基于支持向量机与分层遗传算法的斜拉桥全结构损伤分步识别

Step-by-step damage detection for cable-stayed bridge based on support vector machine and hierarchic genetic algorithm

为了实现斜拉桥全结构的损伤识别,提出一种支持向量机与分层遗传算法相结合的分步识别方法。该方法首先按结构的材料特性将斜拉桥分为主梁、索塔、拉索三类子结构,利用支持向量机的分类特性判定损伤的来源,确定损伤属于某一类子结构;然后,应用分层遗传算法对子结构中的单元进行损伤位置与损伤程度的识别。以实验室独塔斜拉桥模型作为研究对象进行数值仿真,结果表明:采用支持向量机方法能较准确的对主梁、索塔、拉索三类子结构的损伤进行分类,确定损伤的来源;分层遗传算法能快速有效的完成斜拉桥某一子结构中损伤单元的定位与识别;两种算法结合的分步识别方法,实现了斜拉桥全结构的损伤识别,同时分步识别策略减少了支持向量机训练样本与遗传算法中初始种群的规模,提升了寻优效率。

In order to achieve the whole structure damage detection for cable-stayed bridge,a new kind of step-bystep damage detection method based on support vector machine( SVM) and hierarchic genetic algorithm was proposed in this paper. Firstly,according to the material characteristics of the structure,the cable-stayed bridge would be divided into three substructures,such as main girder,pylon and stay cable substructure,and Support Vector Machine method could be used to detect the source of damage to determine the damage belongs to one kind of substructure; Secondly,the element belonged to damage substructure was further identified by hierarchic genetic algorithm to determinate the exact location and extent of damage. The numerical simulation for the single-tower cable-stayed bridge in laboratory was made and it is shown that SVM can identify the source of damage exactly and hierarchic genetic algorithm can quickly and effectively completes the identification of the damage element in the substructure and the whole structure damage detection of cable-stayed bridge can be achieved step by step. The num-ber of training samples of SVM and the size of the initial population of hierarchic genetic algorithm are reduced and the efficiency optimization is enhanced in this method.