Application Of Data Mining In Wind Turbine Fault Diagnosis

With the global warming,the available energy is declining.Renewable energy has an irreplaceable position in national energy strategies because of its cleanliness and safety.Wind turbines are a huge reserve of renewable energy,and more and more scholars are putting more energy into the research of wind turbines.Nowadays,data mining has spurred the innovation of new technologies,which have been widely applied and penetrated into all walks of life,and the wind power industry is also among them.This paper combines the monitoring and evaluation of wind turbine generators operating conditions,SCADA systems and data mining.By using the actual operating data of the wind turbine generators,real-time detection and evaluation of the wind turbine performance indicators have been done.Excessive winding temperature and gearbox temperature are important features of wind turbine generators failure.The temperature trend analysis method is used to monitor the gearbox state.The nonlinear state estimation method is improved on the flower pollination algorithm based on the beetle search,and it is used to establish the temperature prediction model of the gearbox,moreover,the experimental simulation is carried out.The experiment shows that the improved BAS-NSET algorithm is more accurate than the NSET algorithm.The value is closer to the true value and the residual is smaller.Aiming at the problem of wind turbine winding temperature prediction,this paper establishes a 3-layer BAS-BP network,taking multiple characteristics of wind turbine wheels as input units,and using the actual operation data of Beizhen Wind Farm of Guodian and Wind Power Development Co.,Ltd.as a sample for training and prediction.It is predicted to output the winding temperature prediction under different working conditions,and the UP82-1500 in the power plant is taken as an example to compare and analyze the predicted result and the measured data.The results show that the model is fast and accurate,and the applicability of BAS-BP neural network in wind turbine winding temperature prediction is verified.In this paper,the validity of the proposed BAS-NSET model and BAS-BP model for wind turbine fault prediction and the high prediction accuracy are verified.These two methods have strong applicability to such fault prediction.This paper has 17 pictures,7 tables,and 51 references.