基于时序分析的拉伸件成型裂纹的特征参数提取  

Extraction of characteristic parameters of acoustic emission signals for molded metal drawing parts based on time-series analysis and MATLAB

作　　者：骆志高[1] 叶红英[1] 胥爱成[1] 

机构地区：[1]江苏大学机械工程学院,江苏镇江212013

出　　处：《振动与冲击》2012年第17期120-123,144,共5页Journal of Vibration and Shock

基　　金：科技型中小企业创新基金(09C26213201011)

摘　　要：研究金属拉伸制件声发射信号特征参数的提取对判断成型制件的质量尤为重要。研究采用时序分析和MATLAB对经过小波包分解的声发射信号进行特征参数的提取。在此过程中利用时间序列的自相关系数和偏相关系数的拖尾性及截尾性来判断模型的类型;采用FPE(Final Prediction Error,最终预测误差0029准则对模型进行定阶;以最小二乘估计法对模型的参数进行估计。最后,根据所建立的自回归谱模型提取声发射信号的特征参数。研究结果是通过此方法成功提取到了金属拉抻制件声发射信号的特征参数及其分布图,为成型制件质量的判断提供了有利的依据。Extraction of characteristic parameters of acoustic emission signals for metal drawing parts is particularly important to judge the quality of molded parts. Time-series analysis and MATLAB were adopted to process the acoustic emission signals decomposed with the wavelet packet in order to extract their characteristic parameters. In the process, the type of the model was judged by using the trailing property and the truncated property of time series; the order of the model was chosen by using FPE （final prediction error） rule; the model parameters were estimated with the least-square estimation method. Finally, the characteristic parameters of the acoustic emission signal were extracted based on the established auto-regressive spectrum model. The characteristic parameters of the acoustic emission signals for molded metal drawing parts were extracted successfully by means of this method, and its profile provided a basis for judging the quality of molded parts.

关 键 词：声发射 特征参数 时序分析 MATLAB 

分 类 号：TB535[理学—物理] TG386.4[理学—声学]