摘要
针对串联电弧高频数据提取困难的问题,提出基于时间序列生成对抗网络(Time-GAN)的串联电弧电流超分辨率重建方法,旨在对时域采样的电流信号进行超分辨率处理,以获取更高频的电流信号。首先搭建光伏系统下低压电弧故障真型实验平台,用不同采样率采集相关数据;然后分析不同采样率对数据时频域采集的影响;再对低采样率的时序数据进行超分辨率重构,生成高频的电流采集数据;最后与高采样率电流信号对比。实验结果表明,基于Time-GAN的超分辨率算法能有效提升电弧电流的采样频率。
Extracting high-frequency data related to series arc faults presents considerable challenges.To solve this problem,we propose a super-resolution reconstruction method for series arc current based on Time-GAN(Time-Generative Adversarial Network).This method aims to perform super-resolution processing on time-domain sampled current signals to obtain higher frequency time-domain signals,thereby achieving higher frequency domain signals.Firstly,a true-scale experimental platform for low-voltage arc faults in photovoltaic systems was established,and relevant data were collected at different sampling rates.Secondly,we analyzed the impact of different sampling rates on time-frequency domain data collection.Then,we applied the super-resolution algorithm to the low sampling rate time series data to generate high sampling rate time-frequency domain data,and compared it with the original high sampling rate time-frequency domain data.Experimental results demonstrate that the Time-GAN-based time super-resolution algorithm significantly enhances the sampling frequency of current signals.
作者
龙葵
牟炜淇
李秋惠
刘嘉威
陈旭凯
苏盛
Long Kui;Mou Weiqi;Li Qiuhui;Liu Jiawei;Chen Xukai;Su Sheng(Dongguan Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Dongguan 523000,China;State Key Laboratory of Disaster Prevention and Reduction for Power Grid,Changsha University of Science and Technology,Changsha 410014,China)
出处
《太阳能学报》
北大核心
2025年第11期260-271,共12页
Acta Energiae Solaris Sinica
基金
中国南方电网有限责任公司科技项目(031900KC23040012(GDKJXM20230372))。
关键词
串联电弧
电气火灾
信号采样
Time-GAN算法
时频域信号重建
光伏系统
series arc
electrical fire
signal sampling,Time-GAN algorithm
time-frequency domain signal reconstruction
photovoltaic system
作者简介
通信作者:苏盛(1975-),男,博士、教授,主要从事低压用电安全和电力气象灾害方面的研究。eessheng@163.com。
