采用互信息与随机森林算法的用户用电关联因素辨识及用电量预测方法

发布时间：2018-09-12 14:54

【摘要】：随着智能电网的不断发展,电力企业积累的大量数据为用户用电量精细化预测提供了数据基础。针对与用电量预测相关的大数据种类多、体量大、维度高和生成速度快等特点,在研究用户用电特性评价指标的基础上,提出海量用户用电特性子空间聚类分析方法,挖掘用户多种用电模式。根据不同用电模式对用户进行群体划分,并利用互信息矩阵从区域及行业经济数据、气候条件,以及电力价格等方面辨识与用户群体用电量相关联的因素,进而构建基于随机森林算法的用电量大数据预测模型。该文方法可以有效识别不同用户群体的用电关联因素,规避用电模式差异性为用电量预测带来的不利影响。仿真结果表明,该方法具有较高的预测精度,且适用于大数据分析处理。
[Abstract]:With the continuous development of smart grid, a large amount of data accumulated by power enterprises provide a data basis for accurate prediction of power consumption. In view of the characteristics of big data, which is related to power consumption prediction, such as many kinds, large volume, high dimension and fast generation speed, based on the study of the evaluation index of user's power consumption characteristics, a cluster analysis method of massive user's power consumption characteristic subspace is put forward. Excavate a variety of user power patterns. According to the different power consumption modes, the users are divided into groups, and the mutual information matrix is used to identify the factors related to the electricity consumption of the user group from the aspects of regional and industry economic data, climate conditions, and electricity price, etc. Furthermore, the forecasting model of power consumption big data based on stochastic forest algorithm is constructed. The method can effectively identify the power consumption correlation factors of different user groups and avoid the adverse effects of different power consumption modes on power consumption prediction. The simulation results show that this method has high prediction accuracy and is suitable for big data analysis.
【作者单位】：电力传输与功率变换控制教育部重点实验室(上海交通大学);中国电力科学研究院;
【基金】：国家863高技术研究发展计划项目(2015AA050203) 国家电网公司科技项目(520900150037)~~
【分类号】：TM714

【相似文献】