摘要
为解决传统水电预测模型因原始数据处理方式单一而导致预测精度低的问题,提出了趋势导向的极限学习机预测模型。模型首先对小水电群的出力历史数据进行了平滑及异常数据修复处理,接着从处理后的功率曲线中提取功率变化的趋势,并将趋势作为极限学习机模型的新特征输入,为极限学习机的预测提供一个正确且唯一的预测趋势,极大地提升了小水电发电量预测的准确率。模型已应用于广西全区小水电群的发电量预测,具有广泛的应用前景。
A trend-guided extreme learning machine(TG-ELM)prediction model was proposed to solve the problem of low prediction accuracy caused by single processing mode of original data in the traditional hydropower prediction model.Firstly,historical data of the small hydropower group was smoothed,and abnormal data was repaired.Then,the trend of power change was extracted from the processed power curve.This trend,taken as the new feature input of the ELM model,provided a correct and unique prediction trend for the ELM,thus greatly improving the accuracy of prediction of small hydropower generating capacity.The proposed model was applied to generation prediction of small hydropower groups across Guangxi and could have a broad application prospect.
作者
陈晓兵
吴剑锋
黄馗
练椿杰
江雄烽
祝云
张弛
Chen Xiaobing;Wu Jianfeng;Huang Kui;Lian Chunjie;Jiang Xiongfeng;Zhu Yun;Zhang Chi(Guangxi Power Grid Co.,Ltd.,Nanning Guangxi 530023,China;Guangxi Key Laboratory of Power System Optimization and Energy-saving Technology(Guangxi University),Nanning Guangxi 530004,China)
出处
《电气自动化》
2020年第2期73-75,共3页
Electrical Automation
关键词
极限学习机
趋势导向
小水电群
功率预测
数据平滑
extreme learning machine
trend guide
small hydropower group
power prediction
data smoothing
