基于改进AlexNet-GRU深度学习网络的配电网短期负荷预测方法  

Short⁃term Load Prediction Method of Distribution Network Based on Improved AlexNet⁃GRU Deep Learning Network

作　　者：朱海南 李丰硕 孙华忠 兰栋 吴俊勇[2] ZHU Hainan;LI Fengshuo;SUN Huazhong;LAN Dong;WU Junyong(State Grid Shandong Electric Power Company Weifang Power Supply Company,Shandong Weifang 261021,China;School of Electrical Engineering,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]国网山东省电力公司潍坊供电公司,山东潍坊261021 [2]北京交通大学电气工程学院,北京100044

出　　处：《电力电容器与无功补偿》2023年第4期48-54,61,共8页Power Capacitor & Reactive Power Compensation

基　　金：国网山东省电力公司科技项目(52060419005C)。

摘　　要：负荷预测作为电力系统规划的重要环节,对于确保电网稳定运行、实现电力供需平衡等方面具有十分重要的作用。本文提出了一种基于改进AlexNet-GRU深度学习网络的配电网短期负荷预测方法。通过聚类将日负荷曲线分为不同日类型;然后根据聚类结果,建立基于改进AlexNetGRU深度学习网络的配电网短期负荷预测模型,并与传统的负荷预测方法进行对比。对某地区2013年的负荷进行预测结果表明,本文所提方法可以有效提高预测精度。As an important part of power system planning,load prediction plays a very important role in ensuring stable operation of power grid,achieving the balance of power supply and demand as well as implementing reasonable operation modes.In this paper,a short⁃term load forecasting method for distribu⁃tion network based on improved AlexNet⁃GRU deep learning network is proposed.First,the daily load curve is clustered,the operating days throughout the year are divided into five categories and different types of day are identified.At the same time,Pearson correlation analysis method is used to screen the main weather factors that affect load prediction.Then,the load data and the main weather influencing fac⁃tors are integrated.Finally,an improved AlexNet⁃GRU deep learning network distribution network load prediction model is set up.Different prediction methods are used to predict the load of a certain area in 2013,and the results show that the proposed method has significant advantages in accuracy prediction.

关 键 词：聚类 AlexNet 门控循环单元 负荷预测 

分 类 号：TM715[电气工程—电力系统及自动化] TP18[自动化与计算机技术—控制理论与控制工程]