Research on Low Voltage AC Series Fault Arc Identification Based on Spiking Neural Network

doi:10.16628/j.cnki.2095-8188.2025.04.002

Abstract

Abstract:

In order to solve the problems of misjudgment, missing judgment and slow identification speed when the series fault arc is suddenly connected to the circuit, a low voltage AC series fault arc identification method based on spiking neural network is proposed. The current arc signal is preprocessed by Markov transition field, and the fault arc identification model is constructed by introducing Poisson coding and spiking neurons. The results show that the proposed method can effectively encode the fault arc image, and the accuracy of fault detection is higher than that of the traditional neural network, which can reach more than 98%.

Key words: fault arc, spiking coding, neuron model, spiking neural network

CLC Number:

TM501.2

CHEN Zukuai, LI Jing, SHEN Yuwen, JIANG Song. Research on Low Voltage AC Series Fault Arc Identification Based on Spiking Neural Network[J]. LOW VOLTAGE APPARATUS, 2025, 0(4): 7-14.

Figures/Tables 13

References 19

[1]	聂云良, 赵宇, 李瑞. 基于改进Mayr模型与Comsol的串联故障电弧研究与火灾风险评估[J]. 电器与能效管理技术, 2022(11):74-80.
[2]	KIM J C, NEACSU D O, LEHMAN B, et al. Series AC arc fault detection using only voltage waveforms[C]// 2019 IEEE Applied Power Electronics Conference and Exposition, 2019:2385-2389.
[3]	杨欣怡. 电弧故障断路器的研究[D]. 沈阳: 沈阳工业大学, 2018.
[4]	汪倩, 陈思磊, 孟羽, 等. 低压直流系统故障电弧检测技术研究综述[J]. 高压电器, 2023, 59(5):93-103.
[5]	郭敏, 吴宁, 郭小璇, 等. 基于多特征融合和SVM的串联电弧故障检测方法[J]. 电力电容器与无功补偿, 2024, 45(3):13
[6]	高小庆, 陈晓军, 鲍光海. 考虑故障电弧电流高频特征的阻抗电弧模型[J]. 电器与能效管理技术, 2021(3):6-12.
[7]	李瑞新, 刘树鑫, 曹云东. 基于振动信号的交流接触器运行状态识别[J]. 电器与能效管理技术, 2022(6):39-44.
[8]	康乐, 田云博, 杨騉, 等. 串联故障电弧电流时频域特性的仿真与实验研究[J]. 高压电器, 2023, 59(2):104-112,119.
[9]	陈思磊, 王源丰, 孟羽, 等. 包络微分算子增强直流故障电弧特征及其半监督式辨识方法[J]. 高压电器, 2023, 59(7):145-155.
[10]	林靖怡, 武建文, 李奎. 基于主成分分析和决策树的故障电弧识别方法[J]. 电器与能效管理技术, 2023(4):1-7.
[11]	单潇洁, 郑昕. 基于小波分析的低压电弧特性识别模型的研究[J]. 电器与能效管理技术, 2021(6):7-14.
[12]	刘官耕, 杜松怀, 苏娟, 等. 低压电弧故障防护技术研究与发展趋势[J]. 电网技术, 2017, 41(1):305-313.
[13]	陈照, 王尧, 牛峰, 等. 基于神经网络和D-S证据理论的故障电弧检测[J]. 高压电器, 2019, 55(3):29-36.
[14]	ZHENG N, MAZUMDER P. Online supervised learning for hardware-based multilayer spiking neural networks through the modulation of weight-dependent spike-timing-dependent-plasticity[J]. IEEE Transactions on Neural Networks and Learning Systems, 2017, 29(9):4287-4302.
[15]	WANGZ, OATES T. Encoding time series as images for visual inspection and classification using tiled convolutional neural networks[C]// Workshops at the Twenty-ninth Aaai Conference on Artificial Intelligence, 2015:1-7.
[16]	PETER U D, MATTHEW C, MASAMI T, et al. Unsupervised learning of digit recognition using spike-timing-dependent plasticity[J]. Frontiers in Computational Neuroscience, 2015(9):99.
[17]	FENG J. Behaviors of spike output jitter in the integrate-and-fire model[J]. Physical Review Letters, 1997, 79(22):4505.
[18]	IZHIKEVICH E M. Simple model of spiking neurons[J]. IEEE Transactions on Neural Networks, 2003, 14(6):1569-1572. doi: 10.1109/TNN.2003.820440 pmid: 18244602
[19]	杨志春, 裴冀南, 牛健人. 脉冲神经网络的稳定性及其脉冲控制[J]. 电子科技大学学报, 2007(增刊1):367-369.

负载编号	使用负载	试验环境
1	真空吸尘器	7.6 A/230 V
2	开关电源	4.0 A/230 V
3	电容起动电机	7.5 A/230 V
4	荧光灯	5.1 A/230 V
5	卤素灯	5.6 A/230 V
6	电动手持工具	2.9 A/230 V

神经元模型	优点	缺点	适用的场景
SRM模型	计算能力强	对噪声敏感	大型网络和抗噪要求不高的仿真
Izhikevich模型	结合生物合理性高和计算效率高的优点,性价比高	仍有较高的模型复杂度	大型网络和模型简单的仿真
LIF模型	计算模型简单	脉冲发射方式单一	大型网络和模型复杂度不高的仿真

模型	准确率/%	损失
CNN	91.80	0.413
SRM	97.27	0.167
Izhikevich	98.33	0.022
LIF	98.74	0.025

算法	分断时间/s
SNN	0.100
CNN	0.136