Novel fuzzy predictive PID control for a class of nonlinear systems

Most plants are nonlinear in practice, the proper T-S fuzzy model is obtained by a recursive fuzzy identification method to describe the real complex nonlinear systems. First, a fuzzy-clustering algorithm is used to competitively learn the centers of the input-areas online. Second, the radii of input areas are determined based on the superposed degree of the predetermined rules, while the parameters in the inference can be achieved by the RLS algorithm, then the fuzzy local linear model can be used as the CARIMA model at every instant to predict the output of the plant. Finally, a tuning algorithm for PID controller parameters is presented based on the idea of generalized predictive control (GPC) and the relation between the GPC and PID controller. The optimal index includes the weighting term of the square of error between the predicted process steady-state value and the desired-state value. The stability of the system has also been studied. The simulation results both in theory and a paper pulp process show the effectiveness of the proposed method.