基于大数据的公路路面养护智能决策研究  

Research on Intelligent Decision of Highway Pavement Maintenance Based on Big Data

作　　者：李辉 LI Hui(China Merchants Highway Beijing-Tianjin-Tanggu Engineering Construction Management(Tianjin)Co.,Ltd.,Tianjin 300051)

机构地区：[1]招商公路京津塘工程建设管理(天津)有限公司,天津300051

出　　处：《公路交通技术》2023年第2期82-88,共7页Technology of Highway and Transport

摘　　要：在制定运营公路路面养护计划或实施养护工程过程中,公路管理者常忽略全生命周期养护成本,且因缺乏科学的养护决策辅助技术延误最佳养护时机,应这方面的实际要求提出基于路面大数据分析的养护决策分析方法,主要包括建立养护大数据库、变长分段法划分路段单元、通过随机森林算法预测路况衰变、多层感知算法输出养护方案、非支配排序遗传算法推荐路网级养护最优方案。结果表明:1)养护数据标准制定及养护数据清洗是大数据算法建立的基础;2)决策分析过程涉及的变长分段法、多种预测模型的偏离度对比分析、考虑静动态大数据的神经网络模型更具有工程实际意义;3)通过与实际养护工程项目对比,本文研究的大数据养护决策系统生成的决策方案匹配率达到93%,表明该大数据养护决策系统具有较好的应用前景及实际工程价值。In the process of formulating operational pavement maintenance plans or implementing maintenance projects,highway managers often ignore the cost of the full life-cycle maintenance,and the lack of scientific maintenance decision support technology delays the best time for maintenance.Therefore,a maintenance decision analysis method based on pavement big data analysis should be proposed,according to the actual requirements in this regard.It mainly includes establishing a large maintenance database,dividing section units by variable length section method,predicting the decline of road conditions by the random forest algorithm,outputing the maintenance scheme by the multi-layer perception algorithm,and recommending the optimal plan for road network level maintenance by the non dominated sorting genetic algorithm.The results indicate that:1)formulating standards for maintenance data and cleaning maintenance data are the basis for conducting big data algorithm calculations;2)The variable length segmentation method of decision analysis process design,the comparative analysis of deviation degree of various prediction models,and the neural network model considering static and dynamic big data have more practical engineering significance;3)By comparing the decision plans generated by the big data maintenance decision-making system with the matching situation of actual engineering maintenance projects,the matching rate reached 93%,proving that the big data maintenance decision-making system has good application prospects and practical engineering value.

关 键 词：路面养护决策 随机森林算法 多层感知算法 资金优化分配 

分 类 号：U418.6[交通运输工程—道路与铁道工程]