Research On Self-tuning Control Method For Multivariable Time-delay Systems

For the multivariable system in the practical control engineering, this paper studies theadaptive control algorithm and decoupling algorithm. Because there are many factors, suchas nonlinear, time-varying, coupling, time-delay, non-minimum phase characteristics etc.,in the process control systems, it is difficult to describe controlled plants by the precisemathematical model and design the controller for them. Since satisfactory controlperformance cannot be obtained by most available control method; the scholars in thecontrol field are looking for better control algorithm all the time. Thus they are payingmore attention in the control algorithm of multivariable time-delay and coupling system.The neural network has good characteristics of parallel mechanism, fault tolerance,distributed storage and adaptability. Therefore, this paper combine neural network withself-tuning control algorithm to explore new control algorithm of multi-variable couplingsystems with time-delay. From simple system to complex one, this paper starts step by stepfrom the study of single variable linear systems to study of multivariable system, and fromthe study of nonlinear system to the study of the time-delay system, finally studies thetime-varying delay systems with time-delay. I have got good results about these systems.The main contents of this paper are as follows:(1) Study for single variable linear continuous system and discrete system. Thischapter presents the control structure of good fault tolerance, which not needs to identifylag time of the plant. For the study two kinds of the controlled auto-regressive movingaverage model (CARMA) and controlled auto-regressive integrated moving average model(CARIMA), we research a self-tuning control algorithm which does not need highaccuracy identification to the plant.(2) For the nonlinear time-delay system with saturation characteristics, this paperpresents a multi-compensator control method, which has the advantages of stronganti-interference ability. (3) For the nonlinear system with time-varying delay, this chapter researches aself-tuning control method which makes use of BP neural network to identify the systemand design the controller.(4) For the nonlinear time-varying systems with dead-zone, the article presents anadaptive compensated method which makes use of two RBF neural networks tocompensate the nonlinear systems with dead-zone.(5) For the multivariable linear coupling systems with time-delay, the article studiesparallel compensating decoupling control which the output terminals of the compensatorsconnect to the output terminals of the system.(6) For multivariable nonlinear coupling system, decoupling control is researched thatcombined PID with an improved BP algorithm, which using the particle swarmoptimization algorithm to optimize the BP network initial.(7) Finally, in the MATLAB/Simulink simulation platform, all above-mentionedcontrol strategies are investigated by the large number of simulation. The simulation resultsdemonstrate the superiority and application value of our method.