摘要
为解决风力发电预测不准确的问题,进一步提高风力发电场发电量的预测精度,提出了一种双层XGBoost的风电场发电量预测模型.该模型第一层为数据分析层,利用XGBoost的特征重要性排序分析SCADA数据,通过特征重要性分析排序选择特征变量作为输入;第二层为预测层,利用XGBoost回归算法建立风力发电量预测模型.为表现模型优越性,设置对比实验,将降维后的数据输入随机森林和决策树.结果表明,对比随机森林和决策树,XGBoost模型均方根误差和平均绝对误差较小,准确率可达到96.1%,具有更高的拟合度,可以有效解决非线性变量难以选择的问题,减少输入特征维度,加快预测速度.
In order to solve the problem of the inaccuracy of wind power generation prediction,to further improve the prediction accuracy of wind farm power generation,we proposed a double-layer XGBoost wind farm power generation prediction model.The first layer is the data analysis layer,using XGBoost’s feature importance ordering analysis SCADA data,through the feature importance analysis sorting to select feature variables as input.The second layer is the prediction layer,and the wind power forecasting model is established by using XGBoost regression algorithm.In order to show the superiority of the model,we set up a comparative experiment,input the data into random forest and decision tree after dimension reduction.The results show that,compared with random forest and decision tree,mean square root error and mean absolute error of XGBoost model is small,the accuracy can reach 96.1%,with higher fit.The proposed model can effectively solve the problem of difficult selection of nonlinear variables,reduce the input feature dimension and speed up the prediction speed.
作者
段亚穷
向勉
刘洪笑
周丙涛
曾曙莲
DUAN Yaqiong;XIANG Mian;LIU Hongxiao;ZHOU Bingtao;ZENG Shulian(College of Intelligent Systems Science and Engineering,Hubei Minzu University,Enshi 445000,China)
出处
《湖北民族大学学报(自然科学版)》
CAS
2022年第2期171-175,195,共6页
Journal of Hubei Minzu University:Natural Science Edition
基金
2020年硒食品营养与健康智能技术湖北省工程研究中心开放课题(PT082005).
关键词
电量预测
特征重要性
风力发电
XGBoost
新能源
electric quantity prediction
feature importance
wind power generation
XGBoost
new energy