Inverse Fuzzy Learning Control of Unknown Nonlinear Dynamic Systems

This paper addresses fuzzy inverse learning control for unknown nonlinear processes. It introduces an inverse adaptive level-set-based modeling and an inverse fuzzy functional model-based cancellation feedback approach. The level-set-based modeling uses a learning procedure to find the parameters of antecedent membership functions via gradient descent, while employing a recursive least-squares method based on correntropy to estimate the coefficients of the output functions for fuzzy local control laws. The inverse fuzzy functional model-based-cancellation feedback uses a similar learning scheme to develop an inverse process model and a reference model designed to track the desired trajectory. Simulations are done for open and closed loop control of a highly nonlinear benchmark process in tracking complex trajectories. The results show that the inverse learning controllers suggested in the paper outperforms state-of-the-art inverse controllers reported in the literature.