针对无人机输电线路巡检图像的复杂背景目标检测失准、故障小目标难以被准确检测的问题,提出一种基于深度展开超分网络(deep unfolding super-resolution network,USRNet)与改进YOLOv5x算法的输电线路绝缘子故障检测方法。首先,使用USR...针对无人机输电线路巡检图像的复杂背景目标检测失准、故障小目标难以被准确检测的问题,提出一种基于深度展开超分网络(deep unfolding super-resolution network,USRNet)与改进YOLOv5x算法的输电线路绝缘子故障检测方法。首先,使用USRNet对原始图像进行超分辨率重建,以降低复杂背景干扰实现测试数据集优化;然后,以YOLOv5x检测模型为基础,利用K-means++对标记框进行聚类,生成匹配输电线路故障目标尺寸的锚框;同时,通过更改多尺度特征融合模块结构,在预测端引入一个包含更大特征图的检测头以检测故障小目标;最后,使用有效交并比损失(efficient intersection over union loss, EIOU_Loss)函数优化模型整体性能,并设置对比实验对所提方法进行验证。结果表明,所提方法的均值平均精度(mean average precision, mAP)值达到98.8%,可使输电线路故障检测精度提高到95.4%,从而具有更好的复杂背景目标以及小目标检测性能。展开更多
According to the Fick＇s second law of diffusion, six analytical solutions of chloride profile in concrete were studied and discussed with regard to different boundary and initial conditions. In those analytical solut...According to the Fick＇s second law of diffusion, six analytical solutions of chloride profile in concrete were studied and discussed with regard to different boundary and initial conditions. In those analytical solutions, the most prevailing error-function solution which is based on semi-infinite assumption is the simple one, but may under-estimate the chloride content in concrete and over-rate the life time prediction of concrete structures. The experimental results show that compared with other solutions, the chloride content in concrete predicted by error-function model is the minimum, and the calculation difference produced by different analytical models should not be ignored. The influence of models on chloride content prediction is more than other environment and material coefficients in some time. In order to get a more realistic prediction model, modification to error-function model is suggested based on analysis and calculation examples concerning the boundary and edge effect.展开更多
文摘针对无人机输电线路巡检图像的复杂背景目标检测失准、故障小目标难以被准确检测的问题,提出一种基于深度展开超分网络(deep unfolding super-resolution network,USRNet)与改进YOLOv5x算法的输电线路绝缘子故障检测方法。首先,使用USRNet对原始图像进行超分辨率重建,以降低复杂背景干扰实现测试数据集优化;然后,以YOLOv5x检测模型为基础,利用K-means++对标记框进行聚类,生成匹配输电线路故障目标尺寸的锚框;同时,通过更改多尺度特征融合模块结构,在预测端引入一个包含更大特征图的检测头以检测故障小目标;最后,使用有效交并比损失(efficient intersection over union loss, EIOU_Loss)函数优化模型整体性能,并设置对比实验对所提方法进行验证。结果表明,所提方法的均值平均精度(mean average precision, mAP)值达到98.8%,可使输电线路故障检测精度提高到95.4%,从而具有更好的复杂背景目标以及小目标检测性能。
基金Funded by the National Key Technology R&D Program(No.2011BAG07B04)
文摘According to the Fick＇s second law of diffusion, six analytical solutions of chloride profile in concrete were studied and discussed with regard to different boundary and initial conditions. In those analytical solutions, the most prevailing error-function solution which is based on semi-infinite assumption is the simple one, but may under-estimate the chloride content in concrete and over-rate the life time prediction of concrete structures. The experimental results show that compared with other solutions, the chloride content in concrete predicted by error-function model is the minimum, and the calculation difference produced by different analytical models should not be ignored. The influence of models on chloride content prediction is more than other environment and material coefficients in some time. In order to get a more realistic prediction model, modification to error-function model is suggested based on analysis and calculation examples concerning the boundary and edge effect.
