基于小波分解和极限学习机的柔性PVC基材损伤识别研究  

Damage identification of flexible PVC substrate based on wavelet decomposition and limit learning machine

作　　者：朱平 严宏鑫 ZHU Ping;YAN Hongxin(School of Instrumentation and Electronics,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学仪器与电子学院,太原030051

出　　处：《振动与冲击》2022年第13期220-227,共8页Journal of Vibration and Shock

基　　金：装发快速扶持项目第二阶段(61409220157);装发基础研究项目重大专项(514010504-308)。

摘　　要：针对柔性天线在恶劣机载环境下基底材料容易出现孔洞、裂纹损伤等情况,提出采用兰姆波检测技术进行柔性聚氯乙烯(PVC)基材结构中的损伤探测和识别。该损伤识别方法是将小波分解和极限学习机相结合,对回波信号进行小波阈值去噪和损伤特征提取,从时域、频域和变换域3个层次构建出回波信号的特征向量,训练、测试分类出无损伤、孔洞、槽形损伤类型。搭建了基于兰姆波的柔性PVC基材损伤检测试验平台进行验证,结果表明:柔性基底上的无损伤、孔洞、槽形损伤回波信号经小波阈值去噪、分解后,提取出特征向量,得到的低频缓变回波重构信号特征明显;采用ELM模型进行训练和测试分类出损伤类型,分类精度达到92.56%,与BP神经网络分类方法作对比,ELM分类方法具有较好的分类性能,从而验证了小波分解和极限学习机相结合来识别柔性PVC基材损伤特征的有效性。此检测方法在柔性电子器件早期损伤的现场快速检测领域具有广泛的应用前景。Here,aiming at the situation of flexible antenna substrate being easy to generate holes and crack damages in harsh airborne environment,Lamb wave detection technique was proposed to detect and identify damage in flexible polyvinyl chloride(PVC)substrate structure.This damage identification method could combine wavelet decomposition and limit learning machine to perform wavelet threshold denoising and damage feature extraction for echo signals,and construct feature vectors of echo signals from 3 levels of time domain,frequency domain and transform domain.After training and testing,types of no damage,hole and groove damages were classified.The test platform for damage detection of flexible PVC substrate based on Lamb wave was built for verification.The results showed that echo signals of no damage,hole and groove damages on flexible substrate are denoised and decomposed using wavelet threshold to extract feature vectors,reconstructed low-frequency slowly varying echo signals have obvious features;ELM model is used for training and testing to classify damage types,and the classification accuracy is 92.56%;compared with the BP neural network classification method,the ELM classification method has better classification performance to verify the effectiveness of combining wavelet decomposition and limit learning machine to identify damage characteristics of flexible PVC substrate;this detection method has a wide application prospect in onsite rapid detection field for early damage of flexible electronic devices.

关 键 词：小波分解 极限学习机 柔性基材 损伤识别 

分 类 号：TN431.1[电子电信—微电子学与固体电子学]