Predictive controller design for multivariable process system based on support vector machine model

Based on the predictive control principle and support vector machine theory, this paper presents an intelligent predictive control scheme to solve the control difficulties of industry process with multi-variables. As an example, the rotary kiln calcination is the most important part of cement production including complicated physical and chemical reaction processes with large inertia, pure hysteresis, non-linearity and strong coupling characteristics. Considering the need of advanced process control in cement industry, the main control system structure includes three control loops as the pressure control loop, the burning zone control loop and the back-end of kiln temperature control loop. Based on the analysis of PID and generalised predictive control algorithm, the performance index of generalised predictive control algorithm is restructured into PID form. By analysis of the experimental data, the non-linear regression model based on SVM is introduced. The control algorithm using SVM model is simulated in two cases to derive the responses of system compared with the ordinary PID control algorithm. The simulation results of typical step responses of control variables using the presented control scheme show the effectiveness of the control scheme with better response time and tracking performance compared to traditional PID control.