Research On Fault Early Warning Method For Wind Power Generator

Wind energy is recognized as the world’s sustainable renewable clean energy,China’s wind energy resources are particularly rich.China’s wind power is developing rapidly,the proportion of wind power in the power system is increasing,and will achieve a 15% share in 2020,becoming the third largest energy after thermal power and hydropower.At the same time,wind power management mode has changed from the previous extensive management information gradually to the simple,non centralized and information system management mode.This change will significantly improve the efficiency of wind energy utilization,and enhance the safety level of the operation of the wind farm.This paper mainly studies two problems of wind turbine fault feature extraction and early warning methods.Firstly,normal operation data such as the wind speed,power,generator temperature and other monitoring parameters are extracted from the wind power SCADA system,and then the influence of the generator is analyzed by the Relief algorithm.Finally,the relevant parameters of the larger relevant weights are extracted,and the fault feature information is extracted,and the weights are applied to the improved weighted principal component analysis algorithm.Secondly,according to the historical operation monitoring data,the data matrix is constructed to cover the normal operation space of the generator,and then the generator fault warning principal component model is established.The sliding window algorithm is used to update the model,and the normal data is added to the window in the sliding step size,and the earlier time data is eliminated to ensure the model adaptability.The fault prediction model established by principal component analysis method,if the generator has a potential fault,squared prediction error statistics changed,and will continue beyond the control limit,can be used for early warning of potential fault.Finally,this paper adopts traditional PCA algorithm,traditional APCA algorithm,sliding window KPCA analysis algorithm and data block sliding window KPCA algorithm based on established fan fault early warning model.Fault detection results shows that the improved algorithm can find early fault motor and realizes fault warning more accurately.