Dynamic support vector machine regression based on recurrent strategy

Currently, the research and application of support vector machines regression (SVR) mainly focus on the statistical modeling of the complex industrial process, which is difficult to meet the requirement of real dynamic process control. In order to overcome this shortcoming, a really dynamic recurrent support vector machine regression model (DR-SVR) is proposed, within which inner structure cells are introduced based on recurrent strategy. Then a recursive algorithm for input data with sequence supply rather than batch is given. Finally, the new DR-SVR method is used to identify the dynamic model of the PET process. The simulation results show that the presented DR-SVR model has better performance than the normal SVR in the identification of industrial process.