The Research And Application Of Adaptive Inverse Control Using BP Neural Networks

As a novel control method, adaptive inverse control had got large development in the past decades.Up to now, linear adaptive inverse control method is relatively mature,however, the research and results of nonlinear adaptive inverse control systems are rare. The control performances of nonlinear adaptive inverse control system are mainly depending on the structure design of nonlinear adaptive filters and nonlinear adaptive inverse control. At present, the fast development of neural networks provides adaptive inverse control systems with powerful nonlinear adaptive filters. BP neural networks have features such as simple structure, universal approximation and good generalization ability. So this thesis combines neural networks with nonlinear adaptive inverse control to ensure the dynamic features control and the disturbance elimination control of the nonlinear object.Firstly, a structure of nonlinear control method which based on neural networks is presented, according to the characteristics of the nonlinear object and BP neural networks, two different structures of BP neural networks are designed and driven respectively by BP and BPTM algorithm to approaching the object model and inverse controller. Simulation results show that the object model and inverse controller which obtained by neural networks with high accuracy approximation, thus, Adaptive inverse control system based on BP neural networks can get a good tracking performance and good dynamic performance.Adaptive inverse control with feature of processing object dynamic control and disturbance elimination control separately.On the basis of analyzing and studying the principle of disturbance rejection, the inverse object model is obtained.The disturbance rejection loop is introduced into the existing nonlinear control system and a nonlinear adaptive inverse control structure which can quickly eliminate disturbance is designed. The simulation results prove that the system can quickly eliminate the object disturbance and get a better tracking performance at the same time.Finally, the nonlinear adaptive method which designed before is applied to the nonlinear inversed pendulum.Object model and inverse model of inversed pendulum using BP neural network study algorithm were given. The result indicates that adaptive inverse control has the ability of making the inverted pendulum stabilize in an allowed extent and better control effect.The results show that nonlinear adaptive inverse control suitable for resolving nonlinear problems, so it can be used to solve complex control problems of nonlinear system.