The climate condition of a wind farm has a significant influence on the reliability of wind turbines. The climate condition varies with season in a year and hence the reliability changes in a complex way. The purpose ...The climate condition of a wind farm has a significant influence on the reliability of wind turbines. The climate condition varies with season in a year and hence the reliability changes in a complex way. The purpose of this paper is to model the effect of climate condition on field reliability of wind turbines. The reliability is measured by monthly-averaged mean time between failures(MTBF), and the climate conditions are described by variables of monthly-averaged temperature, relative humidity, rainfall and wind speed. Referring to the physicsof-failure models in accelerated life testing(ALT), we develop a quantitative relation between the MTBF and the climate variables. For a set of field data, the model parameters are estimated by regression, and the insignificant variables are gradually deleted based on the P-value of the regression coefficients. The resulting model is useful for maintenance workload forecasting and preventive maintenance planning, and has a potential to be used in online failure prediction.展开更多
基金the National Natural Science Foundation of China(No.71371035)
文摘The climate condition of a wind farm has a significant influence on the reliability of wind turbines. The climate condition varies with season in a year and hence the reliability changes in a complex way. The purpose of this paper is to model the effect of climate condition on field reliability of wind turbines. The reliability is measured by monthly-averaged mean time between failures(MTBF), and the climate conditions are described by variables of monthly-averaged temperature, relative humidity, rainfall and wind speed. Referring to the physicsof-failure models in accelerated life testing(ALT), we develop a quantitative relation between the MTBF and the climate variables. For a set of field data, the model parameters are estimated by regression, and the insignificant variables are gradually deleted based on the P-value of the regression coefficients. The resulting model is useful for maintenance workload forecasting and preventive maintenance planning, and has a potential to be used in online failure prediction.
