沙漠与绿洲气象2025，Vol.19Issue(5)：36-41,6.DOI:10.12057/j.issn.2097-6801.2405.08001
基于机器学习的风电场风机覆冰发生、消融起始预报研究
Machine Learning-Based Forecasting of Wind Turbines Ice Accretion Occurrence and Melting Initiation in Wind Farms
摘要
Abstract
Using the ice accretion observation records from a wind farm located in a mountainous region of Hubei Province from November 2017 to February 2022,combined with high-precision meteorological reanalysis data,and considering the topographic conditions of the wind turbines for interpolation,hourly meteorological element data for each wind turbine was obtained.Changes in meteorological factors during the onset and melting phases of wind turbine ice accretion were analyzed.The results show that:During the onset phase of ice accretion,the temperature continued to decrease and stabilized after the onset of icing.The relative humidity steadily increased,reaching a peak at the moment of icing onset,followed by a slow decline.The wind speed remained relatively stable with slight fluctuation before and after the onset of ice accretion,maintaining a consistently high level.During the melting phase of wind turbine ice accretion,the temperature continuously increased,peaking at the moment of ice melting.The relative humidity continuously decreased,reaching its lowest value at the time of ice melting.The wind speed remained steady but was significantly lower compared to the icing phase.Based on the ice accretion observation samples,three different methods,AdaBoost,logistic regression,and Fisher discriminant,were used to establish discrimination models for the onset and melting of wind turbine ice accretion.Among these,the model established using the AdaBoost machine learning method exhibited significantly higher accuracy than the other two methods,with an accuracy of 88.3%for the icing onset model and 83.6%for the ice melting model.关键词
山区风电场/风机覆冰/临界值/机器学习Key words
mountain wind farm/wind turbine icing/critical value/machine learning分类
天文与地球科学引用本文复制引用
张荣,孙朋杰,孙舒,许沛华..基于机器学习的风电场风机覆冰发生、消融起始预报研究[J].沙漠与绿洲气象,2025,19(5):36-41,6.基金项目
湖北省自然科学基金联合(2022CFD131) （2022CFD131）
