Research And Application Of Fuzzy Predictive Control In The Three Level System Based On T-S Model

In the industrial control systems, non-linear and time-delay problems are common and can not be avoided, which also have been the hot spots in the study of control. Time-delay problems are the direct causes leading to poor quality even instability of control. In the existing theory there are a lot of control compensation methods which can sovle these problems, on the premise of knowing the exact model of controlled object. However, there is a certain amount of uncertainty in the actual industrial control systems, which can not be mastered and given the precise model. So it is necessary to find a good control tactic of strong anti-interference ability and good robustness to solve these problems with no depending on system model. Predictive control is precisely to meet these conditions: anti-interference ability, strong robustness, fast response, little overshoot and no depending on the precise model. And Dynamic Matrix Control(DMC)is a typical control algorithm of predictive control, which has simple algorithm, less control parameters and can be easily adjusted. The uncertainty of the system can be overcome by using rolling optimization and feedback correction strategy, which not only can solve the time-delay problems in the system, but aslo can be under no static error control in case of model mismatch. In this thesis, Dynamic Matrix Control(DMC)is studied and applied in time-delay system. At the same time, it shows fast trackability, good robustness and strong anti-interference ability compared with classic PID and Smith control through simulation. The precise mathematical model is difficult to establish because of the nonlinear feature. So the predictive control algorithm based on linear system model is aslo difficult to be used in real system.In order to realize high-quality control of nonlinear time-delay systems, a combination of Fuzzy Control based on T-S model and Dynamic Matrix Control (DMC) that is typical in predictive control is introduced in this thesis, in which a thought that realize the universal nonlinearity based on local linearization is under application, taking the measure of linear control to design nonlinear controller in T-S model. Choosing the model vector of step response according to fuzzy rules at current time and combining the algorithm of DMC to get the optimal control parameter of controlled object, the control of nonlinear time-delay systems using predictive control is successfully implemented. And then applies this algorithm to three level system, through simulation results it shows that the presented algorithm is valid. It has excellent sensitivity to the variety of operating conditions and the variety of control variables is smooth. Thus it precipitates application prospect in engineering.