基于红外热像的车站混凝土结构损伤智能检测方法OA

As a typical concrete building,most concrete structures in stations have certain internal defects due to limitations in craftsmanship and material ratios.Therefore,in order to ensure the safety of station personnel,an intelligent detection method for internal damage in concrete based on infrared thermal imaging and mask cyclic convolutional neural network has been proposed.The experimental results show that the highest detection accuracy of this method for different types of defects is 99.7%,and the lowest is 93.3%.The highest intersection ratio is 98.4%,and the lowest is 91.5%.In defect size quantification,the ratio of predicted size to actual size approaches 1,with a maximum relative error of 1.35%and a minimum error of 0.04%,and an average relative error of 0.7%.In practical applications,the average accuracy,average IoU,and average recall of the intelligent detection model are 92.1%,90.2%,and 84.4%,respectively,which have decreased compared to testing,but still remain above a good level.The above results indicate that the intelligent detection model based on infrared thermal imaging and mask cyclic convolutional neural network can effectively achieve accurate detection of internal damage in concrete,effectively ensuring the performance of station concrete structures.