摘要
电力大数据服务是智能电网建设的关键,提出了基于改进AP聚类的用电行为分析方法和基于随机森林的电力负荷预测方法。针对AP聚类分析用电行为存在的复杂度较高问题,利用熵权法确立指标权值,改进相似度计算方式,实现了用户用电行为的快速准确分析。针对电力负荷预测问题,采用模糊C均值构建历史相似日样本集,利用随机森林预测电力负荷。为提高电力数据服务运行效率,构建了基于Spark框架的并行数据处理平台。实验结果表明,提出方法能够有效提取用户的用电行为和预测电力负荷,且性能优于现有方法。
Power system and big data service is the key to the construction of smart grid.This paper proposes a power consumption behavior analysis method based on improved AP clustering and a power load forecasting method based on random forest.Since the high complexity of AP clustering analysis of electricity consumption behavior,entropy weight method is used to establish the index weight,and to improve the similarity calculation method,which realize the fast and accurate analysis of user electricity consumption behavior.Aiming at the problem of power load forecasting,fuzzy c-means is used to construct historical similar daily sample set,and random forest is used to forecast power load.In order to improve the operation efficiency of power data service,a parallel data processing platform based on spark framework is constructed.Experiment results show that the proposed method can effectively extract the user’s electricity consumption behavior and predict the power load,and its performance is better than the existing methods.
作者
孙煜华
李情
张梦清
SUN Yu-hua;LI Qing;ZHANG Meng-qing(Guangzhou Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Guangzhou 510610,China)
出处
《信息技术》
2021年第5期102-108,共7页
Information Technology
关键词
电力大数据
用电行为分析
负荷预测
AP聚类
随机森林
power system and big data
power consumption behavior analysis
load forecasting
AP clustering
random forest
