Design of Adaptive Fractional-Order PID Controller to Enhance Robustness by Means of Adaptive Network Fuzzy Inference System

Abstract

In this paper, a tuning strategy for the design of fractional-order proportional–integral–derivative (PI λ D µ ) controllers is proposed. First, a PI λ D µ controller is designed with genetic algorithm in order to obtain the training data. Then, three Adaptive Network Fuzzy Inference System (ANFIS) structures, related to K p , K i and K d parameters of the PI λ D µ controller, are formed by using the training data. These ANFIS structures are used in the PI λ D µ controller instead of K p , K i and K d parameters, and they are capable of self-tuning during the simulation based on the input signal of the adaptive PI λ D µ controller (ANFIS–PI λ D µ ). Finally, in order to show the control performance and robustness of the proposed parameters adjustment method with ANFIS, simulation results are obtained by using the MATLAB–Simulink program for two different systems and the results obtained from ANFIS–PI λ D µ controller are compared with the results of PI λ D µ and fuzzy logic controller.