A Control Scheme for MIMO System Based on SVM

A scheme for multi-input/multi-output (MIMO) black-box system based on support vector machine (SVM) is developed. By analyzing characters of MIMO system, system inputs are decoupled each other as combination of objective function and system output. After designing transform law of objective function, system inputs are acquired directly as outputs of trained SVMs, therefore recognition and controller design are combined and finished by training SVM. The scheme is proven in linear and nonlinear MIMO systems. Simulation results show that the scheme has higher identification precision and less adjusting time than neural networks (NNs).