| With countries around the world attach great importance to the development and utilization of wind energy,the number of wind farms and wind turbines is increasing rapidly.However,the installation location of wind turbines is remote and the environment is harsh,which causes the components of wind turbines to age seriously and the faults of various subsystems occur frequently.In order to reduce the faults and improve the reliability of wind turbines,this paper gives an effective technology for wind turbine aging assessment and fault detection based on the SCADA data of different wind farms.Firstly,by studying the existing literature,the current status of wind power generation is analyzed,and the internal structure,operating principle,unit classification and common fault manifestations of doubly-fed asynchronous wind turbines as the research objects are discussed in depth.The function,internal structure and development of the SCADA system are briefly described.The state parameters of the SCADA system are enumerated based on historical data in wind farms.Secondly,a fault detection method based on IOWA operator combination prediction model is proposed for aiming at the problem of frequent failures of key components of wind turbines.The parameters related to fault detection are selected according to the grey correlation degree,and the IOWA operator is introduced into a combined model of BPNN and NSET,and the final prediction model is established using the residual optimization problem.The basic window technology is improved and applied to the threshold method to judge the turbine fault,and the problem of threshold setting for predicting temperature error is solved.The front bearing temperature of the generator is taken as the target parameter,and the superiority of the combined model based on the IOWA operator is demonstrated by comparing the evaluation index of the single model,the traditional combined model and the combined model based on the IOWA operator.The reliability of the combined model is verified by selecting four sets of actual data in the wind farm.The result shows that the prediction model based on the IOWA operator can achieve the effectof early warning of unit equipment faults.Finally,an overall aging evaluation method for wind turbines based on information fusion is proposed for the degradation of wind turbine performance caused by aging.The output power,cabin vibration,main bearing temperature,generator rear bearing temperature,gearbox speed and cabin temperature are selected in SCADA data as the aging evaluation criteria.The weights of each evaluation criterion are obtained through the neural network.The information of evaluation criteria is fused to establish an overall aging evaluation model.Wind turbines with different aging degrees are selected to verify the feasibility of the aging evaluation model.The result shows that the aging evaluation model is simple,effective and highly reliable,which provides new ideas for the subsequent research on the health status of wind turbines. |
