摘要
为在桥梁复检时精准推荐子部件和病害类型,避免复检时全面普检效率低和人资物力浪费的问题,提出一种基于协同过滤算法的桥梁复检指引应用方法。该方法基于句向量思想和协同过滤思想,首先分析桥梁与子部件病害类型的关联关系,得出需复检的子部件,同时结合现有海量检测数据,分析不同桥型、不同部件(子部件)和病害类型的关系;然后为复检推荐出关联子部件最有可能发生的病害,引入计算相似度的杰卡德公式,根据桥梁及部件的相似特征,在样本库中寻找最佳相似桥梁和部件,并给出参照评分。以某高速公路桥梁为应用背景,对比分析现场复检病害类型与推荐病害类型,以及复检得分和参照评分。结果表明:指引的子部件病害类型在复检时被检测出的可能性很大,准确率达到75%以上;复检评分与参照评分相差率基本小于10%;整体提升检测效率约50%;该协同过滤算法可以应用于桥梁复检指引。
This study presents a collaborative filtering algorithm developed following the philosophy of sentence vector and collaborative filtering,which can accurately recommend the subcomponents needing extra inspection and deterioration patterns during the inspection of bridges,thereby avoiding the low efficiency of a full inspection and waste of human and material resources.First,the correlation between the bridge and deterioration patterns of subcomponents is analyzed to find the subcomponents that need extra inspection.Additionally,based on the substantial inspection data,the relationship among bridge types,varying components(subcomponents)and deterioration patterns is analyzed.Finally,deteriorations that are most likely to appear to the correlated subcomponents are recommended for reinspection.Following the principle of Jacquard similarity and considering the similar characteristics of the bridge and components,the most similar bridges and components are selected from the sample database,and reference scores are given.For an existing highway bridge,the deteriorations observed on site were compared with the recommended deterioration patterns,and the reinspection score is compared with the reference score.It is shown that the recommended deterioration patterns for the correlated subcomponents are likely to be found,with a probability above 75%,and the differential rate of the reinspection score and reference score is less than 10%.The overall inspection efficiency is improved by 50%,and the collaborative filtering algorithm can be used to guide the inspection of bridges.
作者
斯新华
刘大洋
韩坤林
黄倩文
SI Xinhua;LIU Dayang;HAN Kunlin;HUANG Qianwen(China Merchants Chongqing Highway Engineering Testing Center Co.,Ltd.,Chongqing 400067,China;TY Lin International Engineering Consulting(China)Co.,Ltd.,Chongqing 401121,China)
出处
《世界桥梁》
北大核心
2024年第5期118-124,共7页
World Bridges
基金
国家重点研发计划项目(2023YFB2604400)
重庆市技术创新与应用发展专项面上项目(CSTB2022TIAD-GPX0029)
重庆市技术创新与应用发展专项重点项目(cstc2019jscx-gksbX0008)。
关键词
桥梁工程
协同过滤算法
子部件
病害
复检
指引
技术状况
桥梁检测
bridge engineering
collaborative filtering algorithm
subcomponent
deterioration
reinspection
guidance
technical condition
bridge inspection
