针对无人机在光伏组件巡检任务中红外故障图像识别准确率低、检测速度慢的问题,提出一种特征增强的YOLO v5s故障检测算法。首先对损失函数进行优化,将原有的回归损失计算方法由GIOU(generalized intersection over union)改为功能更加...针对无人机在光伏组件巡检任务中红外故障图像识别准确率低、检测速度慢的问题,提出一种特征增强的YOLO v5s故障检测算法。首先对损失函数进行优化,将原有的回归损失计算方法由GIOU(generalized intersection over union)改为功能更加强大的EIOU(efficient intersection over union)损失函数,并自适应调节置信度损失平衡系数,提升模型训练效果;随后,在每个检测层前分别添加InRe特征增强模块,通过丰富特征表达增强目标特征提取能力。最后,用创建的红外光伏数据集进行对比验证。实验结果表明:本文方法均值平均精度(mean average precision,mAP)为92.76%,检测速度(frame per second,FPS)达到42.37 FPS,其中热斑、组件脱落两种故障类型平均精度分别为94.85%、90.67%,完全能够满足无人机自动巡检的需求。展开更多
V5S8 is an ideal candidate to explore the magnetism at the two-dimensional(2D)limit.A recent experiment has shown that the V5S8 thin films exhibit an antiferromagnetic(AFM)to ferromagnetic(FM)phase transition with red...V5S8 is an ideal candidate to explore the magnetism at the two-dimensional(2D)limit.A recent experiment has shown that the V5S8 thin films exhibit an antiferromagnetic(AFM)to ferromagnetic(FM)phase transition with reducing thickness.Here,for the first time,using density functional theory calculations,we report the antiferromagnetic order of bulk V5S8,which is consistent with the previous experiments.The specific antiferromagnetic order is reproduced when Ueff=2 eV is applied on the intercalated vanadium atoms within LDA.We find that the origin of the magnetic ordering is from superexchange interaction.We also investigate the thickness-dependent magnetic order in V5S8 thin films.It is found that there is an antiferromagnetic to ferromagnetic phase transition when V5S8 is thinned down to 2.2 nm.The main magnetic moments of the antiferromagnetic and ferromagnetic states of the thin films are located on the interlayered vanadium atoms,which is the same as that in the bulk.Meanwhile,the strain in the thin films also influences the AFM-FM phase transition.Our results not only reveal the magnetic order and origin in bulk V5S8 and thin films,but also provide a set of parameters which can be used in future calculations.展开更多
文摘针对无人机在光伏组件巡检任务中红外故障图像识别准确率低、检测速度慢的问题,提出一种特征增强的YOLO v5s故障检测算法。首先对损失函数进行优化,将原有的回归损失计算方法由GIOU(generalized intersection over union)改为功能更加强大的EIOU(efficient intersection over union)损失函数,并自适应调节置信度损失平衡系数,提升模型训练效果;随后,在每个检测层前分别添加InRe特征增强模块,通过丰富特征表达增强目标特征提取能力。最后,用创建的红外光伏数据集进行对比验证。实验结果表明:本文方法均值平均精度(mean average precision,mAP)为92.76%,检测速度(frame per second,FPS)达到42.37 FPS,其中热斑、组件脱落两种故障类型平均精度分别为94.85%、90.67%,完全能够满足无人机自动巡检的需求。
基金Project supported by the National Natural Science Foundation of China(Grant Nos.51922011 and 61888102)the National Key Research&Development Project of China(Grant Nos.2016YFA0202300 and 2019YFA0308500)the Strategic Priority Research Program of Chinese Academy of Sciences(Grant Nos.XDB30000000 and XDB28000000).A portion of the research was performed in CAS Key Laboratory of Vacuum Physics.
文摘V5S8 is an ideal candidate to explore the magnetism at the two-dimensional(2D)limit.A recent experiment has shown that the V5S8 thin films exhibit an antiferromagnetic(AFM)to ferromagnetic(FM)phase transition with reducing thickness.Here,for the first time,using density functional theory calculations,we report the antiferromagnetic order of bulk V5S8,which is consistent with the previous experiments.The specific antiferromagnetic order is reproduced when Ueff=2 eV is applied on the intercalated vanadium atoms within LDA.We find that the origin of the magnetic ordering is from superexchange interaction.We also investigate the thickness-dependent magnetic order in V5S8 thin films.It is found that there is an antiferromagnetic to ferromagnetic phase transition when V5S8 is thinned down to 2.2 nm.The main magnetic moments of the antiferromagnetic and ferromagnetic states of the thin films are located on the interlayered vanadium atoms,which is the same as that in the bulk.Meanwhile,the strain in the thin films also influences the AFM-FM phase transition.Our results not only reveal the magnetic order and origin in bulk V5S8 and thin films,but also provide a set of parameters which can be used in future calculations.
