摘要
文章通过提取海上风电功率预测的最小敏感气象因素集,建立了秒级、分钟级、小时级、日前等多时间尺度的功率预测模型,并采用考虑惯性与日周期性的LSTM网络短期风电功率预测统计适配技术实现了日前短期功率预测,采用插值方法获得了分钟级的超短期功率预测。算例分析表明,文章的多时间尺度功率预测方法具有精度高、误差小、覆盖多时间尺度的优势。
In this paper,by extracting the minimum sensitive meteorological factor set for offshore wind power prediction,a multi-time scale power prediction model is established,such as second-minute,hour-hour,day-ahead,etc.The short-time wind power prediction statistical adaptation technology of LSTM network considering inertia and diurnal periodicity is adopted to realize the short-time power prediction,and the ultra-short-time power prediction of minute is obtained by interpolation method.Numerical examples show that the proposed multi-time scale power prediction method has the advantages of high precision,small error and multi-time scale coverage.
作者
曾东
杨兴旺
吴继秀
卢嘉铭
刘泽健
ZENG Dong;YANG Xingwang;WU Jixiu;LU Jiaming;LIU Zejian
出处
《电力系统装备》
2023年第10期5-8,共4页
Electric Power System Equipment
关键词
聚类
日周期性
遗传算法
LSTM
插值方法
clustering
daily periodicity
genetic algorithm
LSTM
interpolation method