| 引用本文: | 蔡梓文,赵 云,陆煜锌,等.基于变分自编码器的多源数据融合窃电检测方法[J].电力系统保护与控制,2025,53(4):176-187.[点击复制] |
| CAI Ziwen,ZHAO Yun,LU Yuxin,et al.A multi-source data fusion electricity theft detection method based on a variational autoencoder[J].Power System Protection and Control,2025,53(4):176-187[点击复制] |
|
| 摘要: |
| 针对当前窃电检测仅使用单一用电负荷难以捕捉复杂窃电特征，导致窃电检测发生误判，存在误检率高和准确率低下等问题，提出一种融合用电负荷、环境温度、时间以及对应台区相位线损的新型窃电检测方法。首先构建多维度特征提取变分自编码器(variational autoencoder for multi-dimensional feature extraction, MF-VAE)来提取用户用电行为的多维度特征。然后，基于注意力时序卷积网络(attention temporal convolutional networks, ATCN)建立判别模型，再通过膨胀卷积和因果卷积获取多维度窃电行为特征的时序关系。同时，引入卷积注意力模块分配各维度特征的注意力权重，以提高模型的表现和泛化能力。最后采用Softmax分类器完成对多源数据中潜在窃电行为的准确识别。实验结果表明，用该方法提取的窃电行为特征更加丰富和多元化，能够有效降低窃电检测误检率并提高窃电行为判别准确率。 |
| 关键词: 窃电行为判别 多源数据融合 改进时域卷积网络 变分自编码器 注意力机制 |
| DOI：10.19783/j.cnki.pspc.240259 |
| 投稿时间：2024-03-06修订日期：2024-10-02 |
| 基金项目:广东省电网智能量测与先进计量企业重点实验室开放基金项目资助(GPKLIMAMPG-2022-KF-04) |
|
| A multi-source data fusion electricity theft detection method based on a variational autoencoder |
| CAI Ziwen1,ZHAO Yun1,LU Yuxin1,GU Lianqiang1,CHEN Kang2,GAO Yunpeng2 |
| (1. Electric Power Research Institute, CSG, Guangzhou 510530, China; 2. College of Electrical and Information
Engineering, Hunan University, Changsha 410082, China) |
| Abstract: |
| Traditional electricity theft detection methods relying solely on single power consumption loads and struggle to capture complex theft patterns, resulting in high false detection rate and low accuracy. Thus this paper proposes a new electric theft detection method that integrates electric load, ambient temperature, time and corresponding phase line loss of the station area. First, a multi-dimensional feature extraction variational autoencoder (MF-VAE) is constructed to extract the multi-dimensional characteristics of the user’s electricity consumption behavior. Then, an attention temporal convolutional network (ATCN) is employed to establish a discriminator model, where expansion convolution and causal convolution are used to obtain the temporal relationships of multi-dimensional electric theft features. Meanwhile, a convolutional attention module is introduced to assign the attention weight of each dimension feature to improve the performance and generalizability of the model. Finally, a Softmax classifier is used to accurately identify potential power theft behaviors in multi-source data. Experimental results show that the characteristics of electric theft behaviors extracted by the proposed method are more abundant and diversified, which can effectively reduce false detection rates and improve the accuracy of electric theft behavior identification. |
| Key words: electricity theft behavior identification multi-source data fusion improved time-domain convolutional network variational autoencoder attention mechanism |
