基于PSO—LSTM算法的公路桥梁结构损伤识别研究  

Research on Structural Damage Monitoring of Highway Bridges based on PSO-LSTM

作　　者：袁辉 谢庆 计明军 吴炜昌 曾斌 姬生忠 YUAN Hui;XIE Qing;JI Ming-jun;WU Wei-chang;ZENG Bin;JI Sheng-zhong(China Railway South Investment Group Limited,Yangjiang 529500,China;College of Transportation Engineering,Dalian Maritime University,Dalian 116026,China)

机构地区：[1]中铁南方投资集团有限公司,阳江市529500 [2]大连海事大学交通运输工程学院,大连市116026

出　　处：《公路》2025年第2期428-434,共7页Highway

基　　金：国家自然科学基金项目,项目编号71971035。

摘　　要：结构健康监测系统在公路桥梁中得到了广泛应用。利用PSO智能优化算法对LSTM神经网络模型中参数进行优化,设计了PSO—LSTM算法对公路桥梁进行结构损伤识别,以提升桥梁结构工况的实时监测状态和评估效果的准确率。通过建立斜拉桥有限元模型获取静态数据、动态响应数据和振动响应数据等桥梁结构监测数据,建立深度学习的数据库,对桥梁的结构工况进行损伤识别分析。结果表明,PSO—LSTM算法对桥梁结构工况损伤识别的准确率可达到平均98.53%以上的较高水平,优于单一的LSTM深度学习算法和目前应用的PSO—BP神经网络学习算法。为公路桥梁全生命健康周期管理中结构损伤监测提供了一种有效的解决方案。Structural health monitoring system has been widely used in highway bridges.Regarding the structural damage monitoring of highway bridges,the PSO optimization algorithm is employed to optimize the parameters of the LSTM neural network model.A PSO-LSTM algorithm is designed to identify the structural damage of highway bridges so as to enhance real-time monitoring and assessment accuracy of the structural conditions of bridges.A simulation model of a cable-stayed bridge is established to obtain static data,dynamic response data,and vibration response data,serving as a deep learning database for bridge structure monitoring.The results indicate that the PSO-LSTM algorithm achieves a high accuracy rate of over 98.53%on average in identifying damage to bridge structures,surpassing that of individual LSTM deep learning and PSO-BP neural network learning algorithms.The study provides an effective solution to the problem of structural damage monitoring in highway bridges,offering valuable insights for the comprehensive life-cycle management of bridge structures,and has significant reference value for practical applications in bridge engineering projects in future.

关 键 词：桥梁工程 健康养护 桥梁结构损伤监测 PSO—LSTM算法 

分 类 号：U446.2[建筑科学—桥梁与隧道工程]