基于偏最小二乘回归的企业用电量预测研究  

Research on Enterprise Electricity Consumption Prediction Based on Partial Least Squares Regression

作　　者：胡敏 张跃伟 高孝天 HU Min;ZHANG Yuewei;GAO Xiaotian(Shanghai Electrical Apparatus Research Institute,Shanghai 200063,China;Shanghai Electrical Apparatus Research Institute(Group)Co.,Ltd.,Shanghai 200063,China)

机构地区：[1]上海电器科学研究院,上海200063 [2]上海电器科学研究所(集团)有限公司,上海200063

出　　处：《电器与能效管理技术》2023年第6期58-62,共5页Electrical & Energy Management Technology

基　　金：上海市2021年度“科技创新行动计划”高新技术领域项目(215111044000)。

摘　　要：随着电力改革深入推进,电力市场化建设进一步加快。用电量分析与预测成为供电企业及用电企业关注的焦点,另外用电量也可以反映企业当前的运行状况和发展趋势,对生产型企业具有重要的意义。但是,企业月度用电预测存在样本数量较少以及数据精度不足等问题,常见的机器学习方法在该应用中适用性不强。通过采用偏最小二乘(PLS)法解决上述问题,处理了所获取数据之间本身的相关性较高的问题。基于PLS建立了企业用电量预测的回归模型,并通过数据仿真验证了模型的有效性和准确性。With thedevelopment of power reform,the construction of electricity marketization has further accelerated.The analysis and prediction of electricity consumption have become the focus of attention for power supply enterprises and electricity consuming enterprises.Electricity consumption can also reflect the current operation status and development trend of enterprises,which is of great significance to production-oriented enterprises.However,the data of power consumption is less and the data accuracy is insufficient.The common machine learning methods are not suitable for this application.Therefore,the partial least squares(PLS)method is used to solve these problems.At the same time,the problem of high correlation between the acquired data can also be handled.Based on the PLS method,the regression model of enterprise electricity consumption prediction is established,and the validity and accuracy of the model are verified by the data simulation.

关 键 词：机器学习 用电预测 偏最小二乘 高相关性 

分 类 号：TM71[电气工程—电力系统及自动化]