Active Fault-Tolerant Control Strategy Research For Wind Energy Conversion System

Wind energy conversion system is an extremely complex nonlinear system. Besides random of wind speed and directions, fluctuations in the grid parameters and atmospheric conditions, as well as other factors all constitute interference of wind energy conversion system. So, stability of wind energy conversion system has received more and more attentions in recent years. Research carried out on operating principle of wind energy conversion system as well as advanced active fault-tolerant control theory at home and abroad, and studies about active fault-tolerant control for wind energy conversion system have an important practice meaning.Theories of fault diagnosis and fault-tolerant control at home and abroad, as well as research status of fault-tolerant control for wind energy conversion system in recent years are elaborated. Besides, in-depth study of wind energy conversion system mathematical model is carried out. Largest wind energy capture control model of the system is built for simulation analysis. Based on the above work, according to the characteristics that sensor phase fault will cause deviation from the normal value of the electromagnetic torque, dynamic grey prediction model is applied to the electromagnetic torque value prediction, and then determine whether the sensor fault occurs or not. When the fault diagnosis unit detects real-time that sensor phase fault occurs, fault-tolerant control unit responds immediately, and roll of the failed circuit is allocated to the other state feedback circuits for system reconfiguration, making system owning as similar as possible performance before sensor phase fault occurs.For double-fed wind energy conversion system actuator drift and deviation fault, the RBF(Radial Basis Function) neural network is applied to the design of system fault diagnosis unit. And then using RBF adaptive neural network, the actuator drift and deviation fault is reconstructed on-line according to observed system state value. And switch gain of sliding mode controller is adjusted on-line according to the fault observed value, realizing system fault tolerant control. Ensuring that no matter actuator drift fault happened, nor actuator deviation fault happened, fault can be effectively reconstructed by RBF adaptive neural network. System stability can be maintained by compensating the effect of actuator fault automatically.For actuator drift fault and deviation fault of wind energy conversion system, based on the prediction control theory as well as iterative algorithm, a fault prediction observer is designed. Through repeated iteration, the virtual fault introduced is adjusted constantly by the difference between the fault prediction observer output and the actual system output within the time span, so that it can effectively fit the actual value of the actuator fault. Then, structure of sliding fault-tolerant controller is adjusted real-time according to fault observed value. Experimental results show that whatever actuator drift fault or deviation fault occurred, good fault tolerance ability of the system under sliding mode fault-tolerant controller can be guaranteed.