APP下载

基于岭估计和L曲线的损伤识别技术研究

2016-01-18郭惠勇,罗乐,盛懋等

振动与冲击 2015年4期

第一作者 郭惠勇 男,博士,副教授,1971年生

基于岭估计和L曲线的损伤识别技术研究

郭惠勇1, 罗乐2, 盛懋1, 李正良1

(1. 重庆大学 土木工程学院, 重庆400045; 2. 成都城电电力工程设计有限公司,成都610041)

摘要:为了解决结构损伤方程求解的可靠性问题,提出了基于岭估计和L曲线的损伤方程求解方法。首先描述了基于模态应变能灵敏度的结构损伤方程,然后考虑测量噪声等因素易造成病态的损伤方程问题,提出了采用岭估计求解结构的损伤方程,并利用L曲线确定最优的岭参数,最后建立了损伤估计值的修正方法。数值仿真结果表明,基于修正的岭估计和L曲线方法可以较为精确地识别出损伤的位置和程度,而且其定量识别结果明显好于基本的岭估计和L曲线方法以及一般的截断奇异值分解法。

关键词:损伤识别;损伤方程;岭估计;L曲线;位移模态

基金项目:国家自然科学基金(51468058);教育部留学回国人员科研启动基金(20121707)

收稿日期:2014-02-08修改稿收到日期:2014-04-03

中图分类号:TB123文献标志码: A

Structural damage identification method based on ridge estimation and L-curve

GUOHui-yong1,2,LUOLe2,SHENGMao1,LIZheng-liang1(1. School of Civil Engineering, Chongqing University, Chongqing 400045, China;2. Chengdu Chengdian Electric Power Engineering Design Co., Let., Chengdu 610041, China)

Abstract:In order to solve the reliability problem of structural damage equation, a damage detection method based on ridge estimation and L-curve was presented. Firstly, a structural damage equation based on modal strain energy sensitivity was described. Then, considering measurement noise easily making the damage equation be ill, the ridge estimation was utilized to solve the damage equation and L-curve is utilized to identify optimal ridge parameters. Finally, an improved ridge estimation method was proposed to enhance the identification precision. Simulation results demonstrated that the proposed ridge estimation and L-curve methods can detect structural damage locations and level with good accuracy, their quantitative detection results are obviously better than those of the basic ridge estimation and L-curve methods and the truncated singular value decomposition method.

Key words: damage identification; damage equation; ridge estimation; L-curve; displacement modal shape

工程结构在投入使用后,由于环境的作用,常会产生各种损伤。结构的严重损伤会引起建筑物的倒塌或者失效,从而造成巨大的经济和人员生命的损失。因此,对结构损伤的识别研究一直是国际上的研究热点[1-3]。利用损伤前后动力特性的改变,建立相应的损伤方程,从而求解出损伤系数是较为有效的方法[4-5]。但是,在工程实际中,测试数据不可避免的会受到测量噪声、环境因素等的干扰,常会使损伤方程产生病态问题,从而给直接求解带来困难,造成损伤的解极不稳定,与真实值相差较远。……

登录APP查看全文