Fault Detection in Power Distribution Systems Based on Gated Recurrent Attention Network

doi:10.16183/j.cnki.jsjtu.2022.091

Abstract

Abstract:

To improve fault identification accuracy in power distribution systems, a model named gated recurrent attention network is proposed. First, a higher weight is put on the key cycles of fault phase based on the attention mechanism, making the model focus more on these key messages by weight assignment. Then, the gated recurrent network is adopted, which controls the memory transmission with gate signal and constructs the relationship between input waveform and probability of events at different stages to process the waveform sequence, thereby improving recognition accuracy. Experiments based on both simulation and field data show that the proposed method, under the small-sample-learning condition, is much better than other commonly-used classification models, such as support vector machine, gradient boosting decision tree, and convolutional neural network, providing new insights into fault identification technology in power distribution systems.

Key words: power distribution systems, faults identification, attention mechanism, gated recurrent units (GRU)

CLC Number:

TM726

CHEN Haolan, JIN Bingying, LIU Yadong, QIAN Qinglin, WANG Peng, CHEN Yanxia, YU Xijuan, YAN Yingjie. Fault Detection in Power Distribution Systems Based on Gated Recurrent Attention Network[J]. Journal of Shanghai Jiao Tong University, 2024, 58(3): 295-303.

Figures/Tables 8

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

Tab.1

Fig.6

Tab.2

References 27

[1]	张丽虹, 常成, 熊炜, 等. 基于智能控制终端的主动配电网故障处理方法[J]. 电力科学与技术学报, 2020(2): 22-29.
	ZHANG Lihong, CHANG Cheng, XIONG Wei, et al. Research on fault processing of active power distribution network based on intelligent control terminal[J]. Journal of Electric Power Science & Technology, 2020(2): 22-29.
[2]	陈国炎, 李俊均, 陈颖, 等. 数据驱动的配电开关设备交互式诊断平台[J]. 电工电能新技术, 2019, 38(3): 10-17. doi: 10.12067/ATEEE1810004
	CHEN Guoyan, LI Junjun, CHEN Ying, et al. Data-driven and interactive fault diagnosis of distribution switches[J]. Advanced Technology of Electrical Engineering & Energy, 2019, 38(3): 10-17.
[3]	江秀臣, 刘亚东, 傅晓飞, 等. 输配电设备泛在电力物联网建设思路与发展趋势[J]. 高电压技术, 2019, 45(5): 1345-1351.
	JIANG Xiuchen, LIU Yadong, FU Xiaofei, et al. Construction ideas and development trends of transmission and distribution equipment of the ubiquitous power internet of things[J]. High Voltage Engineering, 2019, 45(5): 1345-1351.
[4]	熊思衡, 刘亚东, 方健, 等. 配电线路早期故障辨识方法[J]. 高电压技术, 2020, 46(11): 259-265.
	XIONG Siheng, LIU Yadong, FANG Jian, et al. Detection method of incipient faults of power distribution lines[J]. High Voltage Engineering, 2020, 46(11): 259-265.
[5]	SIDHU T S, XU Z. Detection of incipient faults in distribution under-ground cables[J]. IEEE Transactions on Power Delivery, 2010, 25(3): 1363-1371. doi: 10.1109/TPWRD.2010.2041373 URL
[6]	李泽文, 刘基典, 席燕辉, 等. 基于暂态波形相关性的配电网故障定位方法[J]. 电力系统自动化, 2020(21): 72-79.
	LI Zewen, LIU Jidian, XI Yanhui, et al. Fault location method for distribution network based on transient waveform correlation[J]. Automation of Electric Power Systems. 2020, 44(21): 72-79.
[7]	IZADI M, MOHSENIAN-RAD H. Synchronous waveform measurements to locate transient events and incipient faults in power distribution networks[J]. IEEE Transactions on Smart Grid, 2021, 12(5): 4295-4307. doi: 10.1109/TSG.2021.3081017 URL
[8]	ZHANG W, JING Y, XIAO X. Model-based general arcing fault detection in medium-voltage distribution lines[J]. IEEE Transactions on Power Delivery, 2016, 31(5): 2231-2241. doi: 10.1109/TPWRD.2016.2518738 URL
[9]	蒋碧莺, 荣建, 张军. Logistic分类算法下的配电网故障识别技术研究[J]. 电工技术, 2018, 486(24): 70-71.
	JIANG Biying, RONG Jian, ZHANG Jun. Research on fault identification technology of distribution network based on Logistic classification[J]. Electric Engineering, 2018, 486(24): 70-71.
[10]	郭谋发, 游林旭, 洪翠, 等. 基于LCD-Hilbert谱奇异值和多级支持向量机的配电网故障识别方法[J]. 高电压技术, 2017(4): 1239-1247.
	GUO Moufa, YOU Linxu, HONG Cui, et al. Identification method of distribution network faults based on singular value of LCD-Hilbert spectrums and multilevel SVM[J]. High Voltage Engineering, 2017(4): 1239-1247.
[11]	赵智, 王艳松, 鲍兵, 等. 基于小波神经网络的配电网故障类型识别[J]. 电力系统及其自动化学报, 2007, 19(6): 93-96.
	ZHAO Zhi, WANG Yansong, BAO Bing, et al. Fault type identification in distribution network based on wavelet neural network[J]. Proceedings of The CSU-EPSA, 2007, 19(6): 93-96.
[12]	武光利, 郭振洲, 李雷霆, 等. 融合FCN和LSTM的视频异常事件检测[J]. 上海交通大学学报, 2021, 55(5): 607-614. doi: 10.16183/j.cnki.jsjtu.2020.120
	WU Guangli, GUO Zhenzhou, LI Leiting, et al. Video abnormal detection combining FCN with LSTM[J]. Journal of Shanghai Jiao Tong University, 2021, 55(5): 607-614.
[13]	刘秀丽, 徐小力. 基于特征金字塔卷积循环神经网络的故障诊断方法[J]. 上海交通大学学报, 2022, 56(2): 182-190. doi: 10.16183/j.cnki.jsjtu.2021.001
	LIU Xiuli, XU Xiaoli. A fault diagnosis method based on feature pyramid CRNN network[J]. Journal of Shanghai Jiao Tong University, 2022, 56(2): 182-190.
[14]	张松林, 马栋梁, 王德禹. 基于长短期记忆神经网络的板裂纹损伤检测方法[J]. 上海交通大学学报, 2021, 55(5): 527-535. doi: 10.16183/j.cnki.jsjtu.2020.095
	ZHANG Songlin, MA Dongliang, WANG Deyu. Method for plate crack damage detection based on long short-term memory neural network[J]. Journal of Shanghai Jiaotong University, 2021, 55(5): 527-535. doi: 10.16183/j.cnki.jsjtu.2020.095
[15]	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. New York, USA: ACM, 2017: 6000-6010.
[16]	XIONG S H, LIU Y D, FANG J, et al. Incipient fault identification in power distribution systems via human-level concept learning[J]. IEEE Transactions on Smart Grid, 2020, 11(6): 5239-5248. doi: 10.1109/TSG.5165411 URL
[17]	GRAVES A, LIWICKI M, FERNÁNDEZ S, et al. A novel connectionist system for unconstrained handwriting recognition[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2009, 31(5): 855-868.
[18]	HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780. doi: 10.1162/neco.1997.9.8.1735 pmid: 9377276
[19]	CHO K, VAN MERRIENBOER B, GULCEHRE C, et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation[EB/OL]. (2014-09-03) [2022-03-30]. https://arxiv.org/abs/1406.1078.
[20]	KULKARNI S, ALLEN A J, CHOPRA S, et al. Waveform characteristics of underground cable failures[C]// IEEE PES General Meeting. Minneapolis, USA. IEEE, 2010: 1-8.
[21]	JANNATI M, VAHIDI B, HOSSEINIAN S H. Incipient faults monitoring in underground medium voltage cables of distribution systems based on a two-step strategy[J]. IEEE Transactions on Power Delivery, 2019, 34(4): 1647-1655. doi: 10.1109/TPWRD.61 URL
[22]	谢潇磊, 刘亚东, 孙鹏, 等. 新型配电网线路PMU装置的研制[J]. 电力系统自动化, 2016, 40(12): 15-20.
	XIE Xiaolei, LIU Yadong, SUN Peng, et al. Development of novel PMU device for distribution network lines[J]. Automation of Electric Power Systems, 2016, 40(12): 15-20.
[23]	SASAKI Y. The truth of the f-measure[D]. Manchester: University of Manchester, 2007.
[24]	KE G L, MENG Q, FINLEY T, et al. LightGBM: A highly efficient gradient boosting decision tree[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. New York, USA: ACM, 2017: 3149-3157.
[25]	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6): 84-90. doi: 10.1145/3065386 URL
[26]	CHANG C C, LIN C J. LIBSVM: A library for support vector machines[J]. ACM Transactions on Intelligent Systems & Technology, 2011, 2(3): 1-27.
[27]	CHEN T Q, HE T. Xgboost: Extreme gradient boosting[EB/OL]. (2023-03-31)[2023-08-23]. http://mysql.orst.edu/pub/cran/web/packages/xgboost/vignettes/xgboost.pdf.

模型	仿真类别						平均值
模型	1		2		3		平均值
GRU	0.902		0.941		0.887		0.910
SVM	0.810		0.841		0.815		0.822
GBDT	0.832		0.869		0.844		0.848
CNN	0.871		0.890		0.867		0.876
模型	实验类别						平均值
模型	1	2		3		4	平均值
GRU	0.862	0.871		0.854		0.831	0.855
SVM	0.720	0.704		0.731		0.719	0.719
GBDT	0.762	0.774		0.785		0.747	0.767
CNN	0.790	0.807		0.820		0.787	0.801

模型	仿真类别						平均值
模型	1		2		3		平均值
GRAN	0.940		0.977		0.930		0.949
GRU	0.902		0.941		0.887		0.910
模型	实验类别						平均值
模型	1	2		3		4	平均值
GRAN	0.921	0.940		0.901		0.897	0.915
GRU	0.862	0.871		0.854		0.831	0.855