中文摘要：

介绍了一种基于隐马尔可夫模型的轴承故障音频信号诊断方法。通过对轴承音频信号的Mel频率倒谱系数特征提取，分别采用离散HMM和连续高斯混合密度HMM两种方法进行建模与诊断研究。与CGHMM方法相比，DHMM方法运算速度快，但诊断精度低。而从总体上来看，两种方法都具有运算速度快，诊断精度高的优点。结果表明，本文方法具有很好的应用前景。

英文摘要：

A bearing fault diagnosis scheme based on Hidden Markov Model （HMM） of acoustic signals is introduced in this paper. By abstracting Mel Frequency Cepstrum Coefficients （MFCC） from acoustic signals emitted by bearing, modeling and diagnosing are studied with DHMM and CGHMM distinctly. Compared with CGHMM, it has faster speed, but lower diagnosis precision by DHMM. Both DHMM and CGHMM method have the advantages of fast speed and high diagnosis precision in total. Result demonstrate that the presented method has a great prospect.