《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (12): 3799-3805.DOI: 10.11772/j.issn.1001-9081.2022111796

• 网络空间安全 • 上一篇    下一篇

基于时序异常检测的动力电池安全预警

张安勤1,2, 王小慧1()   

  1. 1.上海电力大学 计算机科学与技术学院,上海 201306
    2.汕头大学地方政府发展研究所,广东 汕头 515063
  • 收稿日期:2022-12-06 修回日期:2023-03-16 接受日期:2023-03-23 发布日期:2023-04-10 出版日期:2023-12-10
  • 通讯作者: 王小慧
  • 作者简介:张安勤(1974—),女,安徽六安人,副教授,博士,主要研究方向:数据挖掘、普适计算;
  • 基金资助:
    广东省人文社会科学重点研究基地—汕头大学地方政府发展研究所开放基金课题(07422002)

Power battery safety warning based on time series anomaly detection

Anqin ZHANG1,2, Xiaohui WANG1()   

  1. 1.College of Computer Science and Technology,Shanghai University of Electric Power,Shanghai 201306,China
    2.Local Government Development Research Institute of Shantou University,Shantou Guangdong 515063,China.
  • Received:2022-12-06 Revised:2023-03-16 Accepted:2023-03-23 Online:2023-04-10 Published:2023-12-10
  • Contact: Xiaohui WANG
  • About author:ZHANG Anqin, born in 1974, Ph. D., associate professor. Her research interests include data mining, ubiquitous computing.
  • Supported by:
    Open Fund Project of Guangdong Province Key Research Base of Humanities and Social Sciences — Local Government Development Research Institute of Shantou University(07422002)

摘要:

电动汽车由于电池内部异常情况无法得到及时预测与预警,易导致事故发生,给驾驶员和乘客的生命和财产安全带来严重威胁。针对上述问题,提出基于Transformer和对比学习的编码器解码器(CT-ED)模型用于多元时间序列异常检测。首先,通过数据增强构造一个实例的不同视图,并利用对比学习捕获数据的局部不变特征;其次,基于Transformer对数据从时间依赖和特征依赖两方面进行编码;最后,通过解码器重构数据,计算重构误差作为异常得分,对实际工况下的机器进行异常检测。在SWaT、SMAP和MSL这3个公开数据集和电动汽车动力电池(EV)数据集上的实验结果表明,所提模型的F1值对比次优模型分别提升6.5%、1.8%、0.9%和7.1%。以上结果表明CT-ED适用于不同实际工况下的异常检测,平衡了异常检测的精确率和召回率。

关键词: 时间序列, 异常检测, 对比学习, 多头注意力, 自动编码器

Abstract:

Abnormal situations inside the vehicle battery cannot be predicted and warned in time, which leads to electric vehicle accidents and brings serious threats to drivers and passengers’ life and property safety. Aiming at the above problem, a Contrastive Transformer Encoder Decoder (CT-ED) model was proposed for multivariate time series anomaly detection. Firstly, different views of an instance were constructed through data augmentation, and the local invariant features of the data were captured by contrastive learning. Then, based on Transformer, the data were encoded from two perspectives of time dependence and feature dependence. Finally, the data were reconstructed by the decoder, and the reconstruction error was calculated as the anomaly score to detect anomalies of the machine under the actual operating conditions. Experimental results on SWaT, SMAP, MSL three public datasets and Electric Vehicle power battery (EV) dataset show that compared to the suboptimal model, the F1-scores of the proposed model increase by 6.5%, 1.8%, 0.9%, and 7.1% respectively.The above results prove that CT-ED is suitable for anomaly detection under different operating conditions, and balancing the precision and recall of anomaly detection.

Key words: time series, anomaly detection, contrastive learning, multi-head attention, autoencoder

中图分类号: