The Research On Preventive Maintenance Strategy Of Wind Turbine Considering Time-varying Working Conditions

With the continuous increase of the total installed capacity of wind power in China and its proportion to the total national power generation capacity,the requirements for the safety guarantee of wind turbine units are higher,and the importance of preventive maintenance of wind turbine units is also increasing.The environment of the wind farm is rich in wind energy resources and the environment is relatively bad.Due to the continuous impact and erosion of complex and changeable working conditions,wind turbine faults occur frequently.At present,the research on preventive maintenance strategy of wind turbines mostly assumes that the working conditions of the wind turbine are constant,or the working conditions do not affect the degradation of unit performance.However,in the actual power generation process,the changeable working conditions will affect the degradation rate of the wind turbine,affect the failure time,and then affect the effectiveness of the preventive maintenance strategy.Given the above problems,from the perspective of the impact of time-varying working conditions on the reliability of wind turbines,based on the data of data acquisition and monitoring control system and the performance degradation model constructed by the multi-stage adaptive Wiener process,this thesis evaluates the working condition of the wind turbine,predicts the remaining life,and puts forward the preventive maintenance strategy of wind turbine considering time-varying working conditions,Provide theoretical and technical support for preventive maintenance of wind farm.The main research contents of this thesis are as follows:(1)Whereas the wind turbine SCADA system records the environment of the unit and the operation state parameters of internal key components,reflecting the working condition of the unit.Based on SCADA data,this thesis adopts the adaptive density-based clustering method with noise under the condition of multi density to clean the SCADA data and eliminate the interference of wrong data;The combination of entropy weight method and cloud model is used to evaluate the working condition of a wind turbine.The above method is verified by using SCADA detection data and maintenance records of the wind farm,and its effectiveness is proved.(2)To improve the timeliness of preventive maintenance,considering the influence of time-varying working conditions on the performance degradation of the wind turbine,this thesis divides the equipment degradation process into stages by using the working conditions obtained by the cloud model,establishes the state degradation model and residual life prediction model based on multi-stage adaptive Wiener process,estimates the initial value of parameters with historical data,and updates the drift coefficient with Kalman filter algorithm,The probability density distribution of remaining life is obtained,which provides support for specifying preventive maintenance strategy.By comparing with the non-phased adaptive model,the advantage of this method’s inaccuracy is verified.(3)Based on the multi-stage degradation model and residual life model,the preventive maintenance strategy of the wind turbine is studied and improved.Firstly,the traditional timebased maintenance and condition-based maintenance are analyzed,the preventive maintenance strategy model is established based on the remaining life,and the preventive maintenance strategy under the goal of optimal average cost per unit time is proposed.Finally,the simulation analysis verifies that the improved preventive maintenance strategy has a better economy.