基于MFCC的汽车敲击异响识别

摘要
图/表
参考文献(17)
相关文章 (8)

全文: PDF (2363 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要现阶段，汽车异响的诊断主要依赖有经验的工程师进行主观评判，存在不准确、易错判、易漏判的问题。本文针对汽车敲击异响实测信号进行统计分析得到梅尔倒谱系数（Mel Frequency Cepstrum Coefficient, MFCC），并以此作为表征异响来源的特征向量，基于最大似然估计法构建其联合概率分布高斯混合模型（Gaussian Mixture Model, GMM），从而针对未知实测异响信号可利用该GMM模型进行似然判别。文章指出了说话人识别技术与敲击异响识别的不同之处即Mel三角滤波器个数和离散余弦变换（DCT）输出系数个数的选取方式，并对方法的可行性进行分析，最后实验加以验证。结果显示此方法的识别率达100%以上，拒绝率达100%以上，为汽车异响的客观评价方法打下基础。
关键词：说话人识别；敲击异响；梅尔倒谱系数；高斯混合模型

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	黄凯1
	郑瑶辰2
	3
	邓兆祥1
	2
	3

关键词 ：说话人识别, 敲击异响, 梅尔倒谱系数, 高斯混合模型

Abstract：At this stage, the diagnosis of vehicle squeak & rattle mainly relies on the subjective judgment of experienced engineers, and there are problems of inaccuracy, easy to make mistakes, and easy to miss. This paper conducts statistical analysis on the measured signals of abnormal sound from vehicle’s rattle noises to obtain Mel Frequency Cepstrum Coefficient (MFCC), and uses this as the feature vectors to characterize the source of abnormal sound, and constructs its joint probability distribution based on the maximum likelihood estimation method Gaussian Mixture Model (GMM), so that the GMM model can be used for likelihood discrimination for unknown measured rattle noise signals. The article points out the difference between speaker recognition technology and rattle noise recognition, namely the selection of the number of Mel triangle filters and the number of discrete cosine transform (DCT) output coefficients, and the feasibility of the method Analyze, and finally verify by experiment. The results show that the recognition rate of this method is more than100%, and the rejection rate is more than100%, which lays the foundation for the objective evaluation method of vehicle squeak & rattle.
Keywords: Speaker recognition; Rattle; Mel Frequency cepstrum coefficient; Gaussian mixture model.

Key words： Speaker recognition Rattle Mel Frequency cepstrum coefficient Gaussian mixture model.

收稿日期: 2020-09-16 出版日期: 2022-07-15

引用本文:

黄凯1，郑瑶辰2,3，邓兆祥1,2,3. 基于MFCC的汽车敲击异响识别[J]. 振动与冲击, 2022, 41(13): 275-282.
HUANG Kai1, ZHENG Yaochen2,3, DENG Zhaoxiang1,2,3. Recognition of vehicle’s rattle based on MFCCs. JOURNAL OF VIBRATION AND SHOCK, 2022, 41(13): 275-282.

链接本文:

http://jvs.sjtu.edu.cn/CN/ 或 http://jvs.sjtu.edu.cn/CN/Y2022/V41/I13/275

[1] Unnikrishnan, Kuttan, Chandrika, et al. Development of an algorithm for automatic detection and rating of squeak and rattle events[J]. Journal of Sound & Vibration, 2010.
[2] Huang H B, Li R X, Huang X R, et al. Identification of vehicle suspension shock absorber squeak and rattle noise based on wavelet packet transforms and a genetic algorithm-support vector machine[J]. Applied Acoustics, 2016, 113(dec.):137-148.
[3] Srinivasan A. Speaker Identification and Verification using Vector Quantization and Mel Frequency Cepstral Coefficients[J]. Research Journal of Applied Sciences Engineering & Technology, 2012, 4(1):33-40.
[4] Mukherjee, H., et al., A lazy learning-based language identification from speech using MFCC-2 features. International Journal of Machine Learning and Cybernetics, 2020. 11(1): p. 1-14.
[5] Nogueira, D.M., et al., Classifying Heart Sounds Using Images of Motifs, MFCC and Temporal Features. Journal of Medical Systems, 2019. 43(6): p. 168.
[6] Upadhya, S.S., A.N. Cheeran and J.H. Nirmal, Discriminating Parkinson diseased and healthy people using modified MFCC filter bank approach. International Journal of Speech Technology, 2019. 22(4): p. 1021-1029.
[7] 刘思思，谭建平，易子馗. 基于MFCC和SVM的车窗电机异常噪声辨识方法研究[J]. 振动与冲击, 2017, 36(5): 102-107.
LIU Sisi,TAN, Jianping, YI Zikui. A window motor abnormal noise identification method based on MFCC and SVM. JOURNAL OF VIBRATION AND SHOCK, 2017, 36(5): 102-107.
[8] 白玲玲等, 隐马尔可夫模型在恶意域名检测中的应用. 计算机工程, 2019. 45(09): 161-168.
Lingling B, Zhenhu N, Fei X, et al. Application of Hidden Markov Model in Malicious Domain Name Detection[J]. computer engineering, 2019.
[9] 王昌海等, 基于加权观测的隐马尔可夫模型. 模式识别与人工智能, 2019. 32(06): 515-523.
Changhai W, Zhehui L I, Bo W, et al. An Improved Hidden Markov Model Based on Weighted Observation[J]. Pattern Recognition and Artificial Intelligence, 2019.
[10] Antony A, Gopikakumari R. Speaker identification based on combination of MFCC and UMRT based features[J]. Procedia Computer Science, 2018, 143:250-257.
[11] 常飞, 乔欣, 张申,等. 基于MFCC特征提取的故障预测与评价方法[J]. 计算机应用研究, 2015, 032(006):1716-1719.
Fei C, Xin Q, Shen Z, et al. Method of failure prediction and evaluation based on MFCC feature extraction[J]. application research of computers, 2015.
[12] Mallouh, A.A., Z. Qawaqneh and B.D. Barkana, New transformed features generated by deep bottleneck extractor and a GMM–UBM classifier for speaker age and gender classification. Neural Computing and Applications, 2018. 30(8): p. 2581-2593.
[13] Peng, K., V.C.M. Leung and Q. Huang, Clustering Approach Based on Mini Batch Kmeans for Intrusion Detection System Over Big Data. IEEE Access, 2018. 6; 6: p. 11897-11906.
[14] Li F, Li K T, Lu K, et al. Random noise suppression and parameter estimation for Magnetic Resonance Sounding signal based on maximum likelihood estimation[J]. Journal of Applied Geophysics, 176.
[15] Reynolds D A, Rose R C. Robust text-independent speaker identification using Gaussian mixture speaker models[J]. IEEE Transactions on Speech & Audio Processing, 1995, 3(1):72-83.
[16] 曹辉, 徐晨, 赵晓,等. 说话人识别中的Mel特征频率倒谱系数[J]. 西北大学学报(自然科学版), 2013, 043(002):203-208.
Hui C, Chen X U, Xiao Z, et al. The Mel-frequency cepstral coefficients in speaker recognition[J]. journal of northwest university(natural science edition), 2013.
[17] 冯辉宗, 王芸芳. 语谱特征的身份认证向量识别方法[J]. 重庆大学学报, 2017(5).
Feng H, Wang Y. An i-vector speaker recognition method based on spectrogram[J]. Journal of Chongqing University, 2017, 40(5):88-94.