Research On Generalized Predictive Control Based On T-S Fuzzy Model

Model-based predictive control (MPC) is one kind of advanced control technique, which was developed from the industrial process. With the combine between the gained productions of MPC and intelligent control, intelligent predictive control has started to present as one of new research domain of model predictive control. Fuzzy identification and control theory have great potential in solving control problems of complex system that is one novel intelligent control methodology. As the widely used T-S fuzzy model, it is essentially one nonlinear model .It has been proved in theory that T-S model can approach arbitrary nonlinear systems according to arbitrary precision. Using the T-S fuzzy model as predictive model can fitly remedy the shortage of single predictive model. This paper applies the T-S fuzzy model into the generalized predictive control, and then presents one new method of fuzzy generalized predictive control. Further research on the fuzzy predictive control will have great signification for theory and practical applications.Firstly, based on the basic principles and characters of fuzzy control and predictive control, the paper analyzes the possibility of integration between them. Give the inducing process of GPC system and the analysis to the choice of the GPC parameters.Secondly, Aim at the problems of sampling the procedure of T-S fuzzy model and adjusting on-line fuzzy space partition and fuzzy regulation of time-varying nonlinear systems, a new modeling method which comprises off-line and on-line identification based on T-S model is presented in this paper. Based on recursive fuzzy clustering method, the structure and parameters in T-S fuzzy system are adjusted on-line conveniently. Present two means of fuzzy generalized predictive control based on partial linear model and global time-varying model. The simulation, analysis and compare between the two fuzzy GPC provide the gist to the practical design of fuzzy predictive control.Lastly, Genetic algorithm is used on fuzzy identification on nonlinear systems, based on fuzzy partition of input space, it adopts adaptive extended gauss function as member function, uses genetic algorithm to optimize its figure to identify the parameters of the fuzzy model. Finally the practicability of this method is demonstrated by the simulation results example.