华东师范大学学报(自然科学版) ›› 2023, Vol. 2023 ›› Issue (5): 135-146.doi: 10.3969/j.issn.1000-5641.2023.05.012

• 数据分析 • 上一篇    

基于Autoformer的电力负荷预测与分析研究

唐利涛1(), 张智勇1, 陈俊1, 许林娜2, 钟佳晨2, 袁培森2,*()   

  1. 1. 南方电网广西电网有限责任公司计量中心, 南宁 530024
    2. 南京农业大学 人工智能学院, 南京 210031
  • 收稿日期:2023-06-30 接受日期:2023-07-22 出版日期:2023-09-25 发布日期:2023-09-15
  • 通讯作者: 袁培森 E-mail:515986188@qq.com;peiseny@163.com
  • 作者简介:唐利涛, 女, 硕士, 高级工程师, 主要研究方向为计量自动化管理及技术. E-mail: 515986188@qq.com
  • 基金资助:
    国家自然科学基金(61877018); 上海市大数据管理系统工程研究中心开放基金(HYSY21022)

Research on Autoformer-based electricity load forecasting and analysis

Litao TANG1(), Zhiyong ZHANG1, Jun CHEN1, Linna XU2, Jiachen ZHONG2, Peisen YUAN2,*()   

  1. 1. Measurement Center of Guangxi Power Grid Co. Ltd., Nanning 530024, China
    2. College of Artificial Intelligence, Nanjing Agricultural University, Nanjing 210031, China
  • Received:2023-06-30 Accepted:2023-07-22 Online:2023-09-25 Published:2023-09-15
  • Contact: Peisen YUAN E-mail:515986188@qq.com;peiseny@163.com

摘要:

新一代电网是未来智能电网发展的主要方向, 而电力负荷精准预测是智能电网的一项重要基础工作. 为了提高智能电力系统负荷预测的精确度, 本文基于自相关机制的预测模型Autoformer, 对负荷数据集进行了特性分析, 在原模型中加入特征提取层, 从编码层数、解码层数、学习率和批量大小等方面优化了模型参数, 实现了周期灵活的负荷预测. 在真实数据集上进行了实验和分析, 实验结果表明, 本文模型在预测效果上表现更好, MAE (mean absolute error)和MSE (mean squared error)分别为0.2512和0.1915, 决定系数为0.9832. 与其他方法相比, 本文方法负荷预测效果更好.

关键词: Autoformer, 电力负荷预测, 自相关机制, 特征提取, 负荷特性

Abstract:

Next-generation power grids is the main direction of future smart grid development, and the accurate prediction of power loads is an important basic task of smart grids. To improve the accuracy of load prediction in smart power systems, this work characterized the load dataset based on an Autoformer, a prediction model with an autocorrelation mechanism; adds a feature extraction layer to the original model; optimized the model parameters in terms of the number of coding layers, decoding layers, learning rate, and batch size; and achieved cycle-flexible load prediction. The experimental results show that the model performs better in prediction, with an MAE, MSE, and coefficient of determination of 0.2512, 0.1915, and 0.9832, respectively. Compared with other methods, this method has better load prediction results.

Key words: Autoformer, power load forecasting, autocorrelation mechanism, feature extraction, load characteristics

中图分类号: