A Simplified Model-Free Self-Evolving TS Fuzzy Controller for Nonlinear Systems with Uncertainties

This paper proposes a self-evolving Takagi-Sugeno fuzzy controller for nonlinear systems with uncertainties. The self-evolving framework is used to add and prune fuzzy rules in an online manner. Our proposed nonlinear controller is model-free and does not depend on the plant dynamics. All adjustable fuzzy parameters are tuned using a sliding surface, which is derived from the gradient descent learning method to minimize the error between the desired and the actual signals. Unlike most of the existing self-evolving controllers, where a hybrid technique is required to determine the control action, our proposed algorithm is able to construct the final control signal, which can be fed directly to control a nonlinear system. The tracking performance of our proposed controller is validated and compared with an adaptive model-based fuzzy controller in the presence of external disturbances, where better tracking results are obtained from our proposed controller.