基于自适应特征频段的光伏故障电弧检测装置  

Photovoltaic Fault Arc Detection Device Based on Adaptive Feature Frequency Band

作　　者：王毅[1,2] 刘志强 沈红伟 李梦娇 WANG Yil;LIU Zhi-qiangl;SHEN Hong-wei;LI Meng-jiao(Chongqing University of Posts and Telecommunications,Chongqing 400065,China;不详)

机构地区：[1]重庆邮电大学,通信与信息工程学院,重庆400065 [2]重庆邮电大学,移动通信技术重点实验室,重庆400065 [3]北京智芯微电子科技有限公司,北京100192

出　　处：《电力电子技术》2024年第1期73-76,80,共5页Power Electronics

基　　金：2022年重庆市技术创新与应用发展专项重点项目(CSTB2022TIAD-KPX0040)。

摘　　要：针对直流故障电弧由于其强烈的随机性而难以检测的问题,提出了一种光伏系统直流电弧检测算法。首先用快速傅里叶变换(FFT)提取信号的频域特征,然后采用幅度熵自适应的方法选取特征频段。但特征频段的频谱分辨率较低,频率泄露严重,因此采用CZT来提高特征频段的频谱分辨率,最后利用极限学习机(ELM)神经网络进行故障电弧识别。并且基于该算法,此处设计了一种基于STM32嵌入式平台的光伏系统直流电弧检测装置,该装置不仅能够在实验室场景下实现故障电弧的检测,还能够在具体的光伏现场中实现高精准度的识别,具有重要的现实意义。Aiming at the problem that DC fault arc is difficult to detect due to its strong randomness,a DC arc detection algorithm for photovoltaic system is proposed.Firstly,fast Fourier transform(FFT)is used to extract the frequency domain features of the signal and then the amplitude entropy adaptive method is used to select the.feature frequency band.However,the spectral resolution of the characteristic frequency band is low and the frequency leakage is serious.Therefore,CZT is used to improve the spectral resolution of the characteristic frequency band.Finally,extreme learning machine(ELM)neural network is used to identify the fault arc.And based on this algorithm,a DC arc detection device of photovoltaic system based on STM32 embedded platform is designed.The device can not only realize fault are detection in the laboratory scene,but also realize high-precision identification in the specific photovoltaic field which has important practical significance.

关 键 词：光伏 极限学习机 电弧检测 

分 类 号：TM615[电气工程—电力系统及自动化]