Nonlinear Multi-step Predictive Control Based on Least Squares Support Vector Machine

A multi-step predictive control algorithm based on least squares support vector machines (LS-SVM) model for complex systems with strong nonlinearity is presented. The nonlinear offline model of the controlled plant is built by LS-SVM with the radial basis function (RBF) kernel. Based on LS-SVM multi-step predictive outputs, the real process multi-step predictive outputs are expanded into Taylor series expansion. This method can be regarded as the second approximation to the process predictive values. By minimizing the multistage cost function, a sequence of future control signals is obtained. Simulation study has shown that this scheme is simple and has good control accuracy and robustness.