东北大学学报（自然科学版）

东北大学学报:自然科学版 ›› 2017, Vol. 38 ›› Issue (2): 290-294.DOI: 10.12068/j.issn.1005-3026.2017.02.028

• 资源与土木工程 • 上一篇    下一篇

基于动力指纹和Bayes数据融合的桥梁损伤识别

孙爽， 梁力， 李明， 李鑫   

  1. (东北大学 资源与土木工程学院， 辽宁 沈阳110819)
  • 收稿日期:2015-09-23 修回日期:2015-09-23 出版日期:2017-02-15 发布日期:2017-03-03
  • 通讯作者: 孙爽
  • 作者简介:孙爽(1984-)，女，辽宁营口人，东北大学博士研究生； 梁力(1955-)，男，辽宁丹东人，东北大学教授，博士生导师.
  • 基金资助:
    国家自然科学基金资助项目(51474048).

Identification of Bridges Damage by Dynamic Fingerprints and Bayes Data Fusion

SUN Shuang， LIANG Li， LI Ming， LI Xin   

  1. School of Resources & Civil Engineering， Northeastern University， Shenyang 110819， China.
  • Received:2015-09-23 Revised:2015-09-23 Online:2017-02-15 Published:2017-03-03
  • Contact: SUN Shuang
  • About author:-
  • Supported by:
    -

RichHTML

0

PDF (PC)

829

摘要： 受测量噪声的影响，采用单一指标评价桥梁安全容易产生误判，因此提出一种基于Bayes理论的桥梁损伤识别方法.该方法将识别过程分解为损伤定位识别与损伤定量识别两部分，首先采用Bayes公式融合归一化的动力指纹，进行损伤位置识别，进而提取损伤处的动力指纹构建Bayes网络，计算各节点的条件概率，从而识别损伤程度.通过简支梁数值模拟验证了该方法具有良好的抗噪性，尤其能够对小损伤准确定位，对程度差别小的损伤准确分类.

关键词: 桥梁损伤识别, 动力指纹, 数据融合, Bayes网络, 测量噪声

Abstract: Using a single index to evaluate bridge safety is likely to draw a false conclusion due to the effect of measurement noise. Thus， an identification method of bridge damage based on the Bayes theory was proposed， which includes location identification and degree identification. First， the Bayes formula was used to fuse the normalized dynamic fingerprints and identify the damage locations. Then， only the fingerprints of damaged locations were extracted to construct the Bayesian network and the conditional probability of every node was calculated in order to identify the damage degrees. By simulating a simple beam， the result shows that the proposed method has a good anti-noise performance， especially can accurately locate tiny damages and distinguish the damages with slight differences.

Key words: bridge damage identification, dynamic fingerprint, data fusion, Bayesian network, measurement noise

中图分类号: