面向CFRP损伤检测的改进L1正则化EIT方法

Damage detection for CFRP based on the modified L1 regularization of EIT

由于碳纤维复合材料(carbon fiber reinforced polymer,CFRP)具有导电特性,因此可利用电阻抗层析成像(electrical impedance tomography,EIT)技术实现结构损伤检测。考虑EIT电极位置、数量有限以及电场软场特性的影响,其灵敏度矩阵呈现非均匀分布特征,导致场域内不同位置成像效果差别较大。该研究基于传统L1正则化的边缘锐化特性,提出了一种改进L1正则化方法。该方法通过分析损伤重建电导率分布规律,发现其在各邻域梯度分布的特征,并利用此特征保留L1正则化可能滤除的有效数据,在保证重建图像锐利边缘的同时,改善了不均匀灵敏度分布导致的区域成像精度差别。为验证所提出方法的有效性,仿真中设计了不同类型、不同相对位置的损伤模型,并且通过搭建的16电极EIT测试系统对损伤CFRP样品进行测试。仿真和试验结果表明,改进后的L1正则化方法不仅能重建更多位置上的损伤图像,而且不同程度地提高了各类损伤重建图像的精度。

Due to the conductivity property of carbon fiber reinforced polymer(CFRP),electrical impedance tomography(EIT)can be used to detect the damage of CFRP made structures.Considering the influences of the position and limited number of EIT electrodes,and soft electric field characteristics,the sensitivity matrix presents the characteristic of non-uniform distribution,resulting in the great difference of imaging effect at different positions of the field.Aiming at the edge sharpening characteristics of traditional L1 regularization,an improved L1 regularization method was proposed.Based on analyzing the conductivity distribution feature of the damage reconstruction,the characteristics of its gradient distribution in each neighborhood area were utilized to retain the effective data that L1 regularization might filter out.Therefore,the sharp edge of the reconstructed image can be retained and the disparity of regional imaging accuracy caused by non-uniform sensitivity distribution is decreased.In order to verify the effectiveness of the proposed method,damage models of different types and relative positions were designed in the simulation.In addition,the damaged CFRP samples were tested on a designed 16-electrode EIT test system.The simulation and experimental results show that the modified L1 regularization method can not only reconstruct more damage images at different positions,but also improve the accuracy of various damage reconstruction images to some extent.