基于BO-KELM的大跨变截面连续箱梁桥损伤识别研究  

Damage Identification of Long-span Variable Cross-section Continuous Box Girder Bridges Using BO-KELM

作　　者：吴庆亮 赵洋 崔书宇 姚兵 高桢赞 WU Qingliang;ZHAO Yang;CUI Shuyu;YAO Bing;GAO Zhenzan(School of Civil Engineering and Transportation,North China University of Water Resources and Electric Power,Zhengzhou 450046,China;Communications Construction Company of CSCEC 7th Division Co.,Ltd.,Zhengzhou 450000,China)

机构地区：[1]华北水利水电大学土木与交通学院,河南郑州450045 [2]中建七局交通建设有限公司,河南郑州450000

出　　处：《华北水利水电大学学报(自然科学版)》2025年第2期112-119,共8页Journal of North China University of Water Resources and Electric Power:Natural Science Edition

基　　金：河南省科技攻关项目(202102310251);中建七局科技研发课题(CSCEC7b-2021-Z-4);华北水利水电大学研究生创新能力提升工程(NCWUYC-2023046,NCWUYC-202315058)。

摘　　要：大跨度变截面连续箱梁桥是我国公路和市政桥梁中经常采用的典型结构型式,在运营阶段,精准识别桥梁损伤是准确评价其结构健康状态的首要条件。针对桥梁缺陷产生的不确定性,提出了一种基于BO-KELM识别大跨度变截面连续箱梁桥损伤的方法。试验和工程应用结果表明:放大曲率模态作为损伤识别指标,可有效识别大型结构单点和多点的轻微损伤位置;基于KELM构建的预测模型并结合BO算法,可快速识别梁桥不同程度的损伤,预测精度可达98.75%,较传统神经网络提高10%。提出的放大曲率模态和BO-KELM损伤识别方法具有对微小损伤识别能力强、损伤程度判别速度快、精度高等特点。研究结果可为大跨度变截面连续箱梁桥损伤位置和损伤程度的快速精准识别提供理论和方法指导。Long-span variable cross-section continuous box-girder bridges represent a typical structural form widely used in Chinese highway and municipal bridges.During operational phases,accurate damage identification serves as the prerequisite for precise structural health assessment.To address uncertainties in bridge defects,this study proposes a BO-KELM(Bayesian Optimization-Kernel Extreme Learning Machine)based method for damage identification in such bridges.Experimental and engineering application results demonstrate that:①The amplified curvature mode,as a damage identification index,effectively detects locations of minor single-point and multi-point damage in large-scale structures;②The prediction model constructed using KELM combined with BO algorithm enables rapid identification of varying damage severities,achieving 98.75%prediction accuracy—a 10%improvement over traditional neural networks.The proposed methodology exhibits distinctive advantages including enhanced sensitivity to minor damage,rapid damage severity assessment,and high precision.These research findings provide valuable references for fast and accurate identification of damage locations and severities in long-span variable cross-section continuous box-girder bridges.

关 键 词：大跨度变截面连续箱梁桥 损伤识别 放大曲率模态 贝叶斯优化 核极限学习机 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TU311[自动化与计算机技术—控制科学与工程]