Control Method On VCM Production Process Based On SVM

Support Vector Machines (SVM) is a new machines learning method in the field of statistics. It has been used in the field of pattern identify effectively. It is a powerful tool in nonlinear system identification. Statistical Learning Theory (SLT) is a theory that specialized in machine learning with finite samples. In addition, SLT has high generalization ability as well. SVM is an important area of SLT, and it possesses good characteristic on modeling. Because of its advantage on modeling of nonlinear system, it not only has a simple model, but also provides a new control theory which is suitable for complex nonlinear system. The research of SVM has obtained some achievement, but SVM method has seldom been used to solve the problem in industry procedure control by now.The produce process of vinyl chloride monomer (VCM) has complicated mechanism, it is nonlinear, uncertainty and time-varying. The process is sensitive to the temperature and the change of temperature, which has higher requirement on system modeling and controller designing.In the paper, according to VCM produce technology and control request, reconstruction solution and design solution is proposed. Based on that, a modeling method of nonlinear systems using support vector machines is proposed. And the modeling of SVM is applied to nonlinear predictive control. A kind of nonlinear predictive control scheme based on the SVM model is presented. Furthermore, the predictive control based on SVM was successfully applied to a model and control for vinyl chloride transform of VCM productive process. The simulation results show that SVM can be trained rapidly, and have good generalization under nonlinear modeling and satisfactory control performance.This paper's work shows that SVM can not only be used in solving the modeling problem such as temperature control of the VCM transform process successfully, but also provide a powerful tool in the complicated industry process control. The algorithm of predictive control based on SVM is effective.