基于仿真和随机森林的光伏阵列故障诊断方法
Photovoltaic Array Fault Diagnosis Method Based on Simulation and Random Forest Algorithms
骆守康 1温克欢 1梁健文 1陈作伟 1谭琨 1张尧1
作者信息
- 1. 南方电网深圳供电局有限公司,深圳 518001
- 折叠
摘要
[目的/意义]基于仿真和机器学习算法的故障诊断理论有助于光伏发电系统故障诊断技术的发展.[方法/过程]本文研究了基于Matlab仿真和随机森林算法结合的光伏故障诊断方法.首先,利用Matlab/Simulink构建光伏发电系统的仿真模型,模拟了光伏正常运行及短路、开路、老化、阴影遮挡等不同故障,获得大量的仿真数据作为样本集;然后,通过对训练样本的特征提取和模型训练,建立随机森林故障诊断模型;最后,利用该故障诊断模型对仿真的光伏进行故障诊断,识别出光伏故障类型.[结果/结论]所提出的光伏故障诊断方法在训练集上诊断准确率达 96.8667%,测试集上准确率达到 90.2977%,为光伏发电系统安全可靠运行提供支持.
Abstract
The fault diagnosis theory based on simulation and machine learning algorithms aids in the development of fault diagnosis technology for photovoltaic power generation systems.This paper studies a photovoltaic fault diagnosis method based on Matlab simulation and the combination of the random forest algorithm.First,a simulation model of the photovoltaic power generation system is constructed using Matlab/Simulink,simulating normal operation as well as various faults such as short circuits,open circuits,aging,and shading,resulting in a large amount of simulation data as the sample set.Then,by extracting features from the training samples and training the model,a random forest fault diagnosis model is established.Finally,this fault diagnosis model is used to diagnose faults in the simulated photovoltaic system,identifying the types of faults.The proposed photovoltaic fault diagnosis method achieves an accuracy of 96.8667%on the training set and 90.2977%on the test set,providing support for the safe and reliable operation of photovoltaic power generation systems.
关键词
光伏系统/故障诊断/Simulink仿真/机器学习Key words
photovoltaic system/fault diagnosis/matlab simulink/machine learning引用本文复制引用
出版年
2024
NETL
NSTL
万方数据