基于损伤诱因反演的高桩码头结构健康评估  

Structural health assessment of high-pile wharves based on damage cause inversion

作　　者：周世良[1] 唐盟涵 孙世泉 徐瑛 谢雷 ZHOU Shiliang;TANG Menghan;SUN Shiquan;XU Ying;XIE Lei(The Southwestern Research Waterway Engineering Institute,Chongqing Jiaotong University,Chongqing 400074,China;Chongqing Shipping Construction Development Co.,Ltd.,Chongqing 401120,China)

机构地区：[1]重庆交通大学西南水利水运工程科学研究院,重庆400074 [2]重庆航运建设发展(集团)有限公司,重庆401120

出　　处：《水利水运工程学报》2025年第1期157-167,共11页Hydro-Science and Engineering

基　　金：重庆市交通科技项目(CQJT2022ZC03);重庆市教委重点科技项目(KJQN201800739)。

摘　　要：现有的码头结构评价方法具有滞后性,无法持续获取结构的安全信息。基于损伤诱因反演研究,综合码头常见的两种失效模式,引入承载力富余系数作为码头结构评价指标,依托机器学习方法开展损伤诱因强度、位置、承载力富余系数的预测。结果表明:粒子群优化神经网络(Particle Swarm Optimization-Back Propagation,PSO-BP)模型对强度反演与承载力富余预测效果最好;粒子群优化支持向量机(Particle Swarm Optimization Support Vector Machine,PSO-SVM)优化模型对损伤诱因位置识别精度最高。本研究中PSO-BP与PSO-SVM模型预测精度较高,为基于损伤诱因反演的码头健康监测方法提供了状态评价指标,可为码头结构长期、动态评价提供参考。Existing methods for evaluating wharf structures are characterized by lag and inability to continuously obtain structural safety information.This study investigates damage cause inversion and integrates two common failure modes of wharves.The load redundancy coefficient is introduced as an evaluation index for wharf structures.Using machine learning methods,the study predicts the intensity,location,and load redundancy coefficient of damage causes.Results indicate that the PSO-BP(Particle Swarm Optimization-Back Propagation)neural network model achieves the best performance in strength inversion and load redundancy prediction.The PSO-SVM(Particle Swarm Optimization-Support Vector Machine)model excels in identifying the location of damage causes.The high predictive accuracy of the PSO-BP and PSO-SVM models in this study provides status evaluation indices for wharf health monitoring based on damage cause inversion,offering a reference for long-term and dynamic evaluation of wharf structures.

关 键 词：高桩码头 机器学习 损伤 反演 承载力 

分 类 号：U656.1[交通运输工程—港口、海岸及近海工程]