Estimated plant’s sensitivity based on data-driving observer for a class of nonlinear discrete-time control systems

Abstract

This paper presents an adaptive controller based on Fuzzy rule emulated network (FREN) for a class of nonlinear discrete-time systems. The learning algorithm of FREN is conducted via plant’s sensitivity estimated by the data-driven observer unit. The novel learning rate for data-driven scheme is proposed with convergence analysis established by Lyapunov direct method. The control direction can be omitted and boundaries of sensitivity can be assumed to be unknown. Only the relation between plant’s output and control effort is required to design this controller within the format of IF--THEN rules. The closed-loop performance can be guaranteed beside of the convergence of tracking error, adjustable parameters and observer’s output. Results from two practical systems, DC-motor current control and pressing force control, demonstrate that the proposed controller is capable of controlling unknown discrete-time systems with satisfactory performance.