新工科背景下基于神经网络的隧道工程缺陷检测研究

Research on tunnel engineering defect detection based on neural network under the background of new engineering

隧道是综合交通运输体系的重要组成部分,是人民便捷生活的基本保障。为响应新工科建设号召,应对新一轮科技革命和产业变革,提升科学化建设要求,基于全卷积神经网络对隧道工程缺陷检测中表面裂缝检测方法进行了探究。通过已有照明拍摄平台采集到的隧道表面图片,采用全卷积神经网络模型对图像中裂缝识别分类,再通过Adam优化器进行细化分割。研究结果表明该模型具有一定的可行性,模型评价结果较好。输出的高精度裂缝骨架可以对缺陷类型进行判定和初步处理,为后续实际缺陷评测和修补工程提供参考。

Tunnels are an important component of the comprehensive transportation system and a basic guarantee for people’s convenient living.In response to the call for new engineering construction,the new round of technological revolution and industrial transformation,and the improvement of scientific construction requirements,this article explores the surface crack detection method in tunnel engineering defect detection based on fully convolutional neural networks.The tunnel surface images collected through existing lighting shooting platforms are used to identify and classify cracks in the images using a fully convolutional neural network,and then refined and segmented using an Adam optimizer.The research results indicate that the model proposed in this paper has a certain degree of feasibility,and the model evaluation results are good.The high-precision crack skeleton output can be used to determine and preliminarily process defect types,providing reference for subsequent actual defect evaluation and repair engineering.