Two-Step Predictive Control Algorithm Based on Least Square Support Vector Machine

According to the nonlinearity of the industrial process, it is difficult for traditional predictive control algorithm to establish an accurate mathematical model. In the paper, a two-step predictive control algorithm based on the least square support vector machine (LS-SVM) is proposed. In this algorithm, the nonlinear system is turned into linear system by adding the appropriate intermediate variables while we consider the coupling of input and output data. Finally, prediction model is constructed by using the previous input and output variables to replace the intermediate variables. The optimal control rule is obtained by using this nonlinear predictive model. Simulation results show the effectiveness of the algorithm.