摘要
针对光伏组件的热斑检测,提出基于改进YOLOv5的光伏组件热斑检测算法。该算法引入空洞卷积构成的Dilated block替换原来的SPP block,有效减小了池化操作带来的热斑信息损失,提高了网络的感受野。同时在多尺度检测过程中加入通道注意力机制,增强了热斑目标区域的重要性,提高了热斑检测性能。并提出由深度可分离卷积构成的轻量化模块,有效减小了模型参数量,提高了检测速度。实验表明,基于改进YOLOv5网络能实现热斑部位的快速准确检测,此方法的AP较原算法提高1.5%达到98.65%,检测速度较原算法提高25%达到31帧/s。
Aiming at the thermal spot detection of photovoltaic modules,an improved YOLOv5 based thermal spot detection algorithm was proposed.In this algorithm,Dilated block composed of cavity convolution is introduced to replace the original SPP block,which effectively reduces the heat spot information loss caused by pooling operation and improves the receptive field of the network.At the same time,the channel attention mechanism is added to the multi-scale detection process,which enhances the importance of the target region of hot spots and improves the detection performance of hot spots.A Lightweight block module composed of deep separable convolution is proposed,which effectively reduces the number of model parameters and improves the detection speed.The experimental results show that the improved YOLOv5 network can achieve fast and accurate detection of hot spots.Compared with the original algorithm,the AP of this method is increased by 1.5%to 98.65%,and the detection speed is increased by 25%to 31 fps.
作者
王道累
肖贝成
姚从荣
赵文彬
朱瑞
Wang Daolei;Xiao Beicheng;Yao Congrong;Zhao Wenbin;Zhu Rui(College of Computer Science and Technology,Shanghai University of Electric Power,Shanghai 201306,China)
出处
《太阳能学报》
EI
CAS
CSCD
北大核心
2024年第8期342-348,共7页
Acta Energiae Solaris Sinica
基金
国家自然科学基金(12172210,61502297)。
关键词
深度学习
目标检测
光伏组件
热斑检测
deep learning
target detection
photovoltaic modules
hot spot detection
