Parameter self-tuning method of SISO tight format model-free controller based on Attention mechanism recurrent neural network

Abstract

The invention discloses a parameter self-tuning method of an SISO tight format model-free controller based on an Attention mechanism recurrent neural network. The method includes: firstly, screening important information from an original input set by an Attention mechanism and calculating the input of a generated recurrent neural network, enabling the recurrent neural network to perform forward calculation to output SISO tight format model-free controller to-be-tuned parameters; performing calculation by a control algorithm to obtain control input of a controlled object, adopting a minimum system error function value as a target, using a gradient descent method, combining the gradient information of the control input for each to-be-tuned parameter, performing system error back propagationcalculation by using a chain rule, and updating the ownership coefficient of the recurrent neural network, thus realizing parameter self-setting of the controller based on the recurrent neural network. According to the parameter self-tuning method of the SISO tight format model-free controller based on the Attention mechanism recurrent neural network, important characteristics of input informationcan be captured, the problem of online tuning of controller parameters is solved, and a good control effect on an SISO system is achieved.

Description

technical field [0001] The invention belongs to the field of automatic control, and in particular relates to a parameter self-tuning method of a SISO compact format model-free controller based on an Attention mechanism cyclic neural network. Background technique [0002] SISO (Single Input and Single Output) systems are widely used in reactors, precision Distillation towers, machines, equipment, devices, production lines, workshops, factories and other controlled objects. With the continuous improvement of the level of science and technology, industrial devices are becoming larger and more complex, making the production process more and more strongly nonlinear and time-varying. It is often difficult to achieve the ideal control effect for complex controlled objects with linear and time-varying characteristics. Model-free controller is a new type of data-driven control model, which has a good control effect on unknown nonlinear time-varying systems, so it has a good applica...

AI Technical Summary

Problems solved by technology

[0008] The object of the present invention is to provide a parameter self-tuning method of a SISO compact format model-free controller based on the Attention mechanism cyclic neural network, so as to solve the parameter online self-tuning problem of the SISO compact format model-free controller

Images