Study Of Fuzzy Predictive Control Method To Monlinear System With Input Constraints

Nonlinearity is the intrinsic characteristics of most systems in the nature,and in the running, the system usually constrained by physical conditions orexternal factors, so the research of constrained nonlinear systems is veryessential. Predictive control is a advanced control algorithm which is developedin the actual industry, and it is gradually improved since it inception, the rollingoptimization and online correction strategy make it has strong adaptability forcomplex nonlinear systems. As we all know, to almost nonlinearity system, it isdifficult to establish accurate mathematical model, T-S fuzzy model has goodapproximation properties, it established a system model rely little on the internalmechanism and determined a set of fuzzy rules through the relationship betweeninput and output.Fuzzy predictive control combined the best of both fuzzy control andpredictive control, using T-S fuzzy model as an approximation for nonlinearsystem, and then adopting the method of predictive control. It does not needexact expression of the model, and can calculate the control law onlineaccording to running status of the system. However, the actual model ofcontrolled object is not ideal, it must consider the complex problems which mayappear in the actual system such as constraints, time-delay and coupling whenusing the fuzzy predictive control in industrial process. This paper mainlystudied fuzzy predictive control for nonlinear systems with input constraints.To the general nonlinear systems with input constraints, first of all, the algorithm identified its T-S fuzzy model, then introduced the diffusion factor inthe predictive control, and according control targets fuzzy satisfaction to adjustthe diffusion factor online, adjustment of diffusion factor directly lead to theperformance of the system change, the system performance changed will affectthe control targets fuzzy satisfaction, both of them influenced each other untilthe system achieved the desired performance. Input diffusion factor unifiedcontrol quantity and control increment constraints, set point diffusion factor tomake the output more stable. So diffusion factor reasonable adjustment canensure satisfied the constraints and obtain satisfactory performance. Look fromthe MATLAB simulation results, online adjust the diffusion factor of the systemperformance is better. In view of the result of MATLAB simulation, the systemperformance with the diffusion factor adjusted online is better.To the nonlinear systems which can be described by the Hammersteinmodel, the paper proposed a new two-step fuzzy generalized predictive control,and made the input constraints to the intermediate variable constraints. Firstly,using the predictive control algorithm for linear element solved intermediatevariable, then to the non-linear element, according to the intermediate variableand prediction error designed the corresponding fuzzy rules to inverse mappingthe input variable. Intermediate variable is not only the input variable of thelinear element, but also the output variable of the nonlinear part, so changing theconstraints is feasible, the change of the input variable within the scope of theconstrained keep continuous and smooth by the rolling optimization strategy ofgeneralized predictive. From the pH control process of the MATLAB simulationexperiment, it can draw the conclusion that the approach is effective.