摘要
文章以珠江黄埔大桥北汊斜拉桥为研究对象,采用有限元模型计算了桥梁在随机车流作用下不同拉索损伤工况所对应的应变响应,进一步采用应变响应统计信息,引入随机森林和梯度提升决策树2种机器学习算法对损伤拉索进行识别,从而提出了一种结合桥梁响应统计信息与机器学习的损伤识别方法。计算结果表明,该方法在损伤定位和定量上均表现出良好的效果,验证了其应用于大跨度桥梁的可行性和有效性。该文提出的方法为有效利用桥梁长期健康监测数据进行损伤识别提供了思路。
Strain responses of the north branch of Huangpu Bridge over Pearl River with different damaged cable scenarios under the action of random traffic flow are calculated using finite element model. With constructing statistical parameters of the strain responses as input for machine learning algorithms such as random forest and gradient boosting decision tree, an effective damage identification method is proposed. The results show that the proposed method exhibits favorable performance for damage localization and quantitative evaluation, demonstrating its applicability in damage detection for long-span bridges. This method provides guide for making effective use of long-term health monitoring data to identify damage for bridges.
作者
蒋雍建
刘逸平
周立成
刘泽佳
程楠
汤立群
JIANG Yongjian;LIU Yiping;ZHOU Licheng;LIU Zejia;CHENG Nan;TANG Liqun(School of Civil Engineering and Transportation,South China University of Technology,Guangzhou 510641,China;State Key Laboratory of Subtropical Building Science,South China University of Technology,Guangzhou 510641,China)
出处
《合肥工业大学学报（自然科学版）》
CAS
北大核心
2019年第12期1667-1671,共5页
Journal of Hefei University of Technology：Natural Science
基金
国家自然科学基金资助项目(11602087)
广东省科技计划资助项目(2015B010131009)
关键词
桥梁健康监测
有限元
损伤识别
统计特征
机器学习
bridge health monitoring
finite element method
damage identification
statistical features
machine learning
