Intelligent Control Applied Research In Nonlinear Control Systems

The methods of controlling nonlinear system is reviewed firstly in this thesis and the history of fuzzy control technology and neural network is also reviewed. In order to overcoming the difficulties of creating model of nonlinear system, a new method of controlling and identifying is proposed in the thesis. With the development of the industry and more and more complicated processes appeared in the industry, the common PID control obviously shows its shortcomings. In recent years, some scholars create complicated fuzzy simultaneous equations to gain model of nonlinear system, theory of the equation is clear, but solution of it is not single or solution is not exist. Some researchers try to create Linguistic fuzzy model, but how to gain the optimal fuzzy rule bases of the given system is very difficult. Taking advantage of the merits of intelligent control and conventional control, two algorithms of controlling is proposed. Firstly, imitate evolution process of biology, overcoming difficulties of GA whose parameters are obtained difficult and fuzzy controller whose fuzzy rule bases are also gained with not easy, using evolutionary programming to optimizing fuzzy rule bases of the given system, moreover, the method is also used in identifying nonlinear system. It provides a new approach of controlling nonlinear system . Secondly, In order to solve conflict of neural network generally used in nonlinear control and construction of it is not gained with easy, a new two steps evolutionary programming is proposed based on the improved method of evolutionary programming , the first step gene represent the construction of neural network and the second step gene represent the parameters of it, the second step is controlled by the first step. the best construction and parameters of network can be obtained at the same time. Thirdly, combining the merit of PID controller whose construction is easy and RBF neural network converge faster, an intelligent PID controller based on the improved evolutionary programming is proposed in the thesis, the given method is better used to improve dynamic and static characteristics of the given system. After lots of computer simulation, the practical experiment of ACotor is completed. So it is concluded that this method not only have the theoretical value but also stronger practical value in a sense. Hope it can be carried out in the control of industrial process in the near future.