3种机器学习方法预报风电功率的对比  

A comparative study of three machine learning methods for forecasting wind power

作　　者：陈金车 王金艳[1] 苏士翔 孙彩霞 谢祥珊 CHEN Jin-che;WANG Jin-yan;SU Shi-xiang;SUN Cai-xia;XIE Xiang-shan(Key Laboratory of Arid Climatic Changing and Disaster Reduction of Gansu Province,College of Atmospheric Sciences,Lanzhou University,Lanzhou 730000,China)

机构地区：[1]兰州大学大气科学学院甘肃省干旱气候变化与减灾重点实验室,兰州730000

出　　处：《兰州大学学报（自然科学版）》2022年第1期124-129,136,共7页Journal of Lanzhou University(Natural Sciences)

基　　金：甘肃省自然科学基金项目(21JR7RA501);国家重点研发计划项目(2020YFA0608402)。

摘　　要：根据甘肃省华家岭风力发电场的风场变化特征,利用风电场2017年8月-2018年7月的风电功率监测数据及同期欧洲中期天气预报中心的数值模式预报资料,用随机森林(RF)算法分析和筛选出主要的预报因子,分别选择RF、极限学习机和支持向量机3种机器学习方法建立预报模型,通过对比预报效果,得出适合的预报方法和模型.结果表明,RF算法的平均预报均方根误差为15.6%,预报效果优于极限学习机和支持向量机(预报均方根误差分别为16.8%和17.2%).RF算法在风电功率的短期预报方面取得了更好的效果,预报值与实际监测值更加接近;基于3种机器学习算法建立的风电功率预报模型的预报结果误差值都会随着风速的增大而增大,随着风速的减小而减小.Based a the study of the wind field change characteristics of Huajialing Wind Farm in Gansu,and using the wind power monitoring data of the wind farm from August 2017 to July 2018 and the numerical weather forecast data of the European Forecast Center during the same period,the main predictors were selected through the random forest(RF)algorithm.RF extreme learning machine and support vector machine were used to establish forecast models,and the forecast effects were tested to get a suitable method and model.The results showed that the prediction effect of RF was better than extreme learning machine and support vector machine,and the average root mean square error of forecast was only15.6%(extreme learning machine and support vector machine were 16.8%and 17.2%respectively).RF could achieve better results in short-term forecasting of wind power,and the forecast data was closer to the actual data.Regardless of the wind power forecast model based on the machine learning algorithm,the error value of the forecast result would increase with the increase in wind speed,and decrease with the decrease in wind speed.

关 键 词：机器学习 随机森林算法 极限学习机 支持向量机 风电功率预报 

分 类 号：P456[天文地球—大气科学及气象学]