Adaptive reinforcement learning interval type II fuzzy fractional nonlinear observer and controller for a fuzzy model of a wind turbine

Abstract

This article presents an innovative control technique for doubly fed induction generator (DFIG) based wind turbine. A DFIG equation is initially presented using fractional order Takagi and Sugeno interval type II fuzzy modeling. Moreover, fault and disturbance are considered in the DFIG model. Then, to design an observer that estimates both fault and disturbance, the fuzzy model of DFIG is decoupled into three subsystems using a different linear coordinate transformation: the state subsystem without disturbance and fault, the subsystem of fault without disturbance, and the subsystem of disturbance without fault. In the following, using the state subsystem, an adaptive reinforcement learning TS interval type II fuzzy fractional sliding mode observer is introduced. The observer gains are adjusted by applying proximal policy optimization, which is a new application of reinforcement learning. Robustness against uncertainties, fast convergence speed of estimation, high degree of freedom and precise estimation are among features of the suggested observer. In addition, only by utilizing the estimated state subsystem and obtained mathematical relationships, the actuator faults and disturbance are estimated. A fractional order sliding mode controller is proposed for the DFIG speed controller. Low chattering characteristics and accurate response are achieved owing to the fractional speed controller. Also, to compensate the influence of disturbances and faults, an adaptive interval type II fuzzy fractional sliding mode controller is used for rotor current regulations, which employs the estimated fault and disturbance. High robustness, fast response time, low tracking error and chattering are some of the advantages of the proposed fault tolerant controller (FTC). The closed-loop stability of controllers is proved by using the Lyapunov theorem. Simulations is accomplished to evaluate the effectiveness of the proposed method. The proposed controller is compared with other controllers. Robustness against parameter uncertainty is also investigated.