Font Size: a A A

Research On Nonlinear Adaptive Inverse Control

Posted on:2016-01-06Degree:MasterType:Thesis
Country:ChinaCandidate:H DongFull Text:PDF
GTID:2428330542957542Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
With the rapid development of modern science and technology and increasing scale of production,a large number of complex nonlinear system appeared in the process of industrial automation production in order to solve the problem of these complex industrial control systems,adaptive inverse control(AIC)as a novel control method has been proposed.So far,linear AIC method is relatively mature,but the research achievements of nonlinear AIC are few.Nonlinear adaptive filter and nonlinear AIC structure is a key determinant of nonlinear AIC performance,and the rapid development of neural networks provides a strong nonlinear adaptive filter to nonlinear AIC.In this thesis,the corresponding nonlinear adaptive filters are designed by using neural network and are applied to the nonlinear AIC system to achieve control of nonlinear systems.Firstly,through extensive review of related literature,the current domestic and foreign research status of the AIC was comprehensively reviewed.The basic concepts and principles of AIC and the neural network filter theory are expounded,to pave the way for the next modeling,inverse modeling and the entire adaptive inverse control system based on neural network filter.Secondly,in the modeling for nonlinear object,in view of the fixed rate of the BP neural network filter modeling method based on local error having low model generalization ability and model accuracy depending on the size of the learning rate,the variable rate of the BP neural network filter modeling method based on global error is used.The simulation show that this method in nonlinear controlled object modeling not only overcomes the problem of modeling accuracy of the fixed rate of the BP neural network filter modeling method based on local error depending on the size of the learning rate,but also the generalization capability of the model has also been greatly improved.Thirdly,in inverse modeling of nonlinear objects,if using BP neural network designs inverse controller,because BP algorithm is relatively complex,the convergence speed is slow and model accuracy is not high.It can't meet the requirements.So the inverse modeling method based on orthogonal neural network is used to design inverse controller.According to this thought,learning algorithms of orthogonal neural network controller are derived in detail.Simulation studies of a time-invariant nonlinear discrete system indicate that the method of inverse modeling can achieve better results in nonlinear systems.Finally,after nonlinear object is modeled and inverse modeled,using the model reference AIC system based on neural network to control the nonlinear controlled object of AIC system.Simulation results show nonlinear AIC system can work well,and the adaptive ability is very strong,explaining that the scheme method is feasible and effective in the nonlinear systems AIC.
Keywords/Search Tags:Nonlinear System, Adaptive Inverse Control, Neural Network, Adaptive Filter, Variable Rate, Inverse Modeling
PDF Full Text Request
Related items