摘要
针对目前光伏组件热斑检测方式较为众多繁杂的问题,总结并系统剖析现存的国内外各种光伏组件传统及智能化的热斑检测方式,将深度学习算法运用于此的最新进展,并着重研究多位研究者对于将各种神经网络、注意力机制以及目标检测模型运用于热斑红外图像的检测上,如多尺度卷积神经网络、基于SpotFPN多尺度特征学习模块等。之后针对光伏组件热斑检测的智能化诊断技术进行实验对比,最后对当前问题展开探讨以及对未来该技术进展的展望。
In view of the many and complicated problems in the current hot spot detection methods of photovoltaic modules,summarizes and systematically analyzes the existing traditional and intelligent hot spot detection methods for PV modules at home and abroad.And the latest progress of applying deep learning algorithms to them.Focusing on the application of various neural networks,attention mechanisms and target detection models to the detection of hot spot infrared images,such as multi-scale CNN,SpotFPN-based multi-scale feature learning modules etc.After that,experiments are conducted to compare the intelligent diagnostic techniques for hot spot detection of PV modules.Finally,the current issues are discussed and the future progress of the technology is expected.
作者
王道累
姚从荣
李超
王卫军
秦宝星
朱瑞
Wang Daolei;Yao Congrong;Li Chao;Wang Weijun;Qin Baoxing;Zhu Rui(College of Energy and Mechanical Engineering,Shanghai University of Electric Power,Shanghai 201306,China;21ST Research Institute of China Electronic Technology Group Corporation,Shanghai 200233,China;Gaussian Robotics,Shanghai 201210,China)
出处
《太阳能学报》
EI
CAS
CSCD
北大核心
2024年第5期527-536,共10页
Acta Energiae Solaris Sinica
基金
国家自然科学基金(12172210,61502297)。
关键词
光伏组件
太阳能发电
红外图像
深度学习
热斑
智能化
PV modules
solar power generation
infrared imaging
deep learning
hot spot
intellectualization
