Predictive Control of Nonlinear LS-SVM Inverse System Based on MFAC Compensation

In order to improve the robustness of the traditional inverse predictive control, An inverse predictive control method of nonlinear least squares support vector machine (LS-SVM) based on model-free adaptive control(MFAC) compensation is proposed. Firstly, the LS-SVM inverse system method is combined with predictive control method, and the MFAC controller is designed to modify the inverse model online at the same time, so as to compensate the influence on the system when the model deviates from the controlled object. Simulation results show that the proposed method not only has good control performance for constant and slow time-varying parameter perturbations and external disturbances, but also has good robust control performance for fast time-varying parameter perturbations and external disturbances.