斜拉索检测机器人的智能表观检测研究

为解决斜拉索检测日益繁重的工作任务,提高检测效率,结合斜拉索检测机器人与智能检测技术对斜拉索的外表进行检测。通过机器人在斜拉索上攀爬收集表面图像,将图像输入到卷积神经网络中进行缺陷识别,以确定缺陷的区域和位置。该项技术已成功应用于广东番禺大桥斜拉索的缺陷检测,结果证实:1)机器人能够灵活地采集到完整的斜拉索外表图像,基于智能算法的图像识别过程能够准确分辨出斜拉索不同的缺陷状态;2)识别正确率超过99%,且检测效率比传统方法提高300%,可以显著解决当前斜拉索检测的难题。

Research on Intelligent Apperance Algorithm of Stay Cable Detection Robot

In order to solve the increasingly heavy task of cable detection and improve the detection efficiency,this paper combines the cable detection robot and intelligent detection technology to detect the appearance of cable. The robot climbs on the stay cable to collect the surface image,and inputs the image into the convolution neural network to identify the defect,so as to determine the area and location of the defect. This technology has been successfully applied to the detection of cable defects in Panyu Bridge,Guangdong province. The results show that the robot can collect the complete cable surface image flexibly,and the image recognition process based on the intelligent algorithm can accurately distinguish the different defect states of the cable. The recognition accuracy is more than 99%,and the detection efficiency is300% higher than the traditional method,which can significantly solve the current problem of cable detection.