Smith predictive compensation method based on sixth-order B-spline wavelet neural network

Abstract

The invention provides a Smith predictive compensation method based on a sixth-order B-spline wavelet neural network, which solves such problems of a Smith predictor as low model precision and unsatisfactory interference suppression effects. The method is characterized in that a differential equation is established for a tested object and subjected to discretization processing, so that system state quantity sampling intervals and learning samples of the sixth-order B-spline wavelet neural network are obtained; and after a structure, input layer weights, a hidden layer function and the number of hidden layer nodes of the neural network are determined, iterative training is carried out to obtain an output layer weight vector and a neural network expression, so that a mathematic model of a Smith predictive compensator is obtained. The method provided by the invention has the advantages that a non-linear controlled object can be modelled; the precision of a process model can be effectivelyimproved; and meanwhile, due to the characteristic of the wavelet neural network that the frequency band is limited, satisfactory interference suppression effects can be achieved.

Description

Technical field: [0001] The present invention is related to the Smith predictive compensation method. Background technique: [0002] Smith predictive control is a control strategy designed for pure lagging systems. In control theory, hysteresis means that the change of the controlled variable lags behind the change of the disturbance in time, which is a very common phenomenon. Pure hysteresis refers to the delay in the transmission of materials, energy or signals due to the limited transmission speed. Generally pure lag refers to the lag caused by the transmission speed limit. Smith predictive control is a kind of pure lag compensation control, which weakens and eliminates pure lag by introducing a compensator connected in parallel with the controlled object. Compensated by the Smith Predictor, the pure hysteresis is moved out of the closed-loop control loop so that it does not adversely affect the system. It can be seen from the displacement theorem of Laplace transform...

AI Technical Summary

Problems solved by technology

However, this method is very sensitive to the error of the process model, and the compensation effect depends on the accuracy of the compensator model. If the error is too large, the control quality will deteriorate, and the system may even become unstable.

Images