中国电力 ›› 2021, Vol. 54 ›› Issue (9): 17-23.DOI: 10.11930/j.issn.1004-9649.202003035

• 电网 • 上一篇    下一篇

基于CNN-BiGRU-NN模型的短期负荷预测方法

曾囿钧1, 肖先勇1, 徐方维1, 郑林2   

  1. 1. 四川大学 电气工程学院,四川 成都 610065;
    2. 国网四川省电力公司绵阳供电公司,四川 绵阳 621000
  • 收稿日期:2020-03-05 修回日期:2020-05-26 出版日期:2021-09-05 发布日期:2021-09-14
  • 作者简介:曾囿钧(1993-),男,通信作者,硕士研究生,从事电力系统负荷预测研究,E-mail:1342382572@qq.com;肖先勇(1968-),男,博士,教授,从事电能质量与优质电力、智能电网与电网安全、超导电力等研究,E-mail:xiaoxianyong@163.com;徐方维(1978-),女,博士,副教授,从事电能质量及谐波等研究,E-mail:xfwlovely@126.com;郑林(1976-),硕士,高级工程师,从事电力系统负荷预测研究,E-mail:zhenglin@163.com
  • 基金资助:
    国家自然科学基金面上资助项目(新一代电力系统中谐波发射水平评估理论与方法,51877141)

A Short-Term Load Forecasting Method Based on CNN-BiGRU-NN Model

ZENG Youjun1, XIAO Xianyong1, XU Fangwei1, ZHENG Lin2   

  1. 1. College of Electrical Engineering, Sichuan University, Chengdu 610065, China;
    2. Mianyang Power Supply Company, State Grid Sichuan Power Supply Company, Mianyang 621000, China
  • Received:2020-03-05 Revised:2020-05-26 Online:2021-09-05 Published:2021-09-14
  • Supported by:
    This work is supported by National Natural Science Foundation of China (Theory and Method of Harmonic Emission Level Evaluation in New Generation Power System, No.51877141)

摘要: 为充分挖掘蕴含在大量采集数据中的有效信息,提高短期负荷预测精度,提出一种基于卷积神经网络(CNN)和双向门控循环单元(BiGRU)、全连接神经网络(NN)的混合模型的短期负荷预测方法,将海量的历史负荷数据、气象信息、日期信息按时间滑动窗口构造特征图作为输入,先利用CNN提取特征图中的有效信息,构造特征向量,再将特征向量作为BiGRU-NN网络的输入,采用BiGRU-NN网络进行短期负荷预测。以2016年举办的全国第九届电工数学建模竞赛试题A题中的负荷数据作为实际算例,实验结果表明:该方法与DNN神经网络、GRU神经网络、CNN-LSTM神经网络短期负荷预测法相比,有更高的预测精度。

关键词: 电力系统, 短期负荷预测, 卷积神经网络, 双向门控循环单元, 卷积神经网络-双向门控循环单元神经网络混合模型

Abstract: In order to fully mine the effective information contained in a large number of collected data and improve the accuracy of short-term load forecasting, a short-term load forecasting method is proposed based on a hybrid model of convolutional neural network (CNN), bidirectional gated recurrent unit (BiGRU), and fully connected neural network (NN). The massive historical load data, meteorological information, and date information are taken to construct feature maps according to time sliding windows. Firstly, the CNN is used to extract valid information from the feature maps to construct feature vectors. And then, by taking the feature vectors as the inputs, the BiGRU-NN network is used to make short-term load forecasting. The load data in the test question A of the Ninth National Electrical Mathematics Modeling Contest held in 2016 are taken as an actual computation example, and the experimental results show that this method has higher accuracy in short-term load forecasting than GRU neural network, DNN neural network, and CNN-LSTM neural network.

Key words: power system, short-term load forecasting, convolutional neural network, bidirectional gated recurrent unit, convolutional neural network-bidirectional gated recurrent unit neural network hybrid model