MIG焊熔透电弧声小波包频带能量特征提取  

Feature extraction of penetration arc sound in MIG welding via wavelet packet frequency-band energy

作　　者：刘立君[1,2] 兰虎[2] 温建力[2] 于忠伟[2] 

机构地区：[1]浙江大学宁波理工学院,浙江宁波315100 [2]哈尔滨理工大学材料科学与工程学院,哈尔滨150080

出　　处：《焊接学报》2010年第1期45-49,共5页Transactions of The China Welding Institution

基　　金：宁波自然科学基金项目(2008A610031);黑龙江省自然科学基金项目(E2007-01);黑龙江省青年骨干教师基金项目(1153G009);哈尔滨市科技创新基金项目(2007RFQXG055)

摘　　要：焊缝熔透状态的监控对焊接过程质量控制和自动化生产的实现具有重要意义,为此建立了M IG焊熔透电弧声试验系统.通过分析射滴与射流过渡电弧声频谱特征,设计了一套针对电弧声频带能量特征提取的流程.首先对焊接过程拾取的电弧声进行小波降噪;其次采用小波包移频算法进行电弧声频带提取,消除了经典小波包迭代算法由于小波包分解过程中的隔点采样而产生的频率混叠现象;然后计算频带能量并构造其特征向量.结果表明,所提取的频带能量特征能较好的反映焊缝熔透状态,为后续基于电弧声的熔透诊断奠定基础.A penetration arc sound experimental system in MIG welding was developed.By analyzing the characteristics of the spectrum of arc sound signal with projected transfer and spray transfer,the procedure of arc sound frequency-band energy extration was designed.Firstly,wavelet transfer is utilized in elimanating noises of arc sound in welding process.Secondly,wavelet packet frequency-shift arithmetic is used to extract arc sound frequency-band energy,which eliminates the frequency aliasing phenomenon caused by the interval sampling in the classical wavelet packet decomposition.The last,frequency-band energy is calculated and the characteristic vector is constructed.The experiments show that the extracted frequency-band characteristics can better reflect the weld penetration.

关 键 词：熔透 电弧声 小波变换 频带能量 特征提取 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]