Nonlinear Model Predictive Control with the Integration of Support Vector Machine and Extremal optimization

In nonlinear model predictive control (NMPC), the system performance is greatly dependent upon the accuracy of the predictive model and the efficiency of the online optimization algorithm. In this paper, a novel NMPC scheme with the integration of Support Vector Machine (SVM) and recently proposed general-purpose heuristic "Extremal Optimization (EO)" is presented. With the superior features of self-organized criticality (SOC), non-equilibrium dynamics, co-evolutions in statistical mechanics and ecosystems respectively, a carefully designed EO based on " horizon based mutation strategy" is used as an online solver to obtain optimal future control inputs of NMPC, in which a multi-step-ahead SVM predictive model is employed. Furthermore, simulation studies on a typical nonlinear system are given to illustrate the effectiveness of the proposed control scheme.