轴
计算机科学
卷积神经网络
动态称重
桥(图论)
人工智能
直线(几何图形)
偏转(物理)
人工神经网络
计算机视觉
实时计算
工程类
数学
结构工程
医学
光学
物理
内科学
几何学
作者
Junki Mori,Junji Yoshida,Koichi Takeya
标识
DOI:10.2749/seoul.2020.281
摘要
<p>Investigating traffic loads and the number of vehicles on bridges is essential in order to grasp factors of deterioration in road bridges. Bridge Weigh-in-Motion (B-WIM) is a method for estimating vehicle axle weight from the response of vehicles passing through a bridge. In this study, we construct a new B-WIM, in which vehicles are tracked from video images and influence line of the bridge is estimated from the response by local buses. As a method of tracking vehicles from video images, we applied Faster Regions with Convolutional Neural Network (Faster R-CNN), which is a method of image processing using deep learning. In addition, influence lines are inversely estimated by the direct search method using deflection responses by local buses. Consequently, the proposed method could estimate axle weights of a large vehicle with over 95 % accuracy.</p>
科研通智能强力驱动
Strongly Powered by AbleSci AI