基于时间序列聚类算法优化下的多变量短期负荷预测模型研究  

Research on Multi-variable Short-term Load Forecasting Model Optimized by Time Series Clustering Algorithm

作　　者：徐洁 XU Jie(China Energy(Guangdong)Comprehensive Energy Company,Guangdong 510000)

机构地区：[1]国能(广东)综合能源有限公司,广东广州510000

出　　处：《能源科技》2024年第2期20-23,共4页Energy Science and Technology

摘　　要：为提高短期负荷预测的精度问题,针对短期负荷预测的特点,采用了对海量序列数做数据增强聚类操作,和外部输入变量(天气因素)并行处理,提出了基于时间序列聚类算法优化下的多变量短期负荷预测模型,并对某电力售电公司进行了实际操作。结果表明:该方法大幅提升了模型的预测精度和实用能力。To improve the short-term load forecasting accuracy,considering the characteristics of short-term load forecast,a multi-variable short-term load forecasting model optimized by a time series clustering algo-rithm is proposed.This model performs data enhancement clustering operations on massive sequences and pro-cesses external input variables(weather factors)in parallel.The model has been implemented in practical op-erations for an electricity sales company.The results show that the method significantly improves the forecast-ing accuracy and utility of the model.

关 键 词：短期负荷预测 时间序列 聚类分析 天气 

分 类 号：F426.61[经济管理—产业经济]