Decoupling Control Of A Multivariable System Based On Fuzzy Neural Network

With the development of the theory of neural network, neural network controltechnology and other advanced technology, gradually infiltrated into every branch ofcontrol theory. Because neural network has good learning ability, fault tolerant ability,generalization ability and nonlinear mapping ability, so it can solve the problem thatthe traditional decoupling method can’t solve, and provide a new thought. In thispaper, for the problems of the multivariable variables, the strong coupling and thepure delay system, a method is proposed that can decouple a multivariable system intosome single variable systems, based on the union of the genetic algorithm, fuzzyneural network.This paper first introduces the present situation and development trend of thedecoupling control system. Then, combining fuzzy neural network and the basictheory of genetic algorithm, the Mamdani based fuzzy neural network mode is studied.The multiple input and multiple output decoupling system model and algorithm arestudied. At the same time, the general system model is put forward to decouple.Thedecoupling algorithm applies to three-dimensional web-fed printing machine speedand tension control system and four-dimensional aviation engine pneumaticthermodynamic system. They are successfully decoupled into three and four singlevariable systems.The traditional method has great difficulty in decoupling the nonlinear system,the variable structure system, the complex system with the coupling relationship orthe changing coupling strength by the time and the load. However, this method that combines the advantages of fuzzy logic and neural network can make up the defectsof the traditional decoupling method and have a good decoupling control performancein complex system, due to the non-linear and the self-learning ability. And thismethod is easy to implement without the precise mathematical model.This method does not need to establish the precise mathematical model and it iseasy to achieve. Finally, the simulation results show that this method has a gooddecoupling control performance.