Direct adaptive control based on LS-SVM inverse model for nonlinear systems

This article proposes an inverse model control scheme for unknown nonlinear systems. The proposed scheme structures the inverse dynamics of the controlled system offline. This is done using a least squares support vector machine (LS-SVM) network that based on the Fuzzy C-Means (FCM) clustering technique. Then the proposed network is updated online to determine the appropriate control action. A linear feedback compensator is added to ensure the closed loop stability. Simulation results of a continuous stirred tank reactor (CSTR) demonstrate the efficiency of the proposed approach.