基于改进双流网络的光伏漏电故障检测算法  

An Algorithm for Detecting Photovoltaic Leakage Faults in Low-Voltage Distribution Substations Based on Two-Stream Network

作　　者：谢小诚 杨文呈 赵彦宏 丁毅 李王宏 杨晨 XIE Xiaocheng;YANG Wencheng;ZHAO Yanhong;DING Yi;LI Wanghong;YANG Chen(Honghe Luxi Power Supply Bureau of Yunnan Power Grid Co.,Ltd.,Honghe 661000,Yunnan,China;Lincang Power Supply Bureau,Lincang 677000,Yunnan,China;Honghe Power Supply Bureau,Honghe 661000,Yunnan,China)

机构地区：[1]云南电网有限责任公司红河泸西供电局,云南红河661000 [2]临沧供电局,云南临沧677000 [3]红河供电局,云南红河661000

出　　处：《电网与清洁能源》2025年第1期146-154,共9页Power System and Clean Energy

基　　金：湖南省自然科学基金项目(2023JJ50341);云南电网有限责任公司临沧供电局科技项目(0510002023030301DL00013)。

摘　　要：针对含光伏电源的低压配电台区中剩余电流受多种因素影响,使正常泄漏电流波动,造成漏电流故障难以被准确识别的问题,提出基于改进双流神经网络的含光伏电源的低压配电台区漏电故障检测算法。首先,通过改进的卷积神经网络结构提取漏电故障波形的空间特征;然后,基于长短期记忆神经网络提取漏电故障波形的时序特征,通过CBAM(convolutional block attention module,CBAM)注意力机制从空间角度提取关键信息,并增强每个通道的特征表达以提取重要特性,从而实现漏电故障精确识别;最后,通过仿真模型进行仿真分析与验证。实验结果表明,该文提出方法可实现漏电故障的高精度检测,且与常用方法相比,所提方法的故障检测准确率和稳定性更高、抗干扰能力更强。To address the issue that the residual current in low-voltage distribution areas with photovoltaic(PV)power sources is influenced by various factors,causing fluctuations in the normal leakage current and making it difficult to accurately identify leakage current faults,an improved dual-stream neural network-based leakage fault detection algorithm for low-voltage distribution areas with PV power sources is proposed in this paper.First,the spatial features of the leakage fault waveform are extracted through the improved convolutional neural network structure;second,the temporal features of the leakage fault waveform are extracted based on the long short-term memory neural network,and the CBAM attention mechanism is used to extract key information from a spatial perspective and enhance the feature expression of each channel to extract important characteristics,thereby achieving precise identification of leakage faults;finally,simulation models are used for simulation analysis and verification.The experimental results show that the proposed method in this paper can achieve highprecision detection of leakage faults,and compared with common methods,the proposed method has higher fault detection accuracy and stability,and stronger anti-interference capabilities.

关 键 词：双流神经网络 注意力机制 漏电故障 剩余电流 光伏电源 

分 类 号：TM72[电气工程—电力系统及自动化]