Nonlinear discrete dynamic system based on support vector machines
基于支持向量机的非线性离散动力学系统控制

A new approach to control a nonlinear discrete dynamic system based on support vector machines (SVM) is proposed in this paper, which depends on the identification of a state space model by SVM. Firstly, a local optimal linearization model is identified at every operating point by least squares support vector machines (LS-SVM), which belongs to the least squares version of SVM. For a linearization model, any linear controller design technique can be applied to design local linear controller at the operating point, and design procedure is repeated at every operating point in the control task. The proposed approach is applied to two typical examples. Pole placement technique is chosen as the linear design technique. Finally, simulation results show that the system has satisfactory tracking performance with reference input because of the desirable ability of SVM.