Adaptive Inverse Control For Siso Nonlinear Systems With Pid Neural Networks

Most of the plants in the industry field are nonlinear systems, and the internal dynamics changes dramatically during the process of the change of working environment. For that reason, it's really hard to describe the plant precisely with analytic expression. As a result of that, classic control theory and modern control theory will not perform perfectly while highly stability and flexibility are required. Adaptive Inverse Control is first denominated in 1986, and it has become more and more popular in the industry for its good adjustability and Robust. There are two major problems during the research field of nonlinear adaptive inverse control: one is how to make adaptive filter converge more quickly and have more simple structure, the other is how to cancel the disturbance. In this paper, some researches have been done on the base of reading many references about adaptive inverse control and neural networks.Firstly, a special dynamics neural network, PID neural network, is employed as adaptive filter for the algorithmic convergence and complexity of the structure of adaptive inverse control using neural networks. Some research has been done to analyze the advantage of such method.Secondly, the emulator, which is used to identify the model and inverse of the plant, and the inverse controller, which is used to control the dynamics of the system, are designed based on adaptive inverse control according to the special structure of PID NN. And find out a perfect method for the plant.Finally, a kind of neural network disturbance canceller is employed for the disturbance of the plant to modify the biased system. This so called BPTM-PIDNN with disturbance canceller will control the dynamics of the plant as well as canceling the disturbance. The simulation results show that this measure can control the nonlinear systems effectively.