Generic Model Predictive Control Strategy Based on Integrated Weighted Least Square Support Vector Machines

In this paper, a new generic model predictive control (PGMC) is proposed. This method combines predictive control and generic model control (GMC), which allowed for favorable robust performance. To get accurate predicted errors, the integrated weighted least square Support Vector Machines (IWLS-SVM) is proposed. Proposed method considers time element of sample data as well as outliers and noises. For the real features of the samples in the proceeding of production, the IWLS-SVM increases response speed and real time ability of the control system. The PGMC based on IWLS-SVM are applied to bending roll control system. The result of a numerical simulation experiment shows the feasibility and effectiveness of this algorithm.