基于CNN-LSTM模型与I-V特性的光伏组串故障识别研究

打开文本图片集

中图分类号：TM615

文献标志码：A

Research on Fault Recognition of Photovoltaic String Driven by CNN-LSTM Modeland I-V Characteristics

Qiu Yiwei1，Zhong Haiwei² （1.Zhejiang Changzheng Vocational& Technical College， Hangzhou 31oo23， China; 2.ZhejiangUniversityof Science & Technology，Hangzhou 31oo23， China）

Abstract：Withthecontiuous expansioofphotovoltaicpowersationscale，hgherrequirementsareputfarwardbypracticalsenariosfor theacuracyandtimelinessinmodulefaultdiagnosis.Toaddressteshortcomingsinaccuracyandeficiencyofexistingfaltetection methods，this paper designs a novel intelligent recognition algorithm based on I -V characteristic curves，convolutional neural networks （CNN），and long short-term memory （LSTM） networks. Firstly， the I ⋅V characteristic curves are analyzed，and the data are cleaned and corrected. Secondly， the I- V data are reconstructed through linear interpolation method to enhance dataset quality.Finaly，a deep neural network ntegatigCandLSTissigedtohancefatureetractioncaabilitis，terebyimprovingfultecogitionface. Experimentalesultsdmostratetateproposdmetodacheveshighcogitioperfomanceinidentifingsventyesofpotovoaic string faults， showcasing good practical application value.

KeyWords：CNN;LSTM;I-Vcharacteristics;fault diagnosis

0引言

光伏发电已成为推动能源绿色转型、构建新型电力系统的关键，而光伏组串则是光伏发电系统的核心组件，但其常年暴露在外，面临多变环境影响以及自身材料老化等问题，导致故障频发，直接影响经济效益，严重时甚至引发火灾。（剩余4908字）