材料科学与工程学报2025，Vol.43Issue(4)：542-549,566,9.DOI:10.14136/j.cnki.issn1673-2812.2025.04.005
基于卷积神经网络电阻抗成像的蜂窝夹芯结构损伤识别
Impact Damage Identification for Honeycomb Sandwich Structures by Using Convolutional Neural Network-based Electrical Impedance Tomography
摘要
Abstract
Considering the susceptibility of honeycomb sandwich structures to impact damage during use and maintenance,a method is proposed to prepare a conductive sensing layer on the surface of the structure and combine it with convolutional neural networks for damage identification.Firstly,the method uses screen printing technology to print the graphene conductive carbon ink on the surface of the honeycomb sandwich structure to form a conductive induction layer,so that the structure has real-time induction function.Electrical simulations of the conductive sensing layer under different damage conditions are performed using the finite element method,generating extensive datasets on the distribution and boundary voltage changes of electrical conductivity.These datasets are used to train the convolutional neural network,and its reliability is verified with a test set.Finally,boundary voltage data of the sensing layer are measured before and after impact damage,and the conductivity change image is reconstructed through the trained network to identify the damage.Experimental results show that compared with traditional algorithms,electrical impedance tomography based on convolutional neural networks offers superior performance in identifying both the size and location of the damage,providing a promising approach for impact damage identification in honeycomb sandwich structures.关键词
蜂窝夹芯结构/冲击损伤识别/导电感应层/电阻抗成像/卷积神经网络Key words
Honeycomb sandwich structure/Impact damage identification/Conductive sensing layer/Electrical impedance tomography/Convolutional neural network分类
通用工业技术引用本文复制引用
李雪峰,周登,严刚..基于卷积神经网络电阻抗成像的蜂窝夹芯结构损伤识别[J].材料科学与工程学报,2025,43(4):542-549,566,9.基金项目
国家自然科学基金项目(12472134) （12472134）
航空航天结构力学及控制全国重点实验室开放课题基金项目(MCMS-E-0423G02) （MCMS-E-0423G02）
