Nonlinear Generalized Predictive Control Based on Online SVR

Generalized predictive controllers (GPCs) have been successfully applied in process control during the last decade. The performance of unstable, non-minimum-phase, or linear processes with dead-time are improved, however, GPCs are limited when used to control real industrial process, because of the real processes is often onlinear. According to the difficult problem of building an accurate model for a nonlinear system, a new generalized predictive control (GPC) algorithm based on online support vector machines (OSVM) is designed. At first, some key parameters of model are identified on-line by OSVM, and then using generalized predictive control strategy to realize predictive control for the system. Through simulating of some experiments, the result shows that the design could predictive accurately, has a good stability and small the steady-state error.