Design And Simulation Of Decoupling Control On Multivariable Coupling System

In recent years, with the rapid development of modern industry, production process becomes more and more complex. Their control systems are often multiple input and output. Coupling among variables becomes a widespread phenomenon in the control system. The existence of coupling, not only make the system very difficult to control, but also greatly reduce the system's controlling quality. In serious cases, the system will not run. Decoupling of coupled object is an important tool to improve the level of automation and meet the growing requirements of the production process. Therefore, decoupling controlling of multi-variable coupled system has great practical significance and application prospects.In addition, as China's rapid economic development, the use of the film is more and more extensive. Their quality requirements are also increasing. Film thickness control system is extremely important part of the film production line. The quality of controlling determines the quality of films. As the film thickness is a typical multi-variable coupled system, the study of its decoupling controlling has more practical significance and application value.At first, the dissertation introduces the source of the project, the research background, and its significance. It summarizes the current situation of decoupling controlling, analysis method on degree of the variable coupling, three relevant decoupling method, PID controller and neural the basic theory of neural networks; Then, on the basis of the adequate research, the dissertation designs and simulates similar-feed-forward decoupling PID controlling of the film thickness system ,and RBF neural network decoupling PID control of a two-variable coupled system; Finally, it gives some discussions and conclusions as well as future study directions. Two major tasks completed in the dissertation are as follows:(1) Design and Simulation of similar-feed-forward decoupling PID Controlling on film thickness system Relative gain array method is used to analysis the coupling of the film thickness control system. A method combining similar-feed-forward decoupling with PID controlling strategy is used to achieve decoupling control dynamically and completely. Compared with inverse nyquist array decoupling PID controlling, this method has the virtue of a perfect control result and decoupling completely without changing the main control channel's characteristics. It has very important significance for industrial process control.(2)Design and simulation of RBF neural network decoupling PID control on a two-variable coupling systemAlgebraic algorithm, which can transform the nonlinear optimization problem into solving a set of linear algebraic equations, is a novel learning algorithm of neural network. Using algebraic algorithm to train RBF neural network can achieve accurate mapping samples of the network, also can guarantees a faster convergence speed. In this thesis, the dynamic RBF neural network is adopted to identify a two-variable coupled system online, and the sensitivity of information acquired is to self-tune the parameters of PID controller. Thus, the decoupling control of system is achieved. The result of simulation shows that this design possesses the advantages of high precision,quick response speed and is of great robustness and adaptability. It has great practical value.In summary, this study of the similar-feed-forward decoupling PID controlling and RBF neural network decoupling PID controlling can effectively achieve the decoupling control of multivariable coupling system. They greatly improve the control quality, and have great practical significance and application.