Neuro-fuzzy control based on on-line least square support vector machines

A novel neuro-fuzzy control structure has been proposed to solve the nonlinear control of industrial system which is referred as the on-line Least Square Support Vector Machines (LSSVM) based on Adaptive Network Fuzzy Inference System (ANFIS) controller since it has been emerged from the ANFIS and SVM. In the proposed controller, an initial control vector is generated by fuzzy neural networks, which will be optimized by on-line LSSVM based controller. And then the optimized control vector will be applied to the controlled system which parameters of neuro-fuzzy network will also be tuned according to the optimized control vector. The simulation results have revealed that the on-line LSSVM based ANFIS controller exhibits considerably high performance by yielding very small transient and steady tracking errors.