The Study On Intelligent Controller Design Method Based On Genetic Algorithm

Modern industry possesses many characteristics, such as non-linearity, large time lag, time-varying and non-accurate mathematic model and so on. It is difficult to obtain perfect control performance; hence we should find new intelligent control strategy. At present, the combination of fuzzy theory, neural network and genetic algorithm (GA) is gradually showing enormous latent capacity in the control field. Based on the combination and improvement of the previous three kinds of control technology, the controller can be automatic designed and optimized with the help of improved genetic algorithm without knowing the mathematic model and prior experience. This paper mainly designs two kinds of controller. The first kind is the fuzzy controller with multiple weighted factors optimized by genetic algorithm. This kind of controller has the advantages of fuzzy control and genetic algorithm. It can realize the fuzzy control rules online self-optimization and improve the self-study performance. The second kind is the neural network controller based on genetic algorithm. It combines the fuzzy control technology, neural network control technology with genetic algorithm. First it uses neural network to construct fuzzy logic system according to the structure equivalence rule and then it finds the optimal weight value of the network by genetic algorithm. With the optimization of the weight values both the membership and fuzzy control rules are optimized.This paper has worked out the control algorithm programs afterwards, and tested the performance of them by doing simulation experiments.The main contents are as follow.(1) Analyze current situation of fuzzy control, neural networks and genetic algorithm (GA).(2) Research the principle of GA and make some strategies to improve GA performance for controller parameters optimization.(3) Research a kind of fuzzy controller whose fuzzy control rules are describe by an analysis expression and present two kinds of new analysis expressions. By adjusting the weighted factors, the fuzzy control rules can be amended accordingly.(4) Combine fuzzy control technology with neural network technology and design a kind of neural network, which is equivalent to fuzzy logic system according tostructure. Thus the amending of membership function and fuzzy control rules can be realized by adjusting the \veight factors.(5) Commingle fuzzy control, neural network with GA and design two kinds of controller. The fuzzy controller with multiple weighted factors and fuzzy neural network controller optimized by GA.(6) Test the performance of controller designed in this paper, by doing simulation experiments with the first order inertia plant having the pure lag and time-variation parameters.(7) Analyze the characteristic of omethoate synthesis system, decide the reasonable control scheme and do simulation research.