华东师范大学学报(自然科学版) ›› 2020, Vol. 2020 ›› Issue (5): 146-155.doi: 10.3969/j.issn.1000-5641.202091013

• 数据中台应用 • 上一篇    下一篇

基于变分自编码器的日线损率异常检测研究

张国芳1, 刘通宇2, 温丽丽1, 郭果1, 周忠新3, 袁培森2   

  1. 1. 国家电网四川省电力公司, 成都 610041;
    2. 南京农业大学 信息科学技术学院, 南京 210095;
    3. 国家电网南瑞南京控制系统有限公司, 南京 211106
  • 收稿日期:2020-08-14 发布日期:2020-09-24
  • 通讯作者: 袁培森,男,博士,副教授,研究方向为大数据处理与分析.Email:peiseny@163.com E-mail:peiseny@163.com
  • 作者简介:张国芳,女,硕士,高级工程师,研究方向为电能量计量、综合能源管控与服务.E-mail:zhanggf1261@sc.sgcc.com.cn

Research on abnormal detection of daily loss rate based on a variational auto-encoder

ZHANG Guofang1, LIU Tongyu2, WEN Lili1, GUO Guo1, ZHOU Zhongxin3, YUAN Peisen2   

  1. 1. State Grid Sichuan Electric Power Company, Chengdu 610041, China;
    2. College of Information Science and Technology, Nanjing Agricultural University, Nanjing 210095, China;
    3. NARI Technology Co. Ltd., Nanjing 211106, China
  • Received:2020-08-14 Published:2020-09-24

摘要: 采用一种基于自编码器的异常检测算法, 实现大规模日线损率数据的异常检测. 变分自编码器是一种利用反向传播算法使得输出值近似等于输入值的神经网络, 使用自编码器将原始日线损率时间序列编码, 在重建过程中记录每个时间点的重建概率, 当重建概率大于指定阈值时就判定其为异常数据. 本文利用真实日线损数据进行实验, 试验结果表明, 基于自编码器的日线损率异常检测算法具有较好的检测效果.

关键词: 自编码器, 异常检测, 日线损率

Abstract: This paper adopts an anomaly detection algorithm based on a self-encoder to achieve anomaly detection of large-scale daily line loss rate data. A variational auto-encoder is a neural network that uses the backpropagation algorithm to make the output value approximately equal to the input value. It uses the auto-encoder to encode the original daily line loss rate time series and records the reconstruction possibility at each time point during the reconstruction process. When the reconstruction possibility is greater than a specified threshold, it is classified as anomaly data. In this paper, experiments were conducted on real daily line loss data. The test results show that the proposed algorithm for abnormal detection of daily line loss rate data based on an auto-encoder has good detection capability.

Key words: auto-encoder, anomaly detection, daily line loss rate

中图分类号: