电工技术学报2024，Vol.39Issue(20)：6526-6543,18.DOI:10.19595/j.cnki.1000-6753.tces.231327
光伏阵列故障诊断技术综述
A Review of Photovoltaic Array Fault Diagnosis Technology
摘要
Abstract
A photovoltaic array is an integral part of a photovoltaic power generation system.Due to its long exposure to the external environment,it is prone to different types and degrees of faults.The fault diagnosis of the photovoltaic array is significant to the safe,reliable,and economic operation of photovoltaic power generation systems.Therefore,this paper summarizes the frontier diagnosis technology of photovoltaic array faults. Firstly,the common fault types and causes of photovoltaic arrays are introduced,including the faults between photovoltaic modules and single-module faults.Furthermore,the operating mechanism of the photovoltaic module is described by a single-diode equivalent circuit model.Based on the model,the voltage and current under different fault types change significantly.The variation trend of voltage and current curves divides these faults into non-mismatch and mismatch faults.For non-mismatched faults,the I-V curve does not change significantly,such as PID faults and diode short circuits.For mismatched faults,the I-V curve has a slope or step-shape trend,such as shadows and hot spots. Secondly,based on the mechanism characteristics and external representation of photovoltaic array faults,this paper summarizes photovoltaic array fault diagnosis methods into two categories:vision and imaging diagnosis methods and electrical characteristic parameter diagnosis methods.Related methods are studied to achieve fault diagnosis by obtaining image and data information through auxiliary detection equipment.Vision and imaging diagnosis methods consist of infrared thermal imaging,visible imaging,electroluminescence,photoluminescence,and machine-learning image diagnosis methods.Electrical characteristic parameter diagnosis methods include I-V curve,physical detection,power loss analysis,voltage-current,and machine learning classification methods.This paper summarizes and analyzes the existing domestic and foreign research.Specifically,the electroluminescence and photoluminescence imaging methods are used to diagnose crack and micro-crack faults.The infrared thermal imaging method diagnoses hot spot faults. The machine-learning image detection method achieves object detection and semantic segmentation of the fault points through the acquired image information.In addition,the I-V curve method is used to analyze and judge the characteristics and trends of the curve,such as shadows and hot spots.The physical detection method diagnoses different modules in photovoltaic arrays,such as grounding and short-circuit faults.Using the equivalent circuit model,the machine-learning method diagnoses open-circuit,line-line,and short-circuit faults based on massive data set. Finally,it is pointed out that the establishment of photovoltaic module fault models,the improvement of fault feature recognition methods,the application of artificial intelligence technology,the health management of photovoltaic array operation state,and the optimization of fault diagnosis strategies need to be further studied in the future.It aims to improve the fault diagnosis capabilities of photovoltaic arrays and provide a technical basis for intelligent operation and maintenance of photovoltaic systems.关键词
光伏阵列故障类型/故障特征/视觉与成像诊断方法/电气特征参数诊断方法Key words
Photovoltaic array fault types/fault characteristics/visual and imaging diagnostic methods/electrical characteristic parameter diagnostic methods分类
信息技术与安全科学引用本文复制引用
王小宇,刘波,孙凯,赵健,陈雷..光伏阵列故障诊断技术综述[J].电工技术学报,2024,39(20):6526-6543,18.基金项目
江苏省碳达峰碳中和科技创新专项资助项目(BE2022038). （BE2022038）
