基于数据融合的结构损伤识别方法研究-工程力学.PDF

第25 卷第1 期 Vol.25 No. 1 工 程 力 学 2008 年 1 月 Jan. 2008 ENGINEERING MECHANICS 16 文章编号：1000-4750(2008)01-0016-06 基于数据融合的结构损伤识别方法研究 *刘 涛，李爱群，缪长青，李枝军 (东南大学土木工程学院混凝土及预应力混凝土结构教育部重点实验室，江苏，南京 210096) 摘 要：为提高结构损伤识别的准确率，在大型复杂土木工程结构的健康监测中引入多传感器数据融合技术，提 出了基于数据融合技术的结构损伤识别方法的实用模型，建立了结构损伤识别方法的评价指标，并通过钢筋混凝 土板试验分析了常用的数据融合方法——Bayes 方法和D-S 证据理论的有效性。分析结果表明：采用数据融合技 术，结构损伤识别方法的评价指数更高，即结构损伤识别的效果更好。 关键词：土木工程；数据融合；结构损伤识别；Bayes 方法；D-S 证据理论 中图分类号：TU311.41; TB123 文献标识码：A STUDY OF STRUCTURAL DAMAGE IDENTIFICATION METHODS BASED ON DATA FUSION * LIU Tao , LI Ai-qun , MIAO Chang-qing , LI Zhi-jun (Key Laboratory of Concrete and Prestressed Concrete Structures of Ministry of Education, College of Civil Engineering, Southeast University, Nanjing 210096, China) Abstract: To improve the validity of structural damage identification, multi-sensor data fusion technique is imported into structure health monitoring for large-scale civil engineering structures. In this paper, practical models for structural damage identification based on data fusion technique are put forward, and an evaluating indicator for structural damage identification methods is developed. Bayes method and D-S evidence theory, which are in common use in data fusion technique, are discussed through an experiment on two RC plates. The analysis results reveal that damage identification evaluating indicators with data fusion technique are higher than those with single structural damage identification method. That is to say, structural damage identification methods based on data fusion yield better