基于小波包变换的复合材料拉挤多腔板振动检测

Vibration Detection of Composite Pultrusion Multicavity Plates Based on Wavelet Packet Transformation

复合材料拉挤多腔板在线生产时有时会产生缺陷,导致生产线故障甚至全面停车,因此需要对在线生产的复合材料进行实时无损检测。为此,采用敲击测试的方法,获取振动信号,利用小波包分析对敲击振动信号进行多层分解,分解后的频带能量作为特征向量。把经过归一化处理的特征向量用支持向量机(SVM)进行分类识别。经过实验验证,识别率达到100%。这表明使用小波包分析和支持向量机结合的方法可以对复合材料拉挤多腔板进行故障诊断,在实际应用中具有较好的前景。

It is easy to produce defects in the online production of composite pultrusion multicavity plates,which may lead to the failure or even complete shutdown of the production line. Therefore,it is necessary to implement real-time nondestructive testing of composite materials produced online. For this purpose,the percussion vibration signals are obtained by means of percussion test,the multilayer decomposition of percussion vibration signals are carried out by wavelet packet analysis,and the decomposed frequency band energy is used as the characteristic vector. The normalized eigenvectors are diagnosed by support vector machine(SVM). After the experimental verification,the diagnosis rate reaches 100%. This shows that the method of wavelet packet analysis and support vector machine can be suitable for fault diagnosis of composite pultrusion multicavity plates.