铁路桥梁损伤的统计模式识别  

Statistical Pattern Recognition of Structural Damage Detection of Railway Bridges

作　　者：单德山[1] 付春雨[1] 李乔[1] 

机构地区：[1]西南交通大学土木工程学院桥梁工程系,四川成都610031

出　　处：《桥梁建设》2011年第1期18-21,共4页Bridge Construction

基　　金：国家自然科学基金(51078316);中央高校基本科研业务费专项资金科技创新项目(SWJTU09CX02)和专题项目(SWJTU09ZT02);抗震工程技术四川省重点实验室开放基金项目(SKZ200906)

摘　　要：针对铁路桥梁结构损伤的特点,提出采用分步识别方案与统计模式识别相结合的方法对其进行识别。将铁路桥梁损伤识别分为损伤预警、损伤定位和损伤程度诊断3个基本问题,再将其各分成3个步骤分别采用统计模式识别的两类分类器、多类分类器及回归机分类器进行损伤诊断。采用某铁路连续梁桥模型试验验证所提方法的正确性,结果表明该方法在抗噪声能力、求解方法及求解思路上与优化识别方法明显不同,其具有良好的推广能力和较强的抗噪声能力,可应用于实际铁路桥梁的损伤识别。In the light of the features of structural damage of railway bridges,the combined method of the step-by-step damage detection and statistical pattern recognition is proposed to be used for recognition of the damage.The damage detection recognition of the bridges is divided into three basic issues,that is,the damage early warning,damage locating and damage extent diagnosing.Each of these issues is further divided into three steps and the damage is diagnosed respectively by the binary-class pattern classification,multi-class pattern classification and support vector regression of the statistical pattern recognition.The correctness of the proposed method is verified by the model tests of a railway continuous girder bridge and the results indicate that the method is apparently different from the optimization recognition method in aspect of the anti-noise capability,solution methods and solution ideas.The method has good value of popularization,good anti-noise capability and can be applied to the damage detection recognition of practical railway bridges.

关 键 词：铁路桥 损伤 诊断 模式识别 模型试验 

分 类 号：U448.13[建筑科学—桥梁与隧道工程] U446.3[交通运输工程—道路与铁道工程]