中国邮电高校学报(英文) ›› 2017, Vol. 24 ›› Issue (6): 67-73.doi: 10.1016/S1005-8885(17)60243-7

• Artificial Intelligence • 上一篇    下一篇

Anomaly detection in smart grid based on encoder-decoder framework with recurrent neural network

Zheng Fengming, Li Shufang, Guo Zhimin, Wu Bo, Tian Shiming, Pan Mi   

  1. 1. 北京邮电大学信息与通信工程学院
    2. 北京邮电大学
    3. 国网河南省电力公司电力科学研究院
    4. 中国电力科学研究院
  • 收稿日期:2017-04-28 修回日期:2017-09-28 出版日期:2017-12-30 发布日期:2017-12-01
  • 通讯作者: 郑凤鸣 E-mail:zfm@bupt.edu.cn
  • 基金资助:
    Business Integration and Data Sharing Service Technology Based on Through Information of Operation and Distribution (2016 state Grid Technology Project).

Anomaly detection in smart grid based on encoder-decoder framework with recurrent neural network

Zheng Fengming, Li Shufang, Guo Zhimin, Wu Bo, Tian Shiming, Pan Mi   

  1. 1. 北京邮电大学信息与通信工程学院 2. 北京邮电大学 3. 国网河南省电力公司电力科学研究院 4. 中国电力科学研究院
  • Received:2017-04-28 Revised:2017-09-28 Online:2017-12-30 Published:2017-12-01
  • Contact: Feng-Ming ZHENG E-mail:zfm@bupt.edu.cn
  • Supported by:
    Business Integration and Data Sharing Service Technology Based on Through Information of Operation and Distribution (2016 state Grid Technology Project).

摘要: Anomaly detection in smart grid is critical to enhance the reliability of power systems. Excessive manpower has to be involved in analyzing the measurement data collected from intelligent motoring devices while performance of anomaly detection is still not satisfactory. This is mainly because the inherent spatio-temporality and multi-dimensionality of the measurement data cannot be easily captured. In this paper, we propose an anomaly detection model based on encoder-decoder framework with recurrent neural network (RNN). In the model, an input time series is reconstructed and an anomaly can be detected by an unexpected high reconstruction error. Both Manhattan distance and the edit distance are used to evaluate the difference between an input time series and its reconstructed one. Finally, we validate the proposed model by using power demand data from University of California, Riverside (UCR) time series classification archive and IEEE 39 bus system simulation data. Results from the analysis demonstrate that the proposed encoder-decoder framework is able to successfully capture anomalies with a precision higher than 95%.

关键词: smart grid, encoder-decoder framework, anomaly detection, time series mining

Abstract: Anomaly detection in smart grid is critical to enhance the reliability of power systems. Excessive manpower has to be involved in analyzing the measurement data collected from intelligent motoring devices while performance of anomaly detection is still not satisfactory. This is mainly because the inherent spatio-temporality and multi-dimensionality of the measurement data cannot be easily captured. In this paper, we propose an anomaly detection model based on encoder-decoder framework with recurrent neural network (RNN). In the model, an input time series is reconstructed and an anomaly can be detected by an unexpected high reconstruction error. Both Manhattan distance and the edit distance are used to evaluate the difference between an input time series and its reconstructed one. Finally, we validate the proposed model by using power demand data from University of California, Riverside (UCR) time series classification archive and IEEE 39 bus system simulation data. Results from the analysis demonstrate that the proposed encoder-decoder framework is able to successfully capture anomalies with a precision higher than 95%.

Key words: smart grid, encoder-decoder framework, anomaly detection, time series mining