关键词:
螺栓故障
3D点云
ICP
RANSAC
摘要:
城市轨道交通作为一种城市内交通运输方式,因其快速高效、低碳环保、运力强大等优点得到了国家产业政策的大力支持,近几年来,我国城市轨道交通发展迅速、列车速度不断提高,其安全问题愈发重要。列车长时间的运行,由于震动、碰撞、老化等问题,列车底部的紧固螺栓会出现松动、甚至掉落丢失的情况,存在发生重大安全事故的风险。而现阶段采取的人工巡检的方法存在效率低、漏检多、标准不一等诸多限制。为了避免造成人员伤亡和财产损失,亟需研究出一种快速、准确的螺栓故障检测方法。本文采用智能列检机器人采集列车车底螺栓图像,合成点云数据,通过ICP算法与模版数据点云配准、RANSAC平面分割消除异常点,计算螺栓表面点云的数量与平面距离判断列车车底螺栓是否正常,实验表明,该方法可以有效的识别出列车车底螺栓丢失和3 mm以上松动故障,真实故障识别率为100%。对减少人工成本、排除列车安全隐患、保障人民生命财产安全具有重要意义。Urban rail transit, as an intra-city transportation mode, has been strongly supported by the national industrial policy due to its advantages of high speed, high efficiency, low carbon and environmental protection, and strong transportation capacity. Its security issues are becoming more and more important. When the train runs for a long time, due to vibration, collision, aging and other problems, the fastening bolts at the bottom of the train will be loose, or even lost, and there is a risk of major safety accidents. However, the manual inspection method adopted at this stage has many limitations, such as low efficiency, many missed inspections, and different standards. In order to avoid casualties and property losses, it is urgent to develop a fast and accurate bolt fault detection method. In this paper, the intelligent train inspection robot is used to collect the images of the bolts on the bottom of the train, synthesize the point cloud data, and use the ICP algorithm to register the point cloud with the template data, RANSAC plane segmentation to eliminate abnormal points, and calculate the number of point clouds on the bolt surface and the plane distance to judge the train bottom. Whether the bolts are normal or not, the experiment shows that this method can effectively identify the loss of the bolts on the bottom of the train and the loose faults of more than 3mm, and the real fault recognition rate is 100%. It is of great significance to reduce labor costs, eliminate hidden dangers of train safety, and ensure the safety of people’s lives and properties.