摘要
针对金属薄板在工艺加工和成型的过程中会产生分层、孔洞、裂缝等内部缺陷的问题,采用兰姆波对钢材薄板进行损伤检测,提出一种基于小波包变换和支持向量机的金属薄板微弱损伤信号特征提取与识别方法。采用小波包变换对接收到的回波信号进行特征提取和筛选,回波信号经过小波包阈值降噪预处理,从时域、频域和时频域三个角度构造出25个特征值,并通过主成分分析进行降维处理提取出损伤特征向量,将样本数据集输入至粒子群优化后的支持向量机识别模型中进行损伤识别。结果表明:模型分类出无损伤、孔洞、槽型、焊点信号损伤类型,识别准确率达到91.67%,与BP神经网络方法进行分类对比,SVM识别模型具有较好的分类性能,从而验证了小波包变换和支持向量机相结合来识别金属薄板微弱损伤信号识别分类的准确性和有效性。
In response to the problem of internal defects such as delamination,holes and cracks in thin metal plates during process of processing and forming,lamb waves are used to detect damage in thin metal plates of steel,and a weak damage signal identification method based on wavelet packet transform and support vector machine is proposed for thin metal plates.The wavelet packet transform is used to extract and filter the features of the received echo signal.The echo signal is pre-processed by wavelet packet threshold noise reduction,which constructs 25 feature values from three perspectives of time domain,frequency domain and time-frequency domain.Then,the damage feature vector is extracted by principal component analysis and the sample data set is in⁃put to the particle swarm optimized support vector machine recognition model for damage identification.The results show that the model classifies no damage,hole,slot type,and welds joint signal damage types,and the recognition accuracy reaches 91.67%.The SVM recognition model has better classification performance compared with the BP neural network method for classification,thus verifying the accuracy and effectiveness of combining wavelet packet transform and support vector machine to identify weak damage signal recognition classification of thin metal plates.
作者
严宏鑫
朱平
孙旺
YAN Hongxin;ZHU Ping;SUN Wang(School of Instrument and Electronics,North University of China,Taiyuan 030051)
出处
《舰船电子工程》
2023年第3期167-173,共7页
Ship Electronic Engineering
基金
装发快速扶持项目第二阶段(编号:61409220157)资助。
关键词
小波包分解
支持向量机
粒子群优化
特征向量
金属薄板
损伤识别
wavelet packet decomposition
particle swarm optimization
support vector machine
eigenvectors
sheet metal
damage identification
