A feedforward tracking control method based on neural network inverse model

By using neural network inverse model to construct a dynamic feedforward channel in the feedforward-feedback control system, and realizing information interaction between feedforward and feedback channels, the problems of rough feedforward effect and unstable adjustment in the prior art are solved, and the control quality of the system and the rapidity and accuracy of the adjustment process are improved.

CN114779648BActive Publication Date: 2025-03-28NORTH CHINA ELECTRIC POWER UNIV
3 Cites 0 Cited by

Patent Information

Application Number
CN202210542099.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-05-18
Publication Date
2025-03-28
Estimated Expiration
2042-05-18

AI Technical Summary

Technical Problem

The feedforward-feedback control system in the prior art has problems such as rough feedforward effect, inability to adapt to the dynamic characteristics of nonlinear controlled objects, and lack of information interaction between feedforward and feedback channels, resulting in instability in regulation.

Method used

The feedforward tracking control method based on the neural network inverse model is adopted to build a dynamic feedforward channel through the feedforward controller and the prediction model, and the tracking controller is used to realize the information interaction and adjustment linkage between the feedforward and feedback channels.

Benefits of technology

The control quality of the system is improved, the feedforward action is adapted to the full range of working conditions of the controlled object, the feedback reciprocating adjustment and the reciprocating action of the actuator is avoided, and the speed and accuracy of the adjustment process is improved.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention relates to a feedforward tracking control method based on a neural network inverse model, wherein the control process of the method is as follows: the set value of a controlled object is input into a feedforward controller to obtain the output of the feedforward controller; the output of the feedforward controller is input into a prediction model to obtain the output of the prediction model; the difference between the prediction model output and the output of the controlled object is input into a tracking controller to obtain the output of the tracking controller; the sum of the output of the feedforward controller and the output of the tracking controller is the control quantity of the controlled object, and the control quantity is input into the controlled object to obtain the output of the controlled object; the feedforward controller and the prediction model are implemented by a neural network model, and the feedforward controller is the inverse of the prediction model. The feedforward tracking control method includes two control channels, feedforward and tracking, wherein the feedforward channel is used as the main body for rapid response, and the tracking channel is used as compensation to achieve fine adjustment and zero-difference adjustment, taking into account both the rapidity and accuracy of the control process.
Need to check novelty before this filing date? Find Prior Art