阅读说明：本技术 一种基于迭代学习自适应mrac控制器的电机控制方法 (Motor control method based on iterative learning self-adaptive MRAC controller ) 是由 史敬灼 徐丹旸 刘姝贝 刘悦琪 徐浩然 于 2021-08-11 设计创作，主要内容包括：本发明属于电机控制技术领域,具体涉及一种基于迭代学习自适应MRAC控制器的电机控制方法。本发明采用迭代学习自适应MRAC控制器的电机的转速进行闭环控制；其中,所述迭代学习自适应MRAC控制器包括MRAC控制器和迭代学习控制器；所述迭代学习控制器根据上一迭代控制过程中迭代学习控制器的输出值、上一迭代控制过程中的转速误差值、以及当前迭代控制过程中的转速给定值,对当前迭代控制过程中MRAC控制器的控制参数的自适应律中所包括的自适应增益进行自动调节。采用迭代学习思想来自适应调整正对角矩阵Γ值的方法,不仅能够解决如何确定Γ值的问题,而且能够有效改善系统控制性能,通过学习来增强控制器对对象模型误差的适应能力。(The invention belongs to the technical field of motor control, and particularly relates to a motor control method based on an iterative learning self-adaptive MRAC controller. The invention adopts the rotation speed of the motor of the iterative learning self-adaptive MRAC controller to carry out closed-loop control; wherein the iterative learning adaptive MRAC controller comprises an MRAC controller and an iterative learning controller; and the iterative learning controller automatically adjusts the adaptive gain included in the adaptive law of the control parameters of the MRAC controller in the current iterative control process according to the output value of the iterative learning controller in the previous iterative control process, the rotating speed error value in the previous iterative control process and the rotating speed given value in the current iterative control process. The method for adaptively adjusting the gamma value of the positive diagonal matrix by adopting the iterative learning idea not only can solve the problem of how to determine the gamma value, but also can effectively improve the control performance of the system, and enhances the adaptability of the controller to the model error of the object by learning.)

1. A motor control method based on an iterative learning adaptive MRAC controller is characterized by comprising the following steps:

closed-loop control is carried out by adopting the rotation speed of a motor of the iterative learning self-adaptive MRAC controller;

wherein the iterative learning adaptive MRAC controller comprises an MRAC controller and an iterative learning controller; the MRAC controller adjusts and controls the rotating speed of the motor in the current iteration control process; the iterative learning controller automatically adjusts the self-adaptive gain included in the self-adaptive law of the control parameters of the MRAC controller in the current iterative control process according to the output value of the iterative learning controller in the previous iterative control process, the rotating speed error value in the previous iterative control process and the rotating speed given value in the current iterative control process, wherein the rotating speed error value is the difference value of the rotating speed given value and the rotating speed actual value.

2. The iterative learning adaptive MRAC controller based motor control method of claim 1, wherein the iterative learning controller automatically adjusts the adaptive gain using the following formula:

wherein r is₀Is an initial value of the adaptive gain; r is_k(i) And y_rk(i) Respectively setting values of the adaptive gain r value and the rotating speed at the moment i in the kth iterative control process; coefficient lambda_PA proportional learning gain; r: (_k-1)(i) The value of the adaptive gain r at the moment i in the k-1 iteration control process is obtained; e.g. of the type_I(k-1)(i +1) is a rotating speed error value at the moment of i +1 in the k-1 th iteration control process; y is_rk(i) And setting the rotating speed at the moment i in the k iterative control process.

3. The iterative learning adaptive MRAC controller based motor control method of claim 1, wherein the adaptive law of the control parameters of the MRAC controller is:

wherein theta is a control parameter, and theta^T＝[k₀ d₀]，k₀For feed forward gain, d₀Is the feedback gain; Γ is a positive oriented diagonal matrix,r is the adaptive gain; e.g. of the type_k(i) Is the output error value at time i in the kth iterative control process, and e_k(i)＝y_mk(i)-y_k(i)，y_mk(i)、y_k(i) Respectively obtaining a reference model output value and a rotating speed actual value at the moment i in the kth iterative control process; the MRAC control includes the reference model.

4. The iterative learning-based adaptive MRAC controller motor control method of claim 3, wherein the adaptive law of the MRAC controller control parameters is designed by selecting a Lyapunov function as:

in the formula, V is the selected Lyapunov function; p and gamma are positive definite symmetric matrixes; e.g. of the type_kIs an augmented error vector;is a parametric error vector.

5. The iterative learning adaptive MRAC controller-based motor control method according to any of claims 1-4, wherein the motor is an ultrasonic motor.

Technical Field

The invention belongs to the technical field of motor control, and particularly relates to a motor control method based on an iterative learning self-adaptive MRAC controller.

Background

The ultrasonic motor has a series of advantages of compact structure, easy miniaturization, fast response and braking, good control characteristic, high positioning precision, small noise and the like, and has wide application scenes and use values in the high and new technical fields of aviation, aerospace, medical treatment, precision instruments and the like.

Due to the special operation mechanism of the ultrasonic motor, the operation characteristic of the ultrasonic motor shows obvious nonlinear and time-varying characteristics, and ideal control performance is not easy to obtain. In an effort to overcome these disadvantages of the ultrasonic motor itself, and to obtain a control performance and operational stability that meet the application expectations, research into a control strategy thereof has gradually become complicated. Many complex controllers, such as neural network controllers, fuzzy neural network controllers, etc., have been proposed and applied to ultrasonic motors in sequence.

However, for reasons such as cost, debugging, and system maintenance, it is always desirable that the control system be simple in structure, and for these reasons, the model reference adaptive control strategy (MRAC) based on input and output variables proposed in 1978 by Narendra Kumpati s and valvani Lena s has been gradually applied to the ultrasonic motor control. The method has wide application range and mature design method, and is widely applied all the time.

However, it also presents a significant problem: in the control strategy, the value of the adaptive gain matrix gamma determines the adaptive adjustment rate of the controller parameter, and directly influences the control performance of the system. However, the control strategy does not provide a method for determining the Γ value, and usually a control system simulation method is adopted to try and correct the Γ value according to expected control performance, and finally a proper value is determined through field debugging. This is often the most time consuming step in applying the control strategy to a practical process and can also lead to undesirable control performance.

Disclosure of Invention

The invention provides a motor control method based on an iterative learning self-adaptive MRAC controller, which is used for solving the problem of non-ideal control performance caused by the fact that a gamma value cannot be accurately given when an MRAC controller is used for controlling a motor.

In order to solve the technical problems, the technical scheme and the corresponding beneficial effects of the technical scheme are as follows:

the invention provides a motor control method based on an iterative learning self-adaptive MRAC controller, which comprises the following steps:

closed-loop control is carried out by adopting the rotation speed of a motor of the iterative learning self-adaptive MRAC controller;

wherein the iterative learning adaptive MRAC controller comprises an MRAC controller and an iterative learning controller; the MRAC controller adjusts and controls the rotating speed of the motor in the current iteration control process; the iterative learning controller automatically adjusts the self-adaptive gain included in the self-adaptive law of the control parameters of the MRAC controller in the current iterative control process according to the output value of the iterative learning controller in the previous iterative control process, the rotating speed error value in the previous iterative control process and the rotating speed given value in the current iterative control process, wherein the rotating speed error value is the difference value of the rotating speed given value and the rotating speed actual value.

The beneficial effects of the above technical scheme are: the invention borrows the idea of iterative learning control, combines the iterative learning controller with the MRAC controller, and utilizes the iterative learning controller to automatically adjust the self-adaptive gain of the MRAC controller on line, namely, the method of adaptively adjusting the positive diagonal matrix gamma value by adopting the iterative learning idea not only can solve the problem of how to determine the gamma value, but also can effectively improve the control performance of the system, and enhances the adaptability of the controller to object model errors by learning.

Further, the iterative learning controller automatically adjusts the adaptive gain using the following formula:

wherein r is₀Is an initial value of the adaptive gain; r is_k(i) And y_rk(i) Respectively setting values of the adaptive gain r value and the rotating speed at the moment i in the kth iterative control process; coefficient lambda_PA proportional learning gain; r: (_k-1)(i) The value of the adaptive gain r at the moment i in the k-1 iteration control process is obtained; e.g. of the type_I(k-1)(i +1) is a rotating speed error value at the moment of i +1 in the k-1 th iteration control process; y is_rk(i) And setting the rotating speed at the moment i in the k iterative control process.

Further, the adaptive law of the control parameters of the MRAC controller is as follows:

wherein theta is a control parameter, and theta^T＝[k₀ d₀]，k₀For feed forward gain, d₀Is the feedback gain; Γ is a positive oriented diagonal matrix,r is the adaptive gain; e.g. of the type_k(i) Is the output error value at time i in the kth iterative control process, and e_k(i)＝y_mk(i)-y_k(i)，y_mk(i)、y_k(i) Respectively obtaining a reference model output value and a rotating speed actual value at the moment i in the kth iterative control process; the MRAC controller includes the reference model.

Further, a lyapunov function selected when the adaptive law of the control parameters of the MRAC controller is designed is:

in the formula, V is the selected Lyapunov function; p and gamma are positive definite symmetric matrixes; e.g. of the type_kIs an augmented error vector;is a parametric error vector.

Further, the motor is an ultrasonic motor.

Drawings

FIG. 1 is a control block diagram of a rotational speed closed-loop control system of an ultrasonic motor of the present invention;

FIG. 2(a) is a graph of a step response of the rotation speed of an ultrasonic motor when the method of the present invention is used to control the rotation speed of the ultrasonic motor;

FIG. 2(b) is a graph showing the variation of r value when the method of the present invention is used to control the rotation speed of the ultrasonic motor;

FIG. 2(c) is a graph showing the feedforward gain k of the controller when the method of the present invention is used to control the rotation speed of the ultrasonic motor₀A graph of the variation of the values;

FIG. 2(d) is a graph showing the feedback gain d of the controller when the method of the present invention is used to control the rotation speed of the ultrasonic motor₀Graph of the change in value.

Detailed Description

The basic concept of the invention is as follows: the invention borrows the idea of iterative learning control, combines the iterative learning controller with the MRAC controller, and utilizes the iterative learning controller to automatically adjust the self-adaptive gain of the MRAC controller on line so as to self-adaptively adjust the value of the positive diagonal matrix gamma through the iterative learning idea.

A motor control method for an iterative learning adaptive MRAC controller according to the present invention will be described in detail with reference to the accompanying drawings and embodiments.

The method comprises the following steps:

in the embodiment of the motor control method based on the iterative learning adaptive MRAC controller, a control block diagram of a rotating speed closed-loop control system of an ultrasonic motor, which is realized by combining the iterative learning controller with the MRAC controller, is shown in fig. 1, and the controller obtained by combining the iterative learning controller with the MRAC controller is referred to as the iterative learning adaptive MRAC controller.

The content within the dashed-dotted line box in fig. 1 is a control block diagram for controlling the rotation speed of the ultrasonic motor by using a standard MRAC controller, and the ultrasonic motor and its driving circuit together constitute a controlled object of the MRAC controller. The iterative learning controller is used for adjusting and controlling the rotating speed of the motor in the current iterative control process. In FIG. 1, y_rk(i)、y_mk(i)、y_k(i)、e_k(i)、e_Ik(i)、r_k(i) Respectively is a given value of the rotating speed at the moment i, a reference model output value, an actual value of the rotating speed of the motor, an output error value, a rotating speed error value and an output value of the iterative learning controller in the process of the kth iterative control, and the values are as follows:

e_Ik(i)＝y_rk(i)-y_k(i) (1)

e_k(i)＝y_mk(i)-y_k(i) (2)

as previously mentioned, the portion of fig. 1 within the dashed box is a standard input and output variable based MRAC speed controller. The controller includes a feed forward gain k₀And a feedback gain d₀Equal two adjustable parameters, using adaptive law pair k₀And d₀And (3) carrying out online adjustment, theoretically realizing complete matching of the reference model and the controlled motor, and enabling the response process of the motor rotating speed to be consistent with the output of the reference model so as to achieve the expected control performance.

The MRAC controller is first designed according to a classical design method.

Choosing a Lyapunov (Lyapunov) function as:

in the formula, P and gamma are positive definite symmetric matrixes; e.g. of the type_kIs an augmented error vector;is a parametric error vector.

The derived adaptive law for the control parameter θ is:

in the formula, e_k(i) The output error at the time point i in the k-th iteration control process is expressed by equation (2).

Taking gamma as the following positive definite diagonal matrix:

in the formula, the adaptive gain r is a positive real number.

The general formula (5) andθ^T＝[k₀d₀]substitution of formula (4) gives k₀And d₀The self-adaptive laws of (1) are respectively:

in the formula, e_k(i)、y_rk(i) And y_k(i) All can be measured, and only the numerical value of r needs to be specified, so that k can be realized by using the formula₀、d₀Adaptive adjustment of (3).

This completes the design of the MRAC controller and an iterative learning controller is described below.

The "iterative learning controller" in fig. 1 is designed to:

in the formula, r₀The initial value of the self-adaptive gain is taken as a smaller value which can enable the MRAC control system to stably operate; r is_k(i) And y_rk(i) Respectively setting values of the adaptive gain r value and the rotating speed at the moment i in the kth iterative control process; coefficient lambda_PA proportional learning gain; e.g. of the type_I(k-1)And (i +1) is an input value (namely a rotating speed error value) of the iterative learning controller at the i +1 moment in the k-1 th iterative control process, and is calculated by using an equation (1).

The design of the iterative learning adaptive MRAC controller is thus completed. After the iterative learning self-adaptive MRAC controller is designed, the designed iterative learning self-adaptive MRAC controller can be adopted to carry out closed-loop control on the rotating speed of the ultrasonic motor, and the motor control method based on the iterative learning self-adaptive MRAC controller is realized.

The method is applied in the following specific examples to illustrate the effectiveness of the method of the invention.

Through programming of a DSP chip, a control system corresponding to the control block diagram of FIG. 1 is realized to verify the effectiveness of the iterative learning self-adaptive MRAC controller. The motor for the experiment is a Shinsei USR60 type traveling wave ultrasonic motor, the driving circuit is of a two-phase H-bridge structure, and a phase shift PWM control mode is adopted.

Setting step given value y of rotating speed_rkThe rotation speed response curve is shown in fig. 2(a), and the change curve of the r value is shown in fig. 2 (b).

In the 6-time step response process shown in fig. 2(a), the 1 st step response is tested by using the model-referenced adaptive speed control system in the dotted-line frame of fig. 1, so as to provide initial memory for the subsequent iterative learning process. It can be seen that the 1 st rotational speed step response has no overshoot, but the response speed is slow, and the adjustment time is 0.4716s, which is much longer than the expected step response adjustment time of 0.12 s. In the subsequent 5 times of iterative learning processes, the step response still has no overshoot, the response speed is gradually increased, the learning effect is obvious, and the adjusting time is 0.1965s, 0.1441s, 0.1310s, 0.1179s and 0.1179s in sequence. The 5 th and 6 th step response curves substantially coincide and have substantially coincided with the desired control performance and have stabilized in this state.

The speed of response of the rotation speed is increased, and is necessarily related to the change condition of the r value. The r value variation curve shown in fig. 2(b) shows that, under the action of the iterative learning control law (8), the r value gradually increases from a fixed value of 0.001 at the 1 st response as the iterative learning process continues, directly resulting in the feedforward gain k shown in fig. 2(c) and 2(d)₀Feedback gain d₀The rate of change of (2) is increased, so that the control action in the dynamic process of step response is realizedWith the enhancement, the rotational speed adjustment time is shortened. Meanwhile, fig. 2(b) also shows that the value of r is continuously changing during one step response. The greater the rotation speed error, the greater the r value, the faster the intensity of the control action increases, causing the rotation speed error to decrease rapidly. Along with the reduction of the rotating speed error, the r value is also gradually reduced, so that the acceleration of the controlled variable is slowed down, and the overshoot can be effectively avoided. As the speed response reaches steady state, the value of r decreases to its initial value of 0.001. The small r value in the steady state can reduce the steady-state fluctuation of the rotating speed caused by random disturbance such as noise.

Therefore, the r value self-adaptive adjusting method is effective, can automatically adjust the r value through iterative learning to enable the system output to approach the reference model output, and can make the dynamic performance meet the expectation through continuous adjustment of the r value in a response process.

In conclusion, the method for adaptively adjusting the value of the positive diagonal matrix gamma by adopting the iterative learning idea can solve the problem of how to determine the value of gamma, effectively improve the control performance of the system and enhance the adaptability of the controller to the model error of the object by learning.

In this embodiment, the controlled object controls the rotation speed of the ultrasonic motor by using an iterative learning adaptive MRAC controller designed for the ultrasonic motor and a driving circuit thereof. As other embodiments, the iterative learning adaptive MRAC controller designed by the invention is also suitable for other types of motors and correspondingly controls the rotating speeds of the other types of motors.