Research And Design Of Intelligent PID Controller Based On Fuzzy Neural Networks

Due to the complication of modern industrial process, and the increase of nonlinearity, time-varying and uncertainty of the practical production processes, the conventional PID controller can no longer meet our requirement. So, in recent years some advanced intelligent control methods have been applied in the PID control field, to find out more effective control strategy.The applications and design procedures of fuzzy neural network controller are researched in this dissertation. The main content is about the intelligence algorithms and its applications in several fields of modern control science, including artificial neural network, gene algorithms and fuzzy logic and its application in model identification, PID parameters design and controller designed.These fellows are the main research points in this dissertation:First, introduced the development and the current research of intelligent control. Second, the basic knowledge of the PID control, fuzzy theory, artificial neural networks and fuzzy neural network is summarized. Present the introduction about the background of the combination of the fuzzy logic and neural network, and the architecture of the regular fuzzy neural network is detailed. Further, the basic knowledge of genetic algorithm, system identification and parameter optimization is studied. Proposed delay time constant identification based on BP neural network, projection algorithm combined with genetic algorithm to indentify the object system parameters, using the genetic algorithms to optimized the parameters of the classic PID controller. Third, the fuzzy neural network based on RBF network is studied and researched, then, the PID type controller based on RBF fuzzy neural network is argued and simulated. Finally, for lack of the former controller, the controller based on the fuzzy CMAC network is studied and designed, and based on this generalized fuzzy neural network, aimed at the weaker time varied non-linear systems the combined control strategy is researched and simulated. Aimed at the strong time varied non-linear systems, a new neural network weight backwards modification method based on modern control stability rule is argued and simulated. Based on the simulations, the control strategy argued in this dissertation has good robustness and be able to against disturb. And the method is concise.