Research On The Prediction Of Short Circuit Faults In The Rotor Windings Of Wind Turbines

The inter-turn short-circuit fault of the rotor winding is an electrical fault that occurs frequently in wind turbines.Although the initial impact of the fault on the wind power system is not heavy,if it is not carried out or not repaired in time,it will further deteriorate the fault and affect stable operation of the system.Therefore,it is a great significance for the effective operation and maintenance of the entire wind farm to carry out predictive research on the rotor winding faults of wind turbines,and to realize the transition from the traditional "post-mortem repair" to "situ repair".This paper summarizes,compares and analyzes the current wind turbine fault prediction methods to select suitable prediction methods to predict the inter-turn shortcircuit faults of the rotor windings of wind turbines.The paper considered competently the complexity and variability of the environment where the wind turbine located.Compared with other prediction methods,the prediction method based on gray theory has better applicability.It has low data requirements but high model efficiency,and it can handle non-stationary time.Therefore,GM(1,1)is selected as the rotor winding fault prediction method.Secondly,the feature extraction methods for wind turbine rotor faults are explained and compared,and the FFT with Rife Vincent(I)-5 window is selected as the feature extraction method for fault prediction.Furthermore,most of the current studies on wind turbine winding failures do not take into account the role of the control system.In this paper,the entire wind power system simulation model including the double-fed wind motor rotor winding inter-turn short-circuit fault module was built under the Matlab/Simulink platform.Finally,by introducing ELM neural network algorithm to improve the shortcomings of GM(1,1)prediction,a rotor fault prediction model based on gray ELM is proposed to improve the prediction accuracy of gray model.Experimental simulations show that the combined gray ELM prediction model compares with gray GM(1,1),ELM network and gray neural network(GNN)and other prediction models,which has better prediction effect on rotor failure of wind turbines and Performance can provide a certain reference for achieving predictive maintenance of wind turbines,improving operating efficiency and reducing maintenance costs.