基于Bayesian理论的无参考信号主动Lamb波损伤定位方法

Baseline free time reversal method for active Lamb wave damage localization following Bayesian theory

目前主动Lamb波损伤检测方法大都基于健康状态下的参考信号以获取损伤散射信号,并采用确定性的方法进行损伤定位,在适用性方面受到较大影响。基于Lamb波的时间反转聚焦原理,考虑损伤定位过程中的不确定性因素,发展了一种基于Bayesian理论的无参考信号主动Lamb波损伤定位方法。根据时反聚焦信号中主、旁瓣波包峰值的相对时刻,提取各测量通道损伤散射信号传播时间与直达波传播的时间之差作为样本,并基于Bayesian理论,获得损伤位置、波速等未知参数的联合后验概率分布,再利用马尔科夫链蒙特卡洛(MCMC)方法在后验分布中对未知参数进行采样估计,得到马尔科夫链的极限分布即未知参数的后验分布。通过对一铝质矩形薄板的数值仿真和模型实验研究结果表明,该方法能够较为准确地识别出损伤位置、波传播速度及其不确定性。

The commonly used active Lamb wave damage detection methods generally use the baseline signal from healthy structure to extract the damage-scattered signal and use deterministic approaches for damage localization.So its performance is usually poor in real applications.By employing the Lamb wave time reversal theory and considering the uncertainties from damage localization procedure,we develop a baseline free active Lamb wave damage localization methodology based on the Bayesian theory.According to the relative time for the wave packets of main band and associated two side bands within the time reversal signals,the time difference between the damage-scattered signal and the direct wave for each measurement path is firstly extracted as sample sets,and then,the Bayesian approach is utilized to obtain the posterior probability density functions（PDF）of the unknown parameters related to the damage location and wave velocity.After that,a Markov Chain Monte Carlo（MCMC）method is employed for sampling the joint posterior distributions of the unknown parameters.The resulting Markov chain is stationary and can represents the posterior distributions of each uncertain parameter.The results from both numerical simulation and experimental investigation for an Aluminum square plate indicate that the damage location,wave velocity and their associated uncertainties can be well identified by the proposed method.