JP2022122018A - Control device designing method and electric inertia control device designing method - Google Patents

Abstract

To provide an electric inertia control device designing method which can calculate a plurality of control parameters of an electric inertia control device of a two-degree-of-freedom control system at a low calculation cost.SOLUTION: A dynamo controller 5 being an object to be designed is a two-degree-of-freedom control system comprising a feedback controller 51 and a feed forward controller 52. The designing method comprises the steps of: acquiring an initial controlled variable y0, an initial target value r0, an initial controlled variable deviation e0, and an initial axial torque measured variable T12_0 which are N-component time series data; and determining design values of control parameters (KPe, KIe, KPr) so that a first reference output φ1(t) obtained by inputting a deviation between a response of a reference target response model Gmr(z-1) to the initial target value r0 and the initial controlled variable y0 to the feedback controller 51 and a second reference output φ2(t) obtained by inputting a fictitious exogenous signal T-12(t) to a reference disturbance response model G'md(z-1) get closer to each other.SELECTED DRAWING: Figure 6

Description

The present invention relates to a control device and a method for designing an electric inertial control device. More specifically, the present invention relates to a design method for a control device and an electric inertia control device, which are two-degree-of-freedom control systems.

In test systems using dynamometers, such as engine bench systems and drivetrain bench systems, specimens such as engines and drivetrains are connected to the dynamometer via shafts. In such a test system, many electric inertia control devices for dynamometers have been proposed for conducting load tests simulating actual road running loads. Here, in order to achieve a predetermined simulated inertia with high accuracy by the dynamometer, it is important to appropriately set the values of the control parameters of the feedback controller such as various gains.

Conventionally, the value of the control parameter is often adjusted by the designer based on experience using the results of estimating the system parameters of the test system, such as the inertia of the specimen and the rigidity of the shaft. However, in such a conventional method, the result of adjusting the value of the control parameter depends on the skill of the designer, the estimation accuracy of the system parameter, and the like. In addition, system parameter values may change due to aging deterioration of the test system, environmental changes, and other factors, and in some cases, it may be difficult to perform system parameter estimation tests in terms of equipment. Therefore, it may be difficult to accurately achieve a predetermined simulated inertia.

On the other hand, in recent years, application of the FRIT (Fictitious Reference Iterative Tuning) method, which is a technique for simply and effectively determining the values of the control parameters of the feedback controller, has been promoted. The FRIT method is a technique for directly calculating control parameter values from closed-loop data (more specifically, time-series data such as manipulated variables and controlled variables) so that the closed-loop transfer characteristics become the desired response characteristics. (See Patent Documents 1 and 2, for example). According to the FRIT method, there is no need to identify the structure of the dynamic characteristic model of the controlled object or the values of the system parameters that characterize the dynamic characteristic model. Values for control parameters can be determined. In particular, Patent Document 2 by the applicant of the present application shows an example in which the FRIT method as described above is applied to a test system.

JP 2015-76024 A Japanese Unexamined Patent Application Publication No. 2019-21079

By the way, the design method disclosed in Patent Document 2 assumes a control device of a one-degree-of-freedom control system including only a feedback controller. In other words, in order to improve the transient response, it is conceivable to combine a feedback controller with a feedforward controller in the control device, but the design method shown in Patent Document 2 is a two-degree-of-freedom control system. No controller is assumed. Moreover, the design method disclosed in Patent Document 2 is based on the assumption that control parameters that minimize the evaluation function are calculated by a genetic algorithm, and the calculation cost is high.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a design method for a control device and an electric inertia control device that can calculate the control parameters of the control device or the electric inertia control device of a two-degree-of-freedom control system at a low computation cost.

(1) According to the present invention, a control amount (for example, a dynamo angular velocity ω(t) to be described later) of a controlled object (for example, a controlled object P to be described later) is set to a predetermined target amount (for example, a target amount r(t) to be described later). , wherein a design target (for example, a dynamo controller 5 described later) is a control amount deviation between the target amount and the control amount (for example, a control amount deviation e ( t))) that outputs a first control input (for example, a feedback input u _fb (t) described later) (for example, a feedback controller 51 described later), and a disturbance to the controlled object (for example, A second control input (for example, a feedforward input u _ff (t) to be described later) according to a measured amount (for example, a shaft torque measured amount T ₁₂ (t) to be described later) that changes according to a throttle opening command signal (to be described later) ), and a control input (for example, a control input u ( t)) is output. The design method is a state in which the values of a plurality of control parameters (for example, gains K _Pe , K _Ie , and K _Pr described later) included in the first and second controllers are set to predetermined provisional values, and the control amount , an initial controlled variable (for example, an initial controlled variable y ₀ (t) to be described later) that is N-component time-series data (N is an integer of 2 or more) of the target amount, the controlled variable deviation, and the measured amount, an initial A target amount (for example, an initial target amount r ₀ (t), which will be described later), an initial control amount deviation (for example, an initial control amount deviation e ₀ (t), which will be described later), and an initial measurement amount (for example, an initial shaft torque measurement, which will be described later) A step of obtaining a quantity T _{12 — 0} (t)) (for example, a closed-loop data obtaining step (S1) described below), and a first reference response model for the initial target quantity (for example, a reference target response model G _mr (z ^{− 1} ))) and the initial control amount is input to the first controller, the first reference output (for example, a first reference output (for example, later described (t)) and _a pseudo exogenous signal (for example, a pseudo exogenous signal T ^~ ₁₂ (t)) is input to a second reference response model (for example, a reference disturbance response model G′ _md (z ⁻¹ ) described later), N-component time-series data output from the second reference response model a step (for example, an optimization step (S5) to be described later) of determining a design value of the control parameter so that a certain second reference output (for example, a second reference output φ ₂ (t) to be described later) is close to , is provided.

(2) In this case, the design method includes an initial control input (for example, an initial control input u ₀ (t )) (for example, a closed-loop data acquisition step (S1) described later), wherein the pseudo exogenous signal is a response of the first controller to the initial control amount deviation and a response to the initial measurement amount It is preferably defined by the combination of the response of the second controller, the initial control input and the initial measured quantity.

(3) In this case, the design value is an evaluation function (for example, is preferably determined to minimize the evaluation function J) of .

(4) In this case, the controlled object includes a first inertial body (for example, the engine 10, which will be described later) that generates a torque (for example, an engine torque T ₁ (t), which will be described later) according to the disturbance, and the control input a second inertia body (for example, a dynamometer 11, which will be described later) that generates a torque (for example, a dynamo torque T ₂ (t), which will be described later) according to , and a connecting shaft 12) described later.

(5) The present invention provides a first inertial body (for example, an engine 10 to be described later) that generates a torque (for example, an engine torque T ₁ (t) to be described later) according to a disturbance (for example, a throttle opening command signal to be described later). ) and a second inertia body (for example, a dynamometer 11 to be described later) that generates a torque (for example, a dynamo torque T ₂ (t) to be described later) corresponding to a control input (for example, a control input u(t) to be described later). , a shaft (for example, a connecting shaft 12 described later) that connects the first inertial body and the second inertial body, and a control amount (for example, a dynamo angular velocity ω(t )), and a shaft torque detector (for example, an encoder 31 to be described later) that outputs the shaft torque of the shaft as a measurement amount (for example, a shaft torque measurement amount T ₁₂ (t) to be described later). For example, a multi-inertia system comprising a shaft torque controller 30 described later) is set as a controlled object (for example, a controlled object P described later), and the second inertial body with respect to the disturbance based on the controlled amount and the measured amount A method for designing an electric inertia control device that outputs the control input so that the behavior becomes a behavior of a simulated inertial body having a predetermined set inertia (for example, set inertia J _set described later), The dynamo control device 5) includes a target quantity calculation unit (for example, a target quantity calculator 50), and a first control input (for example, feedback input u _fb ( t))), and a second control input (for example, a feedforward input u _ff (t) described later) according to the measured quantity. 2 controllers (for example, a feedforward controller 52 to be described later) and a synthesizing unit (for example, a synthesizing unit 53 to be described later) that determines the control input by synthesizing the first and second control inputs. . In the _{design method} , the _control an initial controlled variable (for example, an initial controlled variable y ₀ (t) described later) that is N-component time-series data (N is an integer of 2 or more) of the amount, the target amount, the controlled variable deviation, and the measured amount; Initial target amount (for example, initial target amount r ₀ (t) described later), initial controlled variable deviation (for example, initial controlled variable deviation e ₀ (t) described later), and initial measured amount (for example, initial shaft torque described later) A step of obtaining a measured quantity T _{12 — 0} (t)) (for example, a closed-loop data obtaining step (S1) described later), and a first reference response model for the initial target quantity (for example, a reference target response model G _mr (z ⁻¹ ))) and the initial control amount is input to the first controller, the first reference output (for example, A first reference output φ ₁ (t), which will be described later, and a pseudo exogenous signal (for example, a pseudo exogenous signal T ^~ ₁₂ (t)) is input to a second reference response model (for example, a reference disturbance response model G′ _md (z ⁻¹ ) described later), N-component time-series data output from the second reference response model A step of determining the design value of the control parameter so that the second reference output (for example, a second reference output φ ₂ (t) to be described later) is close to (for example, an optimization step (S5) to be described later) and.

(6) In this case, the design method includes an initial control input (for example, an initial control input u ₀ (t )) (for example, a closed-loop data acquisition step (S1) described later), wherein the pseudo exogenous signal is a response of the first controller to the initial control amount deviation and a response to the initial measurement amount It is preferably defined by the combination of the response of the second controller, the initial control input and the initial measured quantity.

(7) In this case, the design value is an evaluation function (for example, is preferably determined to minimize the evaluation function J) of .

(8) In this case, the transfer function _Gmr of the first reference response model and the transfer function _G'md of the second reference response model are defined by "s" as the Laplace operator and "n" as a predetermined integer, It is preferable that "Lm" be a predetermined real number, "σ" be a predetermined coefficient, and "β" be a predetermined coefficient, and be represented by the following equations (1-1) and (1-2).

(9) In this case, the value of the coefficient β is preferably set such that the frequency band in which the gain is substantially constant on the low frequency side is widened in the frequency response from the measured amount to the controlled amount.

(10) In this case, the value of the coefficient σ is preferably set based on the stability margin.

(1) A design method according to the present invention includes a first controller that outputs a first control input according to a controlled variable deviation between a target variable and a controlled variable, and a measured variable that changes according to a disturbance to a controlled object. and a second controller for outputting a second control input in response to , and a two-degree-of-freedom control system for outputting a control input to be input to the controlled object based on the first and second control inputs. Further, in the design method according to the present invention, the initial control amount, the initial target amount, the initial control amount, which are the N-component time-series data in a state where the values of the plurality of control parameters included in the first and second controllers are set to the provisional values, Quantitative deviation and initial measured quantity are obtained, and these initial control quantities, etc., are close to the first reference output and the second reference output obtained using the first reference response model and the second reference response model. Determine design values for control parameters. In particular, in the present invention, when the deviation between the response of the first reference response model to the initial target amount and the initial control amount is input to the first controller, N-component time-series data output from the first controller is defined as the first reference output, and when a pseudo exogenous signal calculated based on the response of the design object to the initial controlled variable deviation and the initial measured quantity is input to the second reference response model, from this second reference response model The output N-component time-series data is defined as a second reference output. In the design method of the present invention, by defining the first reference output and the second reference output as described above, the difference between the first reference output and the second reference output is a linear function of a plurality of control parameters. Become. For this reason, the calculation for calculating the design values of the plurality of control parameters so that the first reference output and the second reference output are close to each other results in a linear optimization problem. can be done.

(2) In the design method according to the present invention, an initial control input, which is N-component time-series data in a state in which the control parameter is set to a provisional value, is further acquired, and the response of the first controller to the initial control amount deviation and the initial A pseudo-extrinsic signal is defined by a combination of the response of the second controller to the measurand, the initial control input, and the initial measurand. In the design method according to the present invention, by defining the second reference output based on such a pseudo-exogenous signal, it is possible to facilitate calculations for calculating design values of a plurality of control parameters.

(3) In the design method according to the present invention, the design value is determined so as to minimize the evaluation function, which is the N-component sum of squares of deviations between the first reference output and the second reference output. Thereby, design values of a plurality of control parameters can be easily calculated by the method of least squares.

(4) The design method according to the present invention includes a first inertia body that generates torque in response to disturbance, a second inertia body that generates torque in response to control input, and a shaft connecting them. The object of control is a multi-inertia system. This makes it possible to easily design a control device for a two-degree-of-freedom control system that controls a multi-inertia system.

(5) A design method according to the present invention includes: a target amount calculation unit that calculates a target amount for a controlled amount based on a measured amount and a set inertia; a first controller that outputs a control input, a second controller that outputs a second control input according to a measured quantity, a synthesizing unit that determines a control input by synthesizing the first and second control inputs; The object of design is an inertial control device of a two-degree-of-freedom control system. Further, in the design method according to the present invention, after acquiring the initial controlled variable, the initial target variable, the initial controlled variable deviation, and the initial measured variable, which are N-component time-series data, the response of the first reference response model to the initial target variable and Calculated based on the first reference output output from the first controller when the deviation from the initial controlled variable is input to the first controller, and the response of the design target to the initial controlled variable deviation and the initial measured variable The design values of the plurality of control parameters are determined so that the second reference output output from the second reference response model when the pseudo exogenous signal to be output from the second reference response model is input to the second reference response model. As a result, the difference between the first reference output and the second reference output becomes a linear function of the plurality of control parameters, so the design values of the plurality of control parameters can be easily calculated.

(6) In the design method according to the present invention, the initial control input is further acquired, the response of the first controller to the initial controlled variable deviation, the response of the second controller to the initial measured quantity, the initial control input, the initial A pseudo-extrinsic signal is defined by a combination of the measured quantity and In the design method according to the present invention, by defining the second reference output based on such a pseudo exogenous signal, the design values of the plurality of control parameters of the electrical inertial control device to be designed can be easily calculated. be able to.

(7) In the design method according to the present invention, the design value is determined so as to minimize the evaluation function, which is the N-component sum of squares of deviations between the first reference output and the second reference output. This makes it possible to easily calculate the design values of a plurality of control parameters of the electric inertial control device by the method of least squares.

(8) In the design method according to the present invention, the transfer function _Gmr of the first reference response model and the transfer function _G'md of the second reference response model are defined by the above equations (1-1) and (1-2). By setting the coefficients “σ” and “β” of these reference response models, it is possible to specify the frequency band, control stability, etc. that can simulate the set inertia.

(9) In the design method according to the present invention, the value of the coefficient β is set to the frequency band in which the gain is substantially constant on the low frequency side in the frequency response from the measured amount to the controlled amount, that is, the set inertia is simulated by electric inertia control. Set so that the available frequency band is wide. This makes it possible to simulate the set inertia over a wider frequency band.

(10) In the design method according to the present invention, the value of the coefficient σ is set based on the stability margin. This makes it possible to construct an electrical inertial control device with favorable stability.

1 is a diagram showing the configuration of a test system to which a design method according to an embodiment of the invention is preferably applied; FIG. 3 is a block diagram showing the configuration of an approximate model of a controlled object; FIG. 3 is a block diagram showing the configuration of a simple model of a controlled object; FIG. 1 is a block diagram showing a configuration of a two-degree-of-freedom control system constituted by a design object and its control object, and a reference trajectory defined for this two-degree-of-freedom control system; FIG. 4 is a flow chart showing a specific procedure of a design method; It is a block diagram for defining an evaluation function. FIG. 4 is a diagram showing an example of a Bode diagram showing frequency response from measured shaft torque to dynamo angular velocity; It is a figure which shows an example of the Nyquist diagram of a sensitivity function. It is a figure which shows the simulation result using the dynamo control device before design. FIG. 10 is a diagram showing a Bode diagram when the value of coefficient β is changed while the value of coefficient σ is fixed; FIG. 10 is a diagram showing a Bode diagram when the value of coefficient β is changed while the value of coefficient σ is fixed; FIG. 4 is a diagram showing the relationship between parameter Ms and coefficient σ; It is a figure which shows the simulation result using the dynamo control device after design (Example 1). (Example 2) which is a figure which shows the simulation result using the dynamo control device after design.

An embodiment of the present invention will be described in detail below with reference to the drawings.
FIG. 1 is a diagram showing the configuration of a test system S to which the design method of the present invention is preferably applied. The test system S connects an engine 10 as a first inertial body to be tested in the test system S, a dynamometer 11 as a second inertial body, and an output shaft of the engine 10 and an output shaft of the dynamometer 11. A connecting shaft 12 as a shaft, an inverter 20 connected to a dynamometer 11, a throttle actuator 21 connected to the engine 10, a shaft torque detector 30, an encoder 31, and a dynamo control connected to the inverter 20. a device 5; The test system S is a so-called engine bench system that measures various characteristics of the engine 10 by using the dynamometer 11 as a power absorber of the engine 10 while controlling the throttle opening of the engine 10 with the throttle actuator 21. .

In the following description, the control device design method of the present invention is applied to the dynamo control device 5 of the test system S, which is an engine bench system as shown in FIG. Not exclusively. The design method of the present invention is not limited to the engine bench system shown in FIG. 1, but can also be applied to a dynamo controller of a so-called drivetrain bench system in which the test object is a drivetrain.

When a torque current command signal is input as a control input from the dynamo control device 5, the inverter 20 supplies electric power corresponding to the torque current command signal to the dynamometer 11, and a dynamo torque [ Nm] is generated by the dynamometer 11 . In the following, time-series data of dynamo torque generated by the dynamometer 11 is expressed as "T ₂ (t)" as a function of time t.

When a throttle opening command signal corresponding to a command for the throttle opening of the engine 10 is input, the throttle actuator 21 controls the throttle opening of the engine 10 so as to achieve this. Accordingly, the throttle actuator 21 causes the engine 10 to generate an engine torque [Nm] corresponding to the throttle opening command signal. In the following description, the time-series data of the engine torque generated by the engine ₁₀ is expressed as "T1(t)" as a function of time t.

The shaft torque detector 30 generates a shaft torque detection signal corresponding to the shaft torque [Nm], which is the torsional torque of the connecting shaft 12 , and transmits it to the dynamo control device 5 . In the following description, the time-series data of the shaft torque detection signal from the shaft torque detector 30 is expressed as "T ₁₂ (t)" as a function of time t.

The encoder 31 generates a velocity detection signal corresponding to the angular velocity [rad/s] of the output shaft of the dynamometer 11 (hereinafter also referred to as “dynamo angular velocity”) and transmits it to the dynamo controller 5 . In the following description, the time-series data of the speed detection signal from the encoder 31 is expressed as "ω(t)" as a function of time t.

The dynamo control device 5 controls a multi-inertia system composed of the engine 10, the dynamometer 11, the connecting shaft 12, the inverter 20, the throttle actuator 21, the shaft torque detector 30, and the encoder 31. is a control device for controlling the control amount of to a predetermined target amount. In this embodiment, the dynamo control device 5 outputs a throttle opening command input as a disturbance to the controlled object based on a speed detection signal and an axial torque detection signal output from the controlled object as a controlled amount and a measured amount. A so-called electrical inertia control device that outputs a torque current command signal for the dynamometer 11 to the inverter 20 so that the dynamometer 11 behaves in response to the signal as a simulated inertia body having a predetermined set inertia will be described. That is, under the electric inertia control by the dynamo control device 5, the dynamometer 11 behaves as if it has a set inertia with respect to the engine torque generated by the engine 10 in response to the throttle opening command signal. In the following description, the time-series data of the control input output from the dynamo control device 5 is expressed as "u(t)" as a function of time t. Also, hereinafter, the control input and the dynamo torque are assumed to be equal (u(t)=T ₂ (t)).

As shown in FIG. 1, the dynamo control device 5 includes a target amount calculation unit 50 for calculating a target amount for the speed detection signal of the encoder 11 based on the shaft torque detection signal of the shaft torque detector 30, and a target amount calculation unit 50. A feedback controller 51 that outputs a feedback input according to the control amount deviation between the target amount calculated by and the speed detection signal of the encoder 11, and a feedforward control that outputs a feedforward input based on the shaft torque detection signal and a combiner 53 that determines a control input by combining the feedback input and the feedforward input.

For example, the target quantity calculation unit 50 sets the behavior of the dynamometer 11 with respect to the engine torque T ₁ so that the dynamometer 11 realizes a simulated inertia body having the set inertia J _set [kg·m ² ]. Based on the set inertia J _set and the shaft torque T ₁₂ (t), the target amount r(t) for the dynamo angular velocity ω(t) according to the following equation (2) so that the behavior of the simulated inertial body having the inertia J _set Calculate In the following, “Δ” stands for difference operator and is defined by Δ=1−z− ¹ . Also in the following, z ⁻¹ represents the delay operator and is defined by z ⁻¹ x(t)=x(t−1).

The feedback controller 51 controls the error signal e(t) (that is, e(t)=r(t)−ω(t)), which is the deviation between the target quantity r(t) and the dynamo angular velocity ω(t), to be zero. A feedback input u _fb (t) is generated based on a known feedback control law (C _e (z ⁻¹ )) so as to be (see formula (3-1) below). In this embodiment, the feedback controller 51 follows the PI control law characterized by using two control parameters, a proportional gain K _Pe and an integral gain K _Ie , as shown in the following equation (3-2): Although the case where the feedback input u _fb (t) is generated according to the error signal e(t) will be described, the present invention is not limited to this. The control structure of the feedback controller 51 may be any one whose input/output characteristics are defined by one or more control parameters.

The feedforward controller 52 generates a feedforward input u _ff (t) corresponding to the shaft torque T ₁₂ (t) (see formula (4-1) below). In this embodiment, the forward controller 52, as shown in the following equation (4-2), follows the P control law characterized by using the proportional gain K _Pr , which is one control parameter, and the shaft torque T ₁₂ A case of generating a feedforward input u _ff (t) according to (t) will be described, but the present invention is not limited to this. The control structure of feedforward controller 52 may be any one whose input and output characteristics are defined by one or more control parameters.

The synthesizing unit 53 synthesizes the feedback input u _fb (t) and the feedforward input u _ff (t) to generate the control input u(t), and inputs it to the controlled object. In the present embodiment, the synthesizing unit 53 subtracts the feedforward input u _ff (t) from the feedback input u _fb (t) to obtain the control input u(t) as shown in Equation (5) below. Although the case of generation will be described, the present invention is not limited to this. For example, the synthesis unit 53 may calculate the control input u(t) by adding the feedback input u _fb (t) and the feedforward input u _ff (t).

The dynamo control device 5 is characterized by its input/output characteristics by means of a target quantity calculator 50 that calculates a target quantity r(t) according to the set inertia J _set and three control parameters (K _Pe , K _Ie , K _Pr ). A feedback controller 51, a feedforward controller 52, and a synthesizing unit 53 are provided so that the behavior of the dynamometer 11 with respect to the engine torque T ₁ (t) is the behavior of a simulated inertia body having a set inertia J _set . Output the input u(t).

FIG. 2 is a block diagram showing the configuration of an approximation model of the controlled object P of the dynamo control device 5. As shown in FIG. First, in the test system S of FIG. 1, the controlled object P of the dynamo controller 5 can be approximated by a two-inertia system as shown in FIG. That is, the object P to be controlled includes a first inertial body having inertia J ₁ [kg·m ₂ ] and generating engine torque T ₁ (t), and an inertial body having inertia J ₂ [kg·m ₂ ]. A second inertia body that generates dynamo torque T ₂ (t) corresponding to the control input u(t) is defined as a spring loss C ₁₂ [Nm·s/rad] and a spring stiffness K ₁₂ [Nm/rad ] can be approximated by a two-inertia system connected by a shaft with When designing the dynamo control device 5 according to the design method of the present invention, the detailed configuration of the controlled object P and the values of system parameters such as inertia J ₁ and J ₂ , spring loss C ₁₂ , and spring stiffness K ₁₂ are Suppose it is unknown.

The equations of motion of the two-inertia system shown in FIG. 2 are represented by the following equations (6-1) and (6-2).

In addition, in the above equations of motion (6-1) and (6-2), the five transfer functions A(s), B ₁ (s), B ₂ (s), B ₃ (s), and B ₄ (s) is expressed by the following equations (7-1) to (7-5) using system parameters. Also, in the following equation, ωr is the resonance frequency, which is represented by the following equation (7-6) using system parameters.

FIG. 3 is a block diagram showing the configuration of a simple model of the controlled object P. As shown in FIG. This simple model is a further simplification of the approximation model shown in FIG. In the design method according to the present invention, it is possible to design the dynamo control device 5, which is the object of design, without using an approximate model of the object of control P as shown in FIG. However, as will be explained later, some kind of model of the controlled object P is necessary in order to evaluate the Bode diagram, stability margin, etc. of the designed object. Therefore, in the present embodiment, the simplified model shown in FIG. 3, which is a further simplified version of the approximate model shown in FIG.

FIG. 4 shows the configuration of a two-degree-of-freedom control system constituted by the dynamo controller 5, which is the object of design, and its controlled object P, and the reference trajectory y _ref (t) defined for this two-degree-of-freedom control system. It is a block diagram showing.

As shown in FIG. 4 , the dynamo control device 5 includes a target quantity calculator 50 , a feedback controller 51 , a feedforward controller 52 and a combiner 53 . In the controlled object P, the transfer function G(z ⁻¹ ) from the control input u(t) and the shaft torque T ₁₂ (t) to the controlled variable y(t) (that is, the dynamo angular velocity ω(t)) is unknown. Suppose that

In the block diagram shown in FIG. 4, the controlled variable y(t), which is the output of the controlled object P, is represented by the following equation (8).

In the two-degree-of-freedom control system shown in FIG. ₄ , the reference disturbance response model G _md (z ⁻¹ ) and the reference target response model G _mr (z ⁻¹ ), the reference trajectory y _ref (t) for the controlled variable y(t) is expressed by the following equation (9). Here, the reference disturbance response model G _md (z ⁻¹ ) and the reference target response model G _mr (z ⁻¹ ) are reference response models introduced to specify the response to the disturbance and the target amount by the design object.

Here, the feedback controller 51 and the feedforward controller 52 that satisfy the given reference disturbance response, that is, the control amount y(t) calculated by the equation (8) and the reference trajectory calculated by the equation (9) Defining the transfer functions of the feedback controller 51 and the feedforward controller 52 equal to y _ref (t) as C ^* _e (z ⁻¹ ) and C ^* _r (z ⁻¹ ), the reference disturbance response model G _md (z ⁻¹ ) is rewritten by the following equation (10).

Here, in the above equation (10), the function G′ _md (z ⁻¹ ) is a complementary sensitivity function as shown in the following equation (11). Hereinafter, the function G′ _md (z ⁻¹ ) redefined using the transfer function C ^* _e (z ⁻¹ ) as shown in Equation (11) below will be referred to as a reference disturbance response model.

Using the reference disturbance response model G′ _md (z ⁻¹ ) and the reference target response model G _mr (z ⁻¹ ) redefined by Equation (11) as described above, the reference trajectory shown in Equation (9) y _ref (t) is rewritten by the following equation (12).

In the ^design _{method according} to the present ^embodiment , a set of feedback controller 51 (C ^* _{e (} ^{z −1} ₎ ) and Design the feedforward controller 52 (C ^* _r (z ^-1 )).

In this embodiment, the reference target response model G _mr (z ⁻¹ ) and the reference disturbance response model G′ _md (z ⁻¹ ) are given by the following equations (13-1) and (13-2), respectively, in a continuous-time system. is defined to be a binomial model as shown in In the following equations (13-1) and (13-2), "n" is a predetermined integer and "Lm" is an estimated dead time, which are set to predetermined values. Also, in the following equations (13-1) and (13-2), "σ" is a coefficient related to the rise time, "β" is a coefficient related to the target amount response, and each value is set by the procedure described later. be done.

FIG. 5 is a flow chart showing a specific procedure for calculating the design values of the control parameters of the dynamo controller 5 according to the design method of the present invention.

First, in the closed-loop data acquisition step (S1), the designer sets the values of a plurality of control parameters (K _Pe , K _Ie , K _Pr ) included in the feedback controller 51 and the feedforward controller 52 to predetermined provisional values. while inputting disturbance (throttle opening command signal and corresponding engine torque T ₁ (t)) to the controlled object P, the control amount y (t) and the target amount r (t) , control amount deviation e(t), shaft torque measurement amount T ₁₂ (t), and control input u(t). Here, the N-component time-series data is a set of data at N points (t=Ts, 2Ts, . . . , NTs) determined at intervals of a predetermined sampling time Ts. Below, N-component time-series data of the controlled variable y(t), the target variable r(t), the controlled variable deviation e(t), the measured shaft torque T ₁₂ (t), and the control input u(t) are expressed as initial control variable y ₀ (t), initial target variable r ₀ (t), initial control _variable deviation e ₀ (t), initial shaft torque measurement T 12 — 0 (t), and initial control input u, respectively, as a function of time t It is written as ₀ (t).

Next, in the pseudo input signal calculation step (S2), the designer uses the N-component time-series data based on the initial control amount deviation e ₀ (t) and the initial shaft torque measurement amount T ₁₂ — 0 (t) acquired in S1. A pseudo input signal u ^~ (t) is calculated. More specifically, this pseudo input signal u ^~ (t) is the design target for the initial control amount deviation e ₀ (t) and the initial shaft torque measurement amount T _{12_0} (t), as shown in the following equation (14). (that is, the output of the design object when the initial controlled _variable deviation e ₀ (t) and the initial measured shaft torque T 12 — 0 (t) are input to the design object). That is, the pseudo input signal u ^~ (t) subtracts the response of the feedforward controller 52 to the initial measured shaft torque T 12 — 0 (t) from the response of the feedback controller 51 to the initial controlled _variable deviation e ₀ (t). defined by

Next, in the pseudo external signal calculation step (S3), the designer calculates the pseudo input signal u ^~ (t) calculated in S2, the initial control input u ₀ (t) obtained in S1, and the initial measured shaft torque T ₁₂ ^— 0 (t), a pseudo exogenous signal T ₁₂ (t), which is N-component time-series data, is calculated. More specifically, the pseudo exogenous signal T ₁₂ (t) is the sum of the initial control input u ₀ (t) and the initial shaft torque measurement T ₁₂ ^— 0 (t) as shown in the following equation (15). , minus the pseudo input signal u ^~ (t).

Next, in the coefficient setting step (S4), the designer sets the reference target response model G _mr (z ⁻¹ ) and the reference disturbance response model G′ _md (z ⁻¹ ) are set to predetermined provisional values (σ _i , β _j ). Here, i is an integer between 1 and Mi ("Mi" is an arbitrary integer equal to or greater than 1), and j is an integer between 1 and Mj ("Mj" is an arbitrary integer equal to or greater than 1). Also, the total number of combinations M of the provisional values (σ _i , β _j ) of the coefficients is an arbitrary integer equal to or greater than 1 (that is, Mi×Mj=M).

Next, in the optimization step (S5), the designer uses the pseudo external signal T ^~ ₁₂ (t) calculated in S3, the reference target response model G _mr (z ^-1 ) and the reference disturbance Design values of control parameters (K _Pe , K _Ie , K _Pr ) that minimize the evaluation function J defined based on the response model G′ _md (z ⁻¹ ) are calculated.

FIG. 6 is a block diagram for defining the evaluation function J referred to in S5. More specifically, in FIG. 6, in the block diagram shown in FIG. 4, the controlled object P is defined as initial data outputting the sum of the initial control input u ₀ (t) and the initial measured shaft torque T _{12 —} 0 (t). 3 is a block diagram obtained by replacing with an output unit 6. FIG.

The difference between the reference trajectory y _ref (t) defined in FIG. 4 and the initial control amount y ₀ (t) is the pseudo input signal u ^˜ (t) and the pseudo exogenous Introduce the signal T ₁₂ (t) ^and the reference target response model G _mr (z ⁻¹ ) and the reference disturbance response model G′ _md (z ⁻¹ ) shown in equations (13-1) and (13-2) above. Then, the difference signal φ(t) between the first reference output φ ₁ (t) and the second reference output φ ₂ (t) shown in FIG. 6 is rewritten (see equation (16) below).

Here, the first reference output φ ₁ (t) is the response of the reference target response model G _mr (z ⁻¹ ) to the initial target quantity r ₀ (t) (that is, the initial target The output of the reference target response model G _mr (z ⁻¹ ) when the quantity r ₀ (t) is input to the reference target response model G _mr (z ⁻¹ ) and the initial controlled variable y ₀ (t) is defined as N-component time-series data output from the feedback controller 51 when the deviation between is input to the feedback controller 51 .

The second reference output φ ₂ (t) is obtained by inputting the pseudo extraneous signal T ₁₂ (t) into the reference disturbance response model G′ _md (z ⁻¹ ) as shown in the following equation (18) ^. , is defined as N-component time-series data output from this reference disturbance response model G′ _md (z ⁻¹ ).

Therefore, the problem of determining design values (K ^* _Pe , K ^* _Ie , K ^* _Pr ) of a plurality of control parameters so that the reference trajectory y _ref (t) and the initial control amount y ₀ (t) are close to each other is given below. As shown in Equation (19), design values (K ^* _Pe , K ^* _Ie , K ^* _Pr ).

In the optimization step (S5), the designer calculates design values (K ^* _Pe , K ^* _Ie , K ^* _Pr ) of a plurality of control parameters that minimize the evaluation function J shown in Equation (19) above. do.

Note that in this embodiment, by defining the first reference output φ ₁ (t) and the second reference output φ ₂ (t) as described above, the evaluation function J is obtained as shown in the following equation (20-1): Then, it becomes a quadratic function of the parameter vector θ (see formula (20-2) below) based on a plurality of control parameters. In the following formula (20-1), the vector ν is an N-component vector defined by the following formula (20-3), and in the following formula (20-3), the N-component time series data u _T0 (t) is , is the output of the initial data output unit 6 as shown in the following formula (20-4), and in the following formula (20-1), the matrix Φ is shown in the following formulas (20-5) and (20-6) is a 3×N matrix.

In addition, since the evaluation function J becomes a downwardly convex function with respect to the plurality of control parameters, the design values of the plurality of control parameters that minimize this evaluation function J (K ^* _Pe , K ^* _Ie , K ^* _Pr ) Since the problem of determining is reduced to a linear optimization problem, the design values (K ^* _Pe , K ^* _Ie , K ^* _Pr ) of a plurality of control parameters by applying the least squares method are given by the following equation (21) Calculated by Therefore, in the optimization step (S5), the designer calculates the design values (K ^* _Pe , K ^* _Ie , K ^* _Pr ) of the plurality of control parameters by solving the equation shown in Equation (21) below.

Next, in S6, the designer determines whether the design values (K ^* _Pe , K ^* _Ie , K ^* _Pr ) of the control parameters have been obtained under the M sets of temporary values ( _σi , _βj ). . If the determination result in S6 is NO, the designer returns to S4 and calculates the design values of the control parameters under other provisional values (σ _i , β _j ). If the determination result of S6 is YES, the designer proceeds to the process of S7. As a result, M sets of dynamo controllers characterized by M sets of control parameter design values (K ^* _Pe , K ^* _Ie , K ^* _Pr ) are derived.

Next, in the optimum controller selection step (S7), the designer selects one of the M sets of dynamo controllers as the optimum controller. More specifically, the designer evaluates the Bode plots and stability margins of the M sets of dynamo controllers using the simple model of the controlled object described with reference to FIG. ) are determined, and the dynamo controller derived under these optimum values ^{(σ*, β* ) is selected} as the optimum controller.

FIG. 7 is a diagram showing an example of a Bode diagram showing the frequency response from the measured shaft torque _T12 to the dynamo angular velocity ω. As shown in FIG. 7, the frequency response from the measured shaft torque T12 to the dynamo angular velocity ω is approximately constant with a gain of 20log ₁₀ (1/J _set ) on the low frequency side, and a gain of _{20log 10} on the high frequency side. There is a frequency band that is approximately constant at (1/J ₂ ). Also, as shown in FIG. 7, the shape of this frequency response varies with the factor β.

Here, the frequency band in which the gain is approximately constant at 20log ₁₀ (1/J _set ) corresponds to the frequency band in which the set inertia J _set can be achieved. Therefore, in the optimum controller selection process, the designer selects the gain of _{20 log 10 (} 1/ _{J set} ), the optimum value β ^* is determined as the widest frequency band. This makes it possible to design a dynamo controller that can achieve the set inertia J _set over a wide range.

As shown in FIG. 7, even if the value of the coefficient σ is decreased, the frequency band in which the set inertia J _set can be realized can be widened, but if the value of the coefficient σ is decreased too much, it becomes unstable. Therefore, the optimum value σ ^* of the coefficient σ is preferably set based on the stability margin as described with reference to FIG.

FIG. 8 is a diagram showing an example of a Nyquist diagram of the sensitivity function S(jω). In the optimum controller selection step, the designer uses the sensitivity function S(jω) defined based on the simple model shown in FIG. A parameter Ms relating to the margin is calculated (see formula (22-1) below), and the one that gives a parameter Ms close to the value designated by the designer among the plurality of provisional values σ _i is determined as the optimum value σ ^* . As a result, the dynamo controller can be designed in consideration of the stability margin.

In the optimum controller selection step, the designer determines the optimum values (σ ^* , β ^* ) of the two coefficients by the above procedure, and then derives them based on these optimum values (σ ^* , β ^* ). The dynamo controller is selected as the optimum controller, and the process shown in FIG. 5 is terminated.

Next, the effect of the design method according to this embodiment will be described with reference to the simulation results shown in FIGS. 9 to 13. FIG.

FIG. 9 is a diagram showing a simulation result using a dynamo control device before design. Here, the dynamo controller before design is a dynamo controller in which the values of the control parameters (K _Pe , K _Ie , K _Pr ) are set to provisional values, and is used in the closed-loop data acquisition process. Say. More specifically, the values of each control parameter were K _Pe =20.5, K _Ie =20.5, and K _Pr =0.0. In FIG. 9, the control amount y and its target amount r are shown at the top, the shaft torque measurement amount _T12 is shown at the middle, and the control input u is shown at the bottom. As shown in FIG. 9, according to the pre-designed dynamo controller, it takes time for the controlled variable y to follow the target variable r.

10 and ₁₁ show the frequency response from the measured shaft torque T12 to the dynamo angular velocity ω when the value of the coefficient β is varied within a predetermined range while the value of the coefficient σ is fixed at a predetermined value. It is a Bode plot. In the example of FIG. 10, the value of the coefficient σ is 0.005, and the value of the coefficient β is changed from 1.5 to −1.5 in increments of 0.5. The value of σ is 0.2, and the value of coefficient β is changed from 1.5 to −1.5 in increments of 0.5. As is clear from comparison between FIGS. 10 and 11, the frequency band in which the set inertia can be achieved varies depending on the coefficient σ. Also, as shown in FIGS. 10 and 11, the frequency band in which the set inertia can be achieved slightly changes depending on the coefficient β. A case where the value of the coefficient β is set to 0.0 based on the results shown in FIGS. 10 and 11 will be described below.

FIG. 12 is a diagram showing the relationship between the parameter Ms defined by the above equation (22-1) and the coefficient σ. The stability margin increases as the parameter Ms decreases. Also, as shown in FIG. 12, it was confirmed that the parameter Ms tends to decrease as the value of the coefficient σ increases. In the following description, the value of the coefficient σ is set to 0.005 or 0.2 in order to compare control results.

When σ=0.005 and β=0.0, the design values of each control parameter are K _Pe =1.5×10 ³ , K _Ie =1.5×10 ³ and K _Pr =0.68. became. This is hereinafter referred to as the dynamo control system of the first embodiment. When σ=0.2 and β=0.0, the design values of each control parameter were K _Pe =35.1, K _Ie =232.4, and K _Pr =0.0. This is hereinafter referred to as a dynamo control system of the second embodiment.

13 and 14 are diagrams showing examples of simulation results using the designed dynamo control device, respectively. More specifically, FIG. 13 shows the simulation results using the dynamo control system of the first embodiment, and FIG. 14 shows the simulation results using the dynamo control system of the second embodiment.

As is clear from a comparison of FIGS. 13 and 14 with FIG. 9, according to the dynamo controllers of the first and second embodiments, the controlled variable y can quickly follow the target variable r. can be done. Note that the control amount y can follow the target amount r more quickly in the first embodiment than in the second embodiment. This is because the coefficient .sigma. is set to a smaller value in the first embodiment than in the second embodiment. However, if the coefficient σ is set to a small value, the control input u becomes oscillatory. For this reason, it was confirmed that the second embodiment can control more stably than the first embodiment.

As mentioned above, although one Embodiment of this invention was described, this invention is not limited to this. Detailed configurations may be changed as appropriate within the scope of the present invention.

For example, in the above embodiment, the design method of the present invention is applied to the dynamo controller of the test system S that controls a multi-inertia system, but the present invention is not limited to this. The design method of the present invention can be applied to any control device as long as it is a two-degree-of-freedom control system.

S... Test system P... Control object 10... Engine (first inertial body)
11... Dynamometer (second inertial body)
12... Connecting shaft (shaft body)
20... Inverter 21... Throttle actuator 30... Shaft torque detector 31... Encoder (speed detector)
5 ... Dynamo control device (design object)
50... Target amount calculator 51... Feedback controller (first controller)
52 ... feedforward controller (second controller)
53... Synthesizer

Claims (10)

A control device design method for controlling a control amount of a controlled object to a predetermined target amount, comprising:
The object of design includes a first controller that outputs a first control input according to the control amount deviation between the target amount and the control amount, and a first controller that outputs a first control input according to a disturbance to the control object. a second controller that outputs two control inputs, and outputs a control input to the controlled object based on the first and second control inputs;
N-component time series of the controlled variable, the target variable, the controlled variable deviation, and the measured variable in a state in which values of a plurality of control parameters included in the first and second controllers are set to predetermined provisional values; a step of acquiring an initial controlled variable, an initial target variable, an initial controlled variable deviation, and an initial measured variable, which are data (N is an integer equal to or greater than 2);
N-component time-series data that is output from the first controller when the deviation between the response of the first reference response model to the initial target amount and the initial control amount is input to the first controller Output from the second reference response model when a reference output and a pseudo exogenous signal calculated based on the response of the design object to the initial controlled variable deviation and the initial measured quantity are input to the second reference response model and determining the design value of the control parameter so that the second reference output, which is the N-component time-series data obtained by the second reference output, is close to the second reference output. further comprising the step of obtaining an initial control input that is N-component time-series data of the control input with the control parameter set to the provisional value;
The pseudo exogenous signal is obtained by synthesizing the response of the first controller to the initial controlled variable deviation, the response of the second controller to the initial measured quantity, the initial control input, and the initial measured quantity. 2. The method of designing a control device according to claim 1, characterized in that it is defined by: 3. The design value according to claim 1, wherein the design value is determined so as to minimize an evaluation function that is an N-component sum of squares of deviations between the first reference output and the second reference output. How to design a controller. The controlled object includes a first inertial body that generates torque according to the disturbance, a second inertial body that generates torque according to the control input, and a shaft that connects the first and second inertial bodies. 4. The method for designing a control device according to any one of claims 1 to 3, wherein the system is a multi-inertia system comprising: a first inertia body that generates torque in response to a disturbance; a second inertia body that generates torque in response to a control input; a shaft that connects the first inertia body and the second inertia body; A multi-inertia system comprising a speed detector that outputs the speed of two inertial bodies as a controlled variable and a shaft torque detector that outputs the shaft torque of the shaft as a measured quantity, wherein the controlled variable and the measured A method of designing an electric inertia control device for outputting the control input so that the behavior of the second inertial body with respect to the disturbance becomes the behavior of a simulated inertial body having a predetermined set inertia based on the quantity,
The object of design includes a target amount calculator that calculates a target amount for the controlled amount based on the measured amount and the set inertia, and a first control input according to the controlled amount deviation between the target amount and the controlled amount. a second controller that outputs a second control input according to the measured quantity; and a synthesizing unit that determines the control input by synthesizing the first and second control inputs , and
N-component time series of the controlled variable, the target variable, the controlled variable deviation, and the measured variable in a state in which values of a plurality of control parameters included in the first and second controllers are set to predetermined provisional values; a step of acquiring an initial controlled variable, an initial target variable, an initial controlled variable deviation, and an initial measured variable, which are data (N is an integer equal to or greater than 2);
N-component time-series data that is output from the first controller when the deviation between the response of the first reference response model to the initial target amount and the initial control amount is input to the first controller Output from the second reference response model when a reference output and a pseudo exogenous signal calculated based on the response of the design object to the initial controlled variable deviation and the initial measured quantity are input to the second reference response model a second reference output, which is N-component time-series data obtained by the second reference output; further comprising the step of obtaining an initial control input that is N-component time-series data of the control input with the control parameter set to the provisional value;
The pseudo exogenous signal is obtained by synthesizing the response of the first controller to the initial controlled variable deviation, the response of the second controller to the initial measured quantity, the initial control input, and the initial measured quantity. 6. A method of designing an electric inertial control device according to claim 5, characterized in that it is defined by: 7. The design value according to claim 5, wherein the design value is determined so as to minimize an evaluation function that is an N-component sum of squares of deviations between the first reference output and the second reference output. How to design an electric inertial control device. The transfer function _Gmr of the first reference response model and the transfer function _G'md of the second reference response model are defined by "s" as the Laplace operator, "n" as a predetermined integer, and "Lm" as a predetermined Any one of claims 5 to 7, characterized by being represented by the following equations (1-1) and (1-2) with a real number, "σ" as a predetermined coefficient, and "β" as a predetermined coefficient A method of designing an electric inertial control device according to .

9. The value of the coefficient β is set so as to widen the frequency band in which the gain is substantially constant on the low frequency side in the frequency response from the measured amount to the controlled amount. design method for electric inertial control devices. 10. The method of designing an electric inertial control device according to claim 8, wherein the value of the coefficient [sigma] is set based on a stability margin.

Images