基于多源数据融合的高速公路桩号数字化研究及应用

Research and Application of Expressway Stake Digitization Based on Multisource Data Fusion

目前高速公路上普遍存在着设计、施工、运营等资料交接不全或分段设计施工的问题,交通行业相关运营管理部门急需一种能安全、快速、准确地调研桩号现状的方法及相关软硬件装备。基于多源数据融合算法,采用全球卫星定位系统(GNSS)、光电编码器(DMI)、图像识别数据融合的方法,建立了桩号测量及数据校核模型,应用人工智能深度学习对公里桩/百米桩智能提取算法。结果表明:对比人工徒步GNSS打点、车载人工GNSS打点等传统方法,所提方法能有效提升桩号测量效率,通过对道路环境图像和GNSS进行时空同步,将每个公里桩作为校正点,确保每个公里桩之间的误差不进行累积,定位误差不超过5 m。研究成果数据已服务于湖北交投运营集团的智慧养护管理系统底层数据库,为后期高速公路养护设计、检测、施工、信息化等提供了精准数据支撑。

At present,there is a common problem in the transportation industry,especially on highways,due to incomplete handover of design,construction,and operation data or segmented design and construction.The relevant operation management departments in the transportation industry urgently need a safe,fast,and accurate method and related software and hardware equipment to investigate the current situation of pile numbers.This paper was based on a multi-source data fusion algorithm,used the methods of Global Satellite Positioning System(GNSS),Optoelectronic Encoder(DMI),and image recognition data fusion to establish a station measurement and data verification model.Artificial intelligence deep learning was applied to intelligently extract kilometer/hundred meter stations.The results showed that compared with traditional methods such as manual walking GNSS marking and vehicle mounted GNSS marking,this method has improved the efficiency of pile number measurement.By synchronizing the road environment image and GNSS in time and space,each kilometer mark was used as a correction point to ensure that the error between each kilometer mark does not accumulate,and the positioning error does not exceed 5m.The research results data of this paper have been applied to the underlying database of the intelligent maintenance management system of Hubei Jiaotou Operation Group.This provides precise data support for the later maintenance design,testing,construction,and informatization of highways.