基于支持向量机的大跨度拱桥损伤识别方法研究  

Damage identification of a long-span arch bridge based on support vector machine

作　　者：刘春城[1,2] 刘佼[1] 

机构地区：[1]东北电力大学建筑工程学院,吉林132012 [2]大连理工大学海岸与近海工程国家重点实验室,大连116023

出　　处：《振动与冲击》2010年第7期174-178,共5页Journal of Vibration and Shock

基　　金：交通部西部科技项目(200631882350);吉林省教育厅"十一五"科技发展计划项目(200627);吉林市杰出青年科技计划项目(200804)

摘　　要：作为一种新兴的机器学习算法,支持向量机在损伤识别中已显示出其回归能力的优越性。将模态曲率改变率作为损伤识别特征参数,提出了基于支持向量机的大跨度拱桥损伤识别方法。首先应用模态曲率改变率进行损伤定位识别,然后重新构造训练样本,利用最小二乘支持向量机方法进行大跨度拱桥的损伤程度识别,该方法在较少的样本条件下,取得了非常接近目标值的识别效果。通过与RBF神经网络的训练结果进行对比,验证了该方法的精确性。As a new machine learning algorithm,the method of support vector machine (SVM) shows its superiority of the ability of regression in the fields of damage identification.Considering variation ratio of curvature mode as the feature parameters of damage identification,the method of the damage identification of a long-span arch bridge based on SVM was presented.At first,the variation ratio of curvature mode was used to carry on damage location identification.Then,the training samples were reconstructed,the method of least square support vector machine was used to identify the long-span arch bridge damage level,and the identification results were very close to the target values under the condition of small samples.Compared with the results from the RBF neural network,the precision of the proposed method was verified.

关 键 词：支持向量机 模态曲率改变率 损伤识别 拱桥 吊杆 

分 类 号：U448.34[建筑科学—桥梁与隧道工程]