基于时频分析的扬声器异音故障诊断方法  

Loudspeaker Fault Diagnosis Method Based on Time-frequency Transform of Acoustic Signal

作　　者：于德敏[1] 李小明[1] 许增朴[1] 

机构地区：[1]天津科技大学机械工程学院,天津300222

出　　处：《电声技术》2007年第11期24-27,共4页Audio Engineering

基　　金：天津市科技攻关培育项目(06YFGPGX08900)

摘　　要：提出了基于声信号时频分析的扬声器异音故障诊断方法,利用短时傅里叶变换和小波包变换2种时频分析方法将扬声器一维响应信号转化为二维时频图像,并利用3种距离方法对不同类型的扬声器时频图像进行了分析。结果表明,2种时频分析方法可充分反映故障的不同类型,3种距离方法可快速识别扬声器是否合格。A loudspeaker fault diagnosis method based on time-frequency transform of acoustic signal is presented. STFT（Short Time Fourier Transform） and wavelet packet transform are employed to convert the response signal into image signal, and three distance methods are utilized to compare the images of different loudspeakers. The experimental results show that the two time-frequency methods can reflect different kinds of loudspeakers adequately, and the three distance methods can identify the quality of loudspeakers quickly.

关 键 词：扬声器 短时傅里叶变换 小波包变换 图像处理 故障诊断 

分 类 号：TB52[理学—物理]