地铁车站施工机械车辆伤害事故监测及风险评价

唐爽

摘 要

随着地铁建设进程的迅速发展，地铁车站施工规模日益增大，给其安全管理带来了更大的挑战。土石方工程作为地铁车站施工中的关键环节，经常需要用到诸如挖掘机、挖掘装载机、自卸卡车等的机械车辆，以实现土体的开挖与运输。这些活动自如、频繁往来的机械车辆无疑给空间带来了风险，主要表现为其与行走的工人发生碰撞而造成的伤害事故。然而，即使事故总数有所下降，这种碰撞事故在各类伤害事故中的占比却没有明显的下降或改变。这说明需要对其有更深入的认知，并建立更为有效的控制预防措施。

论文通过对地铁车站机械车辆作业的分析，结合事故致因理论，提出事故发生的物理原因和根本原因，试图从这两方面来进行事故的预防控制。一方面，利用机器视觉技术进行事故监测。机器视觉技术是许多国家的研究热点，已经取得了一些有效的成果。本文通过改进的更快区域卷积神经网络来识别施工现场的工人、挖掘机和人-机距离，并建立相应的安全规则表达危险状况，从而为现场管理人员开展工作提供依据。另一方面，通过分析机械车辆伤害事故危险因素，建立人、机、环、管四类因素的风险评价指标体系。然后运用层次分析法和模糊综合评价法对项目发生机械车辆伤害事故的风险进行评价，据此判断其是否有必要采取措施进行整改和预防。

论文所提出的物体识别算法对工人和挖掘机都具有较高的精度，这对事故监测至关重要。事故监测的结果可以帮助现场的管理人员有针对性地进行管控。事故的模糊综合评价则可以从整个项目的角度表达风险状况，为企业、项目管理人员作出管理决策提供依据。

关键词：地铁施工 机械车辆 碰撞 机器视觉 风险评价

Abstract

With the rapid development of metro construction, the expansion of metro station construction scale brings greater challenges to its safety management. As the key link in the building of metro stations, earthwork construction often requires the use of mechanical vehicles such as excavators, backhoe loaders, dump trucks, etc. to achieve excavation and transportation of soil. These mechanical vehicles that move freely and frequently have undoubtedly brought risks to the space, mainly manifested in the injuries caused by collisions between them and workers passing by. However, even if the total number of accidents has decreased, the proportion of such crashes in various types of injuries has not dropped or changed significantly. This shows that it needs to have a deeper understanding of it and establish more effective control and prevention measures.

Based on the analysis of the operation of mechanical vehicles in metro stations and the accident causation theory, the physical and fundamental causes of the accidents are proposed, from which the attempts to prevent and control the accidents are made. On the one hand, machine vision technology is used for accident monitoring. Machine vision technology is a research hotspot in many countries and has achieved some effective results. In this paper, the improved Faster Regional-Convolutional Neural Network is used to identify the workers, excavators and man-machine distances on the construction site, and corresponding safety rules are established to express dangerous situations, so as to provide the basis for site management personnel to carry out their work. On the other hand, by analyzing the risk factors of mechanical vehicle injury accidents, a risk assessment index system including four types of factors as human, machine, environment, and management is established. Then analytic hierarchy process and fuzzy comprehensive evaluation method are applied to evaluate the risk of mechanical vehicle injury accidents in the project, judging whether it is necessary to take measures for rectification and prevention.

The object recognition algorithm presented in this paper has high accuracy for workers and excavators, which is crucial for accident monitoring. The results of the accident monitoring can help the site management personnel to carry out targeted management and control. The fuzzy comprehensive evaluation of accidents can express the risk status from the perspective of the entire project and provide a basis for management decisions made by enterprises and project managers.

Key words：Metro construction Mechanical vehicle  Collision Machine vision Risk assessment