Internal Model Control Based on LS-SVM for a Class of Nonlinear Process

Abstract

Abstract Aiming at a class of nonlinear process, an internal model control (IMC) scheme based on least squares support vector machines (LS-SVM) is proposed in this paper. By using LS-SVM algorithm and selecting the Gaussian kernel function, the internal model and its inversion of the nonlinear process are constructed, and the shortcomings of process identification and modeling based on neural network could be overcome. Then, combing with LS-SVM, the structure of internal model control was designed. The simulation results show that the nonlinear process identification based on LS-SVM has higher precision and better generalization than RBF neural network (NN) method, and IMC based on LS-SVM could achieve a good dynamic performance and robustness.