between the input and the prior input, with the constraints that the input is within the range of the constraints and satisfies the inequality. According to this configuration, the input filter determines the input by solving an optimization problem to find an input that can minimize the difference between the input and a prior input, with the constraint that the input is within a range of constraints and satisfies an inequality that can guarantee the stability of the closed-loop system. Therefore, the input filter can determine an input that satisfies the constraints and the inequality and is closest to the prior input.

[0012] (5) In the control device of the above aspect, the input filter may obtain the prior input from an LQ controller or a model predictive controller. According to this configuration, the prior input can be determined by an LQ controller capable of guaranteeing stability or a model predictive controller capable of considering constraints.

[0013] (6) In the control device of the above form, the reference filter may calculate the equilibrium point by solving an optimization problem that includes variables representing the target value, the equilibrium point, and a first input that is an input corresponding to the equilibrium point, and that finds the first input that is within the range of the constraints and can minimize the difference between the equilibrium point and the target value. According to this configuration, the reference filter calculates the equilibrium point by solving an optimization problem for finding a first input that is within the constraint range and that can minimize the difference between the equilibrium point and the target value. Therefore, the reference filter can calculate the equilibrium point that is closest to the target value within the constraint range.

[0014] (7) In the control device of the above form, the controlled object outputs an output from the controlled object in place of the state of the controlled object, and the control device may further include a target value conversion unit that acquires a target value of the output from the controlled object and converts the acquired target value into a target value for the state of the controlled object, and an input value conversion unit that acquires the input determined by the input filter as an internal input and converts the acquired internal input into the input that is output to the controlled object. According to this configuration, the target value converter converts a target value of an output from the controlled object into a target value of the state of the controlled object. Similarly, the input value converter acquires an input determined by the input filter as an internal input and converts it into an input to be output to the controlled object. Therefore, even if the target value referenced by the reference filter or the input determined by the input filter differs from the input / output format of the controlled object, processing by the reference filter or the input filter can be applied, thereby improving the versatility of the control device.

[0015] The present invention can be realized in various forms, such as an information processing device or ECU (Engine Control Unit) that controls a control object having predetermined constraints, a system including a control device and a control object, a control object having a built-in control device, a control method for these devices and systems, a computer program executed in these devices and systems, a server device for distributing the computer program, and a non-transitory storage medium on which the computer program is stored. [Brief explanation of the drawings]

[0016] [Figure 1]FIG. 1 is an explanatory diagram illustrating a configuration of a control system. [Figure 2] 10 is a flowchart illustrating an example of processing by a control device. [Figure 3] FIG. 1 is a diagram illustrating convergence, divergence, and oscillation of a state. [Figure 4] FIG. 10 is an explanatory diagram illustrating the configuration of a control system according to a second embodiment. [Figure 5] 10 is a flowchart showing an example of processing by a control device according to a second embodiment. [Figure 6] 1A and 1B are diagrams illustrating an example of input and output of a control target according to a conventional control method. [Figure 7] 10A and 10B are diagrams illustrating an example of input and output of a control target by a control device. DETAILED DESCRIPTION OF THE INVENTION

[0017] First Embodiment Fig. 1 is an explanatory diagram illustrating an example of the configuration of a control system 1. The control system 1 according to one embodiment of the present invention includes a control target 20 and a control device 10 that controls the control target 20. The control target 20 has a "predetermined constraint" that will be described later. The control target 20 is also called a "controlled system."

[0018] The controlled object 20 of the first embodiment is a linear system expressed by the following equation (1) or equation (2). The controlled object 20 can be, for example, a motor as a power engine, but any controlled object can be used as long as it has certain constraints and can be expressed by equations (1) and (2). The variables in equations (1) and (2) are a1 to a5 shown below. Note that "n" is any natural number. (a1) x: n-dimensional state vector representing the state of the control object 20, (a2) u: n-dimensional input vector representing the input to the control object 20; (a3) A, B: Model parameters given to the control object 20 and expressed as an n × n matrix, (a4)t: time; (a5)k: Discrete time.

[0019]

number

number

[0020] For the state x, any item representing the state quantity of the controlled object 20 (for example, a motor) can be used. For the input u, any item representing the manipulated variable for the controlled object 20 can be used. For the model parameters A and B, any item characterizing the properties of the controlled object 20 can be used. Equation (2) is a discretization of the time derivative of equation (1).

[0021] The control device 10 obtains an input u that brings the state x of the controlled object 20 closer to a target value r, and outputs the input u to the controlled object 20. The controlled object 20 again outputs the state x corresponding to the input u acquired from the control device 10 to the control device 10. By repeating this control, the control device 10 can bring the state x of the controlled object 20 closer to the target value r. In other words, the target value r is a target value for the state x of the controlled object 20. The target value r may be provided from outside the control device 10, or may be stored in advance inside the control device 10 (for example, in the storage unit 13).

[0022] The control device 10 can be realized as, for example, a personal computer (PC) or an in-vehicle ECU (Electronic Control Unit). The control device 10 includes a CPU, a ROM / RAM, and a communication unit (not shown), and functions as a reference filter 11 and an input filter 12.

[0023] The reference filter 11 is a functional unit that acquires a target value r of the state x of the controlled object 20 and calculates an equilibrium point e that is closest to the target value r within the constraints of the controlled object 20. Details will be described later. The input filter 12 determines an input u that satisfies the constraints of the controlled object 20 and follows the equilibrium point e calculated by the reference filter 11, and outputs it to the controlled object 20. Details will be described later. Here, the equilibrium point e means the point at which the state x of the controlled object 20 converges over time, in other words, the point at which the time change of the state x becomes 0 (steady state). In equation (1), the state x where dx / dt=0 is the equilibrium point e, and in equation (2), x k+1 =x k The state x where x=x is the equilibrium point e. In this embodiment, the reference filter 11 also functions as an "output unit."

[0024] The control device 10 further includes a storage unit 13. The storage unit 13 is a storage medium configured with a hard disk, a flash memory, a memory card, or the like. The storage unit 120 includes an admissible set 131 and model parameters 132. The admissible set 131 is a storage unit that stores an admissible set U that defines the "predetermined constraints" of the control object 20. In the example of this embodiment, the admissible set U stored in the admissible set 131 defines a set of inputs u that satisfy the constraints of the control object 20. The model parameters 132 is a storage unit that stores model parameters A and B of the control object 20 (any item that characterizes the properties of the control object 20).

[0025] Fig. 2 is a flowchart showing an example of processing by the control device 10. The processing shown in Fig. 2 can be started at any given opportunity. For example, the processing shown in Fig. 2 can be started simultaneously with powering on the control device 10.

[0026] In step S10, the reference filter 11 acquires a target value r, a state x, an admissible set U, and model parameters A and B. The target value r can be acquired from an external device or the memory unit 13. The state x can be acquired from the control object 20. The admissible set U can be acquired from the admissible set 131 in the memory unit 13. The model parameters A and B can be acquired from the model parameters 132 in the memory unit 13.

[0027] In step S12, the reference filter 11 calculates an equilibrium point e for the input u and an input u corresponding to the equilibrium point e. e For example, when the controlled object 20 is a system expressed by equation (1), the equilibrium point e=-A -1 Bu e The reference filter 11 can calculate the equilibrium point e closest to the target value r within the range of the constraints (Equation 3: admissible set U) of the controlled object 20 by solving the optimization problem shown in the following equation (3). In equation (3), Q represents a positive definite symmetric matrix. e corresponds to the “first input.” The reference filter 11 can solve the optimization problem shown in equation (3) using a general solver such as the SQP (Sequential Quadratic Programming) method.

[0028]

number

[0029] Equation (3) is the input u such that e = r. e If is included in the admissible set U, then it is a solution to this optimization problem. Also, equation (3) shows that the input u such that e = r e If is not in the admissible set U (i.e., violates the constraint), then input u is in the admissible set U. e The point where e is closest to r (i.e., (er) T It is shown that the point at which Q(er) is minimized is the solution to this optimization problem. Note that step S12 corresponds to the "process (function) of calculating the equilibrium point e."

[0030] In steps S14 and S16, the input filter 12 determines an input u that satisfies the constraints of the controlled object 20 and tracks the equilibrium point e calculated in step S12. The input filter 12 can determine the input u by using the inequality shown in the following equation (4). In equation (4), t represents time, T represents the transpose of a vector, γ represents a positive scalar, and P represents a positive definite symmetric matrix. V in equation (4) is called a Lyapunov function. It is known that when this inequality is always satisfied, the closed-loop system formed by the input filter 12 and the controlled object can be globally asymptotically stable, and the state x of the controlled object can be converged to the equilibrium point e over time. In other words, equation (4) is an inequality that can guarantee the stability of the closed-loop system. In equation (4), T may represent the transpose of a matrix.

[0031]

number

[0032] Hereinafter, an example of a specific method for the input filter 12 to determine the input u using the inequality of Equation (4) will be described. First, in step S14, the input filter 12 acquires a prior input u0 calculated by another controller. In this embodiment, an LQ controller is exemplified as the other controller. However, any controller can be used as the other controller, and for example, a controller that performs model predictive control may be used.

[0033] The LQ controller calculates the prior input u0 by LQ (Linear Quadratic) control. LQ control is a standard technique for controlling unconstrained linear systems. In LQ control, the input corresponding to the target value r is calculated by the input u. r Then, the input u that minimizes the objective function f expressed by the following equation (5) is given by equation (6). lq and Q in Eq. (7) lq is a positive definite symmetric matrix, and P lqis a positive definite symmetric matrix that satisfies the Riccati equation shown in equation (7).

[0034]

number

number

number

[0035] In step S16, the input filter 12 uses the prior input u0 acquired in step S14 to solve the optimization problem shown in the following equation (8), thereby determining the input u that satisfies the constraints of the controlled object 20 (equation 8: admissible set U) and follows the equilibrium point e. In equation (8), R represents a positive definite symmetric matrix. As shown in equation (4), V=1 / 2(xe) in equation (8) T P(xe). The equilibrium point e is calculated by the reference filter 11 in step S12. The input filter 12 can solve the optimization problem shown in equation (8) using a general solver such as the SQP method.

[0036]

number

[0037] It is known that an LQ controller can guarantee stability but cannot handle constraints, and a model predictive controller can take constraints into account but has difficulty in guaranteeing stability. In this regard, the method of this embodiment can achieve both a stability guarantee and the satisfaction of constraints. Steps S14 and S16 correspond to the "step (function) of determining the input u."

[0038] In step S18, the input filter 12 outputs the input u determined in step S16 to the control object 20. Note that step S18 corresponds to the "output process (function)." In step S20, the input filter 12 determines whether or not a processing termination condition is met. Any condition can be used as the processing termination condition, and for example, it can be triggered by power shutdown of the control device 10. If the termination condition is met (step S20: YES), the input filter 12 terminates the processing. On the other hand, if the termination condition is not met (step S20: NO), the input filter 12 transitions the processing to step S10 and repeats the above steps.

[0039] FIG. 3 is a diagram illustrating convergence, divergence, and oscillation of the state x. As described above, according to the control device 10 of the first embodiment, the reference filter 11 acquires the target value r of the state x of the controlled object, and calculates an equilibrium point e (a point to which the state x of the controlled object 20 converges over time) that is close to the acquired target value r within the range of the constraints defined by the admissible set 131. Then, the input filter 12 determines an input u that satisfies the constraints defined by the admissible set 131 and follows the calculated equilibrium point e. In this way, the reference filter 11, an equilibrium point e close to the target value r is calculated in advance, thereby ensuring that a solution in the input filter 12 (i.e., the input u determined by the input filter 12) always exists within the constraint range. The input filter 12 (output unit) outputs the input u calculated in this manner to the control object 20, so that the state x of the control object 20 based on the input u can be converged from the initial state to the equilibrium point e as shown in FIG. 3(A). As a result, according to the first embodiment, in the control device 10 that controls the control object 20 having constraints, it is possible to suppress the occurrence of an event in which the state x of the control object 20 diverges as shown in FIG. 3(B) or an event in which the state x of the control object 20 continues to oscillate as shown in FIG. 3(C), thereby improving the reliability of the control device 10.

[0040] Furthermore, according to the control device 10 of the first embodiment, the input filter 12 determines the input u using inequality (4), which can guarantee the stability of the closed-loop system. Therefore, the closed-loop system including the controlled object 20 and the input filter 12 can be made globally asymptotically stable, and it becomes possible to converge the state x of the controlled object 20 to the equilibrium point e over time.

[0041] Furthermore, according to the control device 10 of the first embodiment, the input filter 12 determines the input u using the inequality shown in equation (4). V that satisfies the inequality shown in equation (4) is called a Lyapunov function, and when the inequality shown in equation (4) is always satisfied, the closed-loop system including the controlled object 20 and the input filter 12 can be made globally asymptotically stable, and it becomes possible to converge the state x of the controlled object 20 to the equilibrium point e over time.

[0042] Furthermore, according to the control device 10 of the first embodiment, the input filter 12 determines the input u by solving an optimization problem (optimization problem shown in equation (8)) that finds an input u that can minimize the difference between the input u and the prior input u0, with the constraints that the input u is within the constraint range and satisfies inequality (4), which can guarantee the stability of the closed-loop system. Therefore, the input filter 12 can determine the input u that satisfies the constraints and inequality (4) and is closest to the prior input u0. Furthermore, according to the control device 10 of the first embodiment, the prior input u0 can be determined by an LQ controller that can guarantee stability or a model predictive controller that can consider the constraints.

[0043] Furthermore, according to the control device 10 of the first embodiment, the reference filter 11 receives the first input u e is within the constraints and the first input u can minimize the difference between the equilibrium point e and the target value r. e Therefore, the reference filter 11 can calculate the equilibrium point e that is closest to the target value r within the range of constraints.

[0044] Second Embodiment 4 is an explanatory diagram illustrating the configuration of a control system 1A according to the second embodiment. The control system 1A includes a control target 20A instead of the control target 20 and a control device 10A instead of the control device 10 in the configuration of the first embodiment.

[0045] The controlled object 20A of the second embodiment is a nonlinear system expressed by the following equation (9) or equation (10). The controlled object 20A can be, for example, an engine as a power plant, but any controlled object can be used as long as it has certain constraints and can be expressed by equations (9) and (10). The variables in equations (9) and (10) are b1 to b3 shown below in addition to a1 to a5 described in the first embodiment. Note that "n" is any natural number. (b1) y: n-dimensional output vector representing the output from the controlled object 20; (b2) v: n-dimensional input vector representing the input to the control object 20; (b3) Ψ, Φ: Model parameters given to the controlled object 20A, invertible nonlinear functions .

[0046]

number

number

[0047] The output y can be any item representing the amount of output from the controlled object 20 (for example, an engine). The input v can be any item representing the amount of operation on the controlled object 20. In the second embodiment, the input u described in a3 of the first embodiment is treated as an internal input, and the input output from the control device 10A to the controlled object 20A is the input v. Equation (10) is a time-discretized version of equation (9).

[0048] The control device 10A of the second embodiment obtains an input v that brings the output y from the controlled object 20A closer to a target value yr, and outputs the input v to the controlled object 20A. The controlled object 20A outputs the output y corresponding to the input v acquired from the control device 10A again to the control device 10A. By repeating this control, the control device 10A can bring the output y of the controlled object 20A closer to the target value yr. In other words, the target value yr is a target value for the output y of the controlled object 20. The target value yr may be provided from outside the control device 10A or may be stored in advance inside the control device 10A.

[0049] In the configuration of the first embodiment, the control device 10A also functions as a target value converter 14 and an input value converter 15. The target value converter 14 acquires a target value yr of an output y from the controlled object 20A, and converts the target value yr into a target value r of a state x of the controlled object 20A. Details will be described later. The input value converter 15 acquires an input u determined by the input filter 12 as an internal input, and converts the input u into an input v to be output to the controlled object 20A. Details will be described later.

[0050] Fig. 5 is a flowchart showing an example of the processing of the control device 10A of the second embodiment. The processing of Fig. 5 can also be started at any trigger, as in Fig. 2. Fig. 5 differs from Fig. 2 in that steps S30 and S32 are executed before step S10, step S34 is executed after step S16, and step S18A is executed instead of step S18.

[0051] In step S30, the target value conversion unit 14 acquires the target value yr, the output y, and the model parameters Ψ and Φ. The target value yr can be acquired from an external device or the storage unit 13. The output y can be acquired from the controlled object 20A. The model parameters Ψ and Φ can be acquired from the model parameters 132 in the storage unit 13, for example.

[0052] In step S32, the target value conversion unit 14 converts the target value yr into a target value r of the state x of the controlled object 20A, and also converts the output y into the state x of the controlled object 20A. Specifically, the target value conversion unit 14 takes the inverse function of Φ in the above-mentioned equation (9) or equation (10). By doing so, the target value yr and the output y can be converted into a target value r and a state x (see equation (11)). Thereafter, the target value conversion unit 14 transmits the converted target value r and state x to the reference filter 11. Thereafter, the reference filter 11 and the input filter 12 execute steps S10 to S16 described in FIG. 2, and the input u is determined.

[0053]

number

[0054] In step S34, the input value converter 15 acquires the input u from the input filter 12 and converts it into the input v. In this embodiment, it is the input v that is directly output to the controlled object 20A, and the input u is not directly output to the controlled object 20A. Therefore, the input u functions as an internal input. The input value converter 15 can convert the input u into the input v by taking the inverse function of Ψ in the above-mentioned equation (9) or equation (10) (see equation (11)).

[0055] In step S18A, the input value converter 15 outputs the input v converted in step S34 to the control object 20A. In this way, in this embodiment, the input value converter 15 also functions as an "output unit."

[0056] Fig. 6 is a diagram showing an example of input and output of the controlled object 20A using a conventional control method. Fig. 7 is a diagram showing an example of input and output of the controlled object 20A using the control device 10A. Figs. 6 and 7 show changes over time in specific values ​​1 to 3 for each of the items shown in c1 to c4 below. Note that for the internal input u, there are upper and lower limit constraints for each of the specific values ​​u1 to u3 of the input u, but for convenience of illustration, these are shown collectively in a single dotted line. (c1) Output y from controlled object 20A, (c2) Input v to the controlled object 20A, (c3) the state x converted by the target value conversion unit 14; (c4) The internal input u calculated by the input filter 12.

[0057] In FIG. 6, the optimization problem of Equation (3) is not solved in step S12, but e = r is set, and the constraint dV / dt≦−γV in the optimization problem of Equation (8) is removed in step S16. The simulation results are considered to be simulations of the conventional control method. In the example of FIG. 6, although the constraint ranges of the input v and the internal input u are observed, it can be seen that the output y and the state x continue to oscillate and do not converge. On the other hand, FIG. 7 shows the results obtained by simulating using the method described in FIG. 5. In the example of FIG. 7, it can be seen that the constraint ranges of the input v and the internal input u are observed, and the state x of the controlled object 20A converges to the equilibrium point e obtained by the reference filter 11, and the output y from the controlled object 20A converges to a value corresponding to the equilibrium point e obtained by the reference filter 11.

[0058] As described above, the control device 10A of the second embodiment can also achieve the same effects as those of the first embodiment. Moreover, according to the control device 10A of the second embodiment, the target value converter 14 converts the target value yr of the output y from the controlled object 20A into a target value r of the state x of the controlled object 20A. Similarly, the input value converter 15 acquires the input u determined by the input filter 12 as an internal input and converts it into an input v to be output to the controlled object 20A. Therefore, even if the target value r referred to by the reference filter 11 or the input u determined by the input filter 12 differs from the input / output format (output y, input v) of the controlled object 20A, the processing by the reference filter 11 and the input filter 12 can be applied, thereby improving the versatility of the control device 10A.

[0059] <Modification of this embodiment> The present invention is not limited to the above-described embodiment, and can be embodied in various forms without departing from the spirit of the present invention. For example, a part of the configuration realized by hardware may be replaced by software, and conversely, a part of the configuration realized by software may be replaced by hardware. In addition, for example, the following modifications are also possible.

[0060] [Variation 1] In the above embodiment, an example of the configuration of the control system 1 has been described. However, the configuration of the control system 1 can be modified in various ways. For example, in the control system 1, the control device 10 and the controlled object 20 may be communicatively connected via a network and located in physically separate locations. For example, the control system 1 may have a configuration different from the illustrated control device 10 and controlled object 20. For example, in the first embodiment, a motor is used as an example of a linear system represented by Equation (1) or Equation (2), and in the second embodiment, an engine is used as an example of a nonlinear system represented by Equation (9) or Equation (10). However, the engine can also be linearly approximated and used as the controlled object 20 of the first embodiment. Similarly, a motor can be nonlinearly approximated and used as the controlled object 20A of the second embodiment.

[0061] [Variation 2] In the above embodiment, an example of the configuration of the control device 10 has been described. However, the configuration of the control device 10 can be modified in various ways. For example, the control device 10 may include a controller (an LQ controller or a model predictive controller) that requests a prior input within the control device 10. For example, the control device 10 may not have a memory unit 13 that stores the admissible set 131 and the model parameters 132, and may acquire the admissible set 131 and the model parameters 132 from outside the control device 10 (for example, from another ECU connected to the control device 10).

[0062] [Variation 3] In the above embodiment, the processes (FIGS. 2 and 5) executed by the control device 10 have been described using an example of a processing procedure. However, these processing procedures can be modified in various ways, and the processing content of each step may be added, omitted, or changed, and the execution order of each step may be changed. For example, the reference filter 11 or input filter 12 of the control device 10 may be configured to be provided in an external device provided outside the control device 10, and the control device 10 may execute the processes described in FIG. 2 or FIG. 5 in cooperation with the external device.

[0063] For example, in step S12, the reference filter 11 may calculate the equilibrium point e without using the optimization problem described in equation (3). In this case, for example, the reference filter 11 can determine the equilibrium point e by randomly picking the equilibrium point e within the range of the admissible set U.

[0064] For example, in steps S14 and S16, the input filter 12 may determine the input u without using the optimization problem described in equation (8). Also, in steps S14 and S16, the input filter 12 may determine the input u without using a prior input u0 calculated by another controller. Also, in steps S14 and S16, the input filter 12 may obtain a prior input u0 calculated by a controller (e.g., a model predictive controller) using a method different from the LQ controller, and perform the above-mentioned processing.

[0065] For example, in steps S18 and S18A, the input u may be output to a separate device (such as a simulator, an information storage medium, or a display device) that is different from the controlled object.

[0066] The present aspect has been described above based on the embodiments and modifications. However, the above-described embodiments are intended to facilitate understanding of the present aspect and are not intended to limit the present aspect. This embodiment may be modified or improved without departing from the spirit and scope of the claims, and equivalents thereof are included in this embodiment. Furthermore, if a technical feature is not described as essential in this specification, it may be deleted as appropriate.

[0067] The present invention can also be realized in the following forms. [Application example 1] A control device for controlling a control object having a predetermined constraint, The control object outputs a state of the control object in response to an input, a reference filter that acquires a target value of the state of the controlled object and calculates an equilibrium point that is closest to the acquired target value within the range of the constraints; an input filter that determines the input that satisfies the constraints and follows the calculated equilibrium point; an output unit that outputs the determined input to the controlled object; A control device comprising: [Application example 2] The control device according to Application Example 1, A control device in which the input filter determines the input using an inequality that includes a variable that represents the equilibrium point and that can guarantee stability of a closed-loop system. [Application example 3] The control device according to Application Example 1 or Application Example 2, The inequality is defined by the following equation: TIFF0007804518000014.tif12169In the above inequality, A control device, wherein x is a vector representing the state of the controlled object, e is a vector representing the equilibrium point, t is time, T represents the transpose of a matrix or vector, γ is a positive scalar, and P is a positive definite symmetric matrix. [Application example 4] The control device according to any one of Application Examples 1 to 3, The input filter comprises: A control device that determines the input by solving an optimization problem that includes variables representing the input and a prior input that is a candidate for the input, and that finds the input that can minimize the difference between the input and the prior input, with the constraints that the input is within the range of the constraints and satisfies the inequality. [Application example 5] The control device according to any one of Application Examples 1 to 4, The input filter comprises: The controller obtains the prior input from an LQ controller or a model predictive controller. [Application Example 6] The control device according to any one of Application Examples 1 to 5, The reference filter is A control device that calculates the equilibrium point by solving an optimization problem that includes variables representing the target value, the equilibrium point, and a first input that is an input corresponding to the equilibrium point, wherein the first input is within the range of the constraints and that can minimize the difference between the equilibrium point and the target value. [Application Example 7] The control device according to any one of Application Examples 1 to 6, The controlled object outputs an output from the controlled object in place of the state of the controlled object, and further a target value conversion unit that acquires a target value of an output from the controlled object and converts the acquired target value into a target value of a state of the controlled object; an input value conversion unit that acquires the input determined by the input filter as an internal input and converts the acquired internal input into the input that is output to the control object; A control device comprising: [Application Example 8] A control method for a controlled object having a predetermined constraint and outputting a state of the controlled object in response to an input, comprising: obtaining a target value for the state of the controlled object, and calculating an equilibrium point that is closest to the obtained target value within the range of the constraints; determining the inputs that satisfy the constraints and track the calculated equilibrium point; outputting the determined input to the control object; A control method for performing the above. [Application Example 9] A computer program for an information processing device, a function of obtaining a target value for a state of a controlled object that has a predetermined constraint and outputs the state of the controlled object relative to an input, and calculating an equilibrium point that is closest to the obtained target value within the range of the constraint; determining the input that satisfies the constraints and follows the calculated equilibrium point; a function of outputting the determined input to the controlled object; A computer program that executes [Explanation of symbols]

[0068] 1,1A...Control system 10, 10A...Control device 11...Reference filter 12...Input filter 13...Storage section 14...Target value conversion section 15...Input value conversion unit 20, 20A...Control target 120...Storage section 131...Admissible set 132...Model parameters

Claims

1. A control device for controlling a control object having a predetermined constraint, The control object outputs a state of the control object in response to an input, a reference filter that acquires a target value of the state of the controlled object and calculates an equilibrium point that is closest to the acquired target value within the range of the constraints; an input filter that determines the input that satisfies the constraints and follows the calculated equilibrium point; an output unit that outputs the determined input to the controlled object; Equipped with the input filter determines the input using an inequality that includes a variable that represents the equilibrium point and that can ensure stability of a closed-loop system; The inequality is defined by the following equation: In the above inequality, A control device, wherein x is a vector representing the state of the controlled object, e is a vector representing the equilibrium point, t is time, T represents the transpose of a matrix or vector, γ is a positive scalar, and P is a positive definite symmetric matrix.

2. The control device according to claim 1, The input filter comprises: A control device that determines the input by solving an optimization problem that includes variables representing the input and a prior input that is a candidate for the input, and that finds the input that can minimize the difference between the input and the prior input, with the constraints that the input is within the range of the constraints and satisfies the inequality.

3. The control device according to claim 2, The input filter comprises: The control device obtains the prior input from an LQ controller or a model predictive controller.

4. The control device according to any one of claims 1 to 3, The reference filter is A control device that calculates the equilibrium point by solving an optimization problem that includes variables representing the target value, the equilibrium point, and a first input that is an input corresponding to the equilibrium point, wherein the first input is within the range of the constraints and calculates the first input that can minimize the difference between the equilibrium point and the target value.

5. The control device according to any one of claims 1 to 3, The controlled object outputs an output from the controlled object in place of the state of the controlled object, and further a target value conversion unit that acquires a target value of an output from the controlled object and converts the acquired target value into a target value of a state of the controlled object; an input value conversion unit that acquires the input determined by the input filter as an internal input and converts the acquired internal input into the input that is output to the control object; A control device comprising:

6. A control method for a controlled object having a predetermined constraint and outputting a state of the controlled object in response to an input, comprising: obtaining a target value for the state of the controlled object, and calculating an equilibrium point that is closest to the obtained target value within the range of the constraints; determining the inputs that satisfy the constraints and track the calculated equilibrium point; outputting the determined input to the control object; Run In the step of determining the input, the input is determined using an inequality that includes a variable that represents the equilibrium point and that can ensure stability of a closed-loop system; The inequality is defined by the following equation: In the above inequality, A control method in which x is a vector representing the state of the controlled object, e is a vector representing the equilibrium point, t is time, T represents the transpose of a matrix or vector, γ is a positive scalar, and P is a positive definite symmetric matrix.

7. A computer program for an information processing device, a function of obtaining a target value for a state of a controlled object that has a predetermined constraint and outputs the state of the controlled object relative to an input, and calculating an equilibrium point that is closest to the obtained target value within the range of the constraint; determining the input that satisfies the constraints and follows the calculated equilibrium point; a function of outputting the determined input to the controlled object; Execute the function of determining the input determines the input using an inequality that includes a variable that represents the equilibrium point and that can ensure stability of a closed-loop system; The inequality is defined by the following equation: In the above inequality, A computer program, wherein x is a vector representing the state of the controlled object, e is a vector representing the equilibrium point, t is time, T represents the transpose of a matrix or vector, γ is a positive scalar, and P is a positive definite symmetric matrix.

Images