Dual-neural network self-learning IPMSM active disturbance rejection control method

Abstract

The invention discloses a dual-neural network self-learning IPMSM active-disturbance-rejection control method, which comprises the following steps: firstly, constructing an objective function based on a position error and a rotating speed error, and carrying out online setting on parameters in a nonlinear expansion state observer and nonlinear state error feedback by respectively utilizing a radial basis function neural network and an error back propagation neural network; the control method gives full play to the superiority of the active disturbance rejection controller, has a better position and rotating speed tracking effect, is high in load resistance and self-adaptive capacity, and better meets the requirements in engineering practice.

Description

technical field [0001] The invention relates to the field of permanent magnet synchronous motor control, in particular to an IPMSM self-learning double neural network self-learning IPMSM self-disturbance rejection control method. Background technique [0002] The traditional permanent magnet synchronous motor vector control strategy generally adopts PID (proportional, integral, differential) control. Different combinations of these three parameters can be adapted to most servo control occasions, so it has a wide range of application effects. In addition, combined control of PID and feedforward, cascaded control composed of multiple PIDs, etc., have been widely used in industrial process control. Although PID is widely used, it also has some inherent defects, such as the contradiction between fast tracking target signal and overshoot, low steady-state accuracy, and poor anti-interference performance, making it difficult to meet the requirements of high-performance control. I...

AI Technical Summary

Problems solved by technology

[0005] Purpose of the invention: In order to overcome the problems of poor anti-interference performance, low steady-state accuracy, and difficulty in achieving ideal control effects by tracking reference signals in the proportional-integral (PI) regulator used in the vector servo control system of the built-in permanent magnet synchronous motor in the background technology , the present invention discloses an IPMSM self-learning dual-neural network self-learning IPMSM ADR control method. The objective function is constructed based on the position error and the rotational speed error, and the radial basis neural network and the error backpropagation neural network are respectively used to observe the nonlinear expansion state. On-line tuning of parameters in controller and nonlinear state error feedback

Images