摘要
光伏阵列随着运行时间的增长,阵列内数量众多的连接线缆、连接头容易产生破损或连接失效等问题,引发直流电弧故障,严重影响系统的安全运行,因此需要采用合适的检测方法进行故障诊断,以及时发现电弧故障。直流电弧故障的检测方法大致可以分为基于物理特性和时频特性两类。前者成本高,难度大,不适合大型光伏系统;后者随着近几年人工智能技术的兴起,大多数是提取直流电弧故障的时频域特征值形成数据集,运用神经网络或智能算法对其进行识别、训练、归纳等,达到检测目的,目前实际应用的检测方法侧重于后者。选用基于时频域特性的集合经验模态分解和支持向量机结合方法进行检测,在MATLAB/Simulink仿真平台搭建光伏阵列模型和直流电弧故障仿真模型,模拟光伏阵列不同位置的串、并联电弧故障,对电流信号进行采集、分析与处理。实验结果表明,支持向量机模型能够较好地对光伏阵列直流电弧故障进行识别和检测,有效区分光伏阵列正常工作状态与故障工作状态。
With the growth of the operation time of the photovoltaic array,a large number of connecting cables and connectors in the array are prone to damage and connection failure,which causes DC arc faults and seriously endangers the safe operation of the photovoltaic system.It is necessary to adopt appropriate detection methods for fault diagnosis in order to find arc faults as soon as possible.The detection methods of DC arc faults can be roughly divided into two categories,based on physical characteristics and time-frequency characteristics.The former is costly,difficult,and not suitable for large-scale photovoltaic systems.With the rise of artificial intelligence in recent years,most of the latter is to extract the time-frequency domain eigenvalues of DC arc faults to form a dataset,and use neural networks or intelligent algorithms to identify,train,and generalize them so as to achieve the purpose of detection,which is focus on practical applications at present.The ensemble empirical mode decomposition and support vector machine combination method based on the characteristics of the time-frequency domain are selected for detection,and the photovoltaic array model and DC arc fault simulation model are built on the MATLAB/Simulink simulation platform to simulate the series and parallel arc faults at different positions of the photovoltaic array,and the current signals are collected,analyzed and processed.Experimental results show that the SVM model can better identify and detect the DC arc faults of photovoltaic arrays,and effectively distinguish the normal working state and fault working state of photovoltaic arrays.
出处
《电动工具》
2024年第3期13-17,19,共6页
Electric Tool
关键词
直流
电弧
故障检测
时频域特性
集合经验模态分解
支持向量机
仿真模型
DC
arc
fault detection
time-frequency domain characteristics
Ensemble Empirical Mode Decomposition(EEMD)
Support Vector Machine(SVM)
simulation model
