Research of predictive control based on OS-LSSVMR

The accuracy of modeling is important for control accuracy and stability in predictive control. So, SVMR (Support Vector Machines Regression), which supports by mathematics theory and has a simple structure and nonlinear modeling properties, compare with other modeling algorithm, such as neural networks, has been applied in the predictive control. Here, OS-LSSVMR, which has good online modeling properties, has been used to model the nonlinear system process based on empirical data in Model Predictive Control, which recede optimization using LM (Levenberg-Marquardt) non-linear least square optimization. The simulation shows that the control algorithm is a good stability and real-time.