Prognostics Health Management Method Of Wind Turbine Blades Based On The Working Condition Approximation

Wind turbine blades are the key component s of energy conversion in wind turbines,which are prone to excessive deformation and cracking of the leading and trailing edges during long-term operation under complex and changeable workin g conditions.It further led to vicious accidents such as blade breakage,collision of blade barrels,tower down,and power grid burnt.Therefore,it is very necessary to carry out research on in-service blade failure Prognostics and Health Management(PHM).Condition monitoring and forecasting are the first priority of PHM,but the existing technology is difficult to achieve real-time monitoring and forecasting of the performance status of in-service blades,resulting in the stagnation of follow-up research on health assessment,fault identification and prognostication,and health management decision.In order to solve this problem,the hydraulic-excited fatigue test bench built by our research team and the 65 k W wind power prototype of the manufacturer in Dezhou,Shandong Province are used.This paper adopts the idea of working condition approximation to realize the condition monitoring and forecasting of in-service blades indirectly,and combines its typical failure modes and working condition characteristics to propose a relatively complete PHM method for in-service blades.Finally,a 65 k W wind turbine blade is taken as the example to verify the rationality and accuracy of the method.The main contributions are as follows:(1)The real-time monitoring and forecasting method for the bending stiffness of in-service blades is proposed.The degradation law and degradation formula of the bending stiffness of the blade were obtained through the working condition approximation experiment under the influence of fact ors such as fatigue load,low-speed impact load,ambient temperature and wind speed distribution.Th e formula is used to analyze the bending stiffness degradation of the blade at different service stages.The analysis results are consistent with the measur ed results and meet the accuracy requirements,and the rationality of the method of working condition approximation and the accuracy of the formula are verified.(2)The real-time monitoring and forecasting method for the crack resistance toughness of the leading and trailing edges of the in-service blades is proposed.The degradation law and formula of the leading and trailing edges crack resistance toughness under various levels of cracking loads,as well as the law of the influence of the environmental temperature on the crack resistance toughness detection value and the degradation rate are obtained through the working condition approximation experiment.The formula is used to analyze the degradation of the crack resistance toughness of the blade at different service stages.The analysis results are consistent with the measured results and meet the accuracy requirements,which verifies the rationality of the method of working conditions approximation and the accuracy of the formula.(3)The method of health assessment and Remaining Healthy Life(RHL)prediction for in-service blade is proposed.A multi-weight model of health assessment method for in-service blades is proposed based on the Grey Relation(GR)model,which is used to evaluate the health statu s of the blades in different service stages in real time,accurately identify the unhealthy operating status of the blades and successfully trigger the fault identification and prediction mechanism.The method can provide alternative models for different e valuation needs and provide credible evaluation intervals for inexperienced evaluators.(4)The method of fault identification and prognostication for in-service blades is proposed.The fault dictionary is compiled and the fault ignition route map is constructed according to the static load calibration records,and combined with the service conditions and performance status to realize the fault status identification.The Local Alignment(LA)method and Markov process(MP)are used to achieve 10 min and 24 h fault prediction.Long-term fault trend prediction is obtained based on wind speed and temperature statistics.The method is used to analyze the fault status of the blades in service at different stages of service.Based on the analysis results of prognostication,the Remaining Useful Life(RUL)of in-service blades is prddicted.The analysis results are consistent with the actual measurement records to verify the rationality and accuracy of the method,thus indirectly verifying the accuracy of the health assessment model.(5)The method of health management decision for in-service blades is proposed.According to the current situation that the existing decision-making model lacks quantitative basis and mapping relationship,the index quantification rule is formulated.Based on the Analytic Hierarchy Process-Fuzzy decision(AHP-F)model,a clear mapping relationship is established between decision goals and alternative strategies,and accurate health management decisions are realized.This method is used to optimize the health management strategies of blades in different service stages.The results show that the daily maintenance mode is no longer applicable after the blade has been in service for more than 6 years.At this time,health management decisions based on health assessment and failure prediction must be carried out to avoid the adverse effects of subjective judgment on the decision results.Finally,a health management interface is designed based on Lab VIEW software to dynamically display the relevant results of the PHM of the in-service blades.