Research On Observer Based Speed Sensor Fault Diagnosis For Wind Turbine

In this paper,the horizontal axis double-fed wind turbine is studied.The fault diagnosis of the speed sensor for measuring the low speed shaft and the speed of the generator is studied.By establishing the model of sensor fault system,the fault diagnosis method based on observer is used to diagnose the sensor fault.First,the development process and current situation of fault diagnosis technology are introduced,and the research condition and existing problems of wind turbine fault diagnosis are analyzed.Then it introduces the fault diagnosis method based on observer,elaborates the principle and design of the two kinds of important diagnostic method of unknown input observer and sliding mode observer,establish the theoretical foundation for the follow-up study of fault diagnosis.Secondly,introduces double-fed wind turbine basic structure,working principle and speed sensor fault type and based on low speed shaft and generator to establish the transmission system speed sensor fault model under the influence of interference and noise.Then introduces the Kalman filter algorithm to design of unknown input observer(UIO)feedback matrix,the residual signal can effectively suppress noise and interference effect,achieve decoupling effect,then uses the maximum likelihood estimation residuals and establishes threshold,the simulation shows that the designed observer can effectively realize the wind turbine fault diagnosis of common speed sensor.Considering the transmission system under sensor failure model is more complex,establishes the reduced order model,then takes saturation function to establish sliding mode observer(SMO),combines the measured difference between sliding mode gain and speed measurement difference,the simulation results demonstrate the effectiveness of the reduced order model of improved sliding mode observer based on speed sensor fault diagnosis.Finally,by analyzing and establishing sensor fault wind turbine model,H_/-H_∞ optimization method is adopted to design the observer feedback matrix,combined with iterative algorithm for solving the optimal balance of robustness and sensitivity.The simulation results show that the optimization design method of fault observer system to achieve the best balance at the same time more sensitive,more robust to interference,which can effectively detect sensor fault of wind turbine.