Research on Fault Detection Method Based on Audio Feature Clustering Algorithm

doi:10.3778/j.issn.1002-8331.2211-0432

Abstract

Abstract: Aiming at the urgent need for online intelligent detection of equipment faults in industrial production, this paper proposes a fault detection method on the strength of the joint optimization of audio feature deep convolutional neural network and clustering algorithm（FDM_DCNN-CA）. The key techniques such as feature location extraction and audio information fusion to clustering analysis in the 3D coordinate system are further elaborated in this paper. The FDM_DCNN-CA has been tested on MIMII, an audio dataset with different background noise, and the results show that the FDM_DCNN-CA achieves good results even on the “valve” equipment which is difficult to be detected. The “valve” equipment has an overall AUC score of more than 91% and is more robust to abnormal audio signals, with an accuracy rate of more than 95%.

Key words: fault detection, audio feature, unsupervised learning, three-dimensional k-means clustering, deep learning

摘要： 针对工业生产中设备故障在线智能检测的迫切需要，提出了一种基于音频特征深度卷积神经网络和聚类算法联合优化的故障检测方法（fault detection method on the strength of the joint optimization of audio feature deep convolutional neural network and clustering algorithm，FDM_DCNN-CA）。该方法详细阐述了三维坐标系中的特征位置归一化提取和音频信息融合聚类分析等关键技术要点。FDM_DCNN-CA在具有不同背景噪声的音频数据集MIMII上进行了测试，结果表明即使在检测难度较大的“阀门”设备上，仍然取得了良好应用结果。“阀门”设备总体AUC分数超过91%，并且在异常音频信号判断上具有更好的鲁棒性，正确率超过95%。

关键词: 故障检测, 音频特征, 无监督学习, 三维[k]-均值聚类, 深度学习

HUANG Yu, ZHOU Wei, GAN Xing, XIE Xiaoyan, HUANG Xinyu. Research on Fault Detection Method Based on Audio Feature Clustering Algorithm[J]. Computer Engineering and Applications, 2023, 59(15): 281-289.

黄羽, 周伟, 甘兴, 谢晓琰, 黄鑫宇. 基于音频特征聚类算法的故障检测方法研究[J]. 计算机工程与应用, 2023, 59(15): 281-289.

References

[1] LIM H，PARK J S，HAN Y.Rare sound event detection using 1D Convolutional recurrent neural networks[C]//Detection and Classification of Acoustic Scenes and Events，2017：80-84.
[2] PHAYE S S R，BENETOS E，WANG Y.Subspectralnet-using sub-spectrogram based convolutional neural networks for acoustic scene classification[C]//2019 IEEE International Conference on Acoustics，Speech and Signal Processing（ICASSP），2019：825-829.
[3] CHEN Y，WEI S，ZHANG Y.Classification of heart sounds based on the combination of the modified frequency wavelet transform and convolutional neural network[J].Medical & Biological Engineering & Computing，2020，58（9）：2039-2047.
[4] JUNG J，SHIM H，KIM J，et al.DCASENet：an integrated pretrained deep neural network for detecting and classifying acoustic scenes and events[C]//2021 IEEE International Conference on Acoustics，Speech and Signal Processing（ICASSP），2021：621-625.
[5] THOIDIS I，GIOUVANAKIS M，PAPANIKOLAOU G.Semi-supervised machine condition monitoring by learning deep discriminative audio features[J].Electronics，2021，10（20）：2471.
[6] YAMAGUCHI M，KOIZUMI Y，HARADA N.Adaflow：domain-adaptive density estimator with application to anomaly detection and unpaired cross-domain translation[C]//2019 IEEE International Conference on Acoustics，Speech and Signal Processing（ICASSP），2019：3647-3651.
[7] KOIZUMI Y，SAITO S，UEMATSU H，et al.Unsupervised detection of anomalous sound based on deep learning and the neyman-pearson lemma[J].IEEE/ACM Transactions on Audio，Speech，and Language Processing，2018，27（1）：212-224.
[8] MEZZA A I，HABETS E A P，MULLER M，et al.Unsupervised domain adaptation for acoustic scene classification using band-wise statistics matching[C]//2020 28th European Signal Processing Conference（EUSIPCO），2021：11-15.
[9] LI Y，LIU M，WANG W，et al.Acoustic scene clustering using joint optimization of deep embedding learning and clustering iteration[J].IEEE Transactions on Multimedia，2019，22（6）：1385-1394.
[10] LI S，GU Y，LUO Y，et al.Enhanced streaming based subspace clustering applied to acoustic scene data clustering[C]//2019 IEEE International Conference on Acoustics，Speech and Signal Processing（ICASSP），2019：11-15.
[11] WANG M，ZHANG X L，RAHARDJA S.An unsupervised deep learning system for acoustic scene analysis[J].Applied Sciences，2020，10（6）：2076.
[12] PUROHIT H，TANABE R，ICHIGE K，et al.MIMII dataset：sound dataset for malfunctioning industrial machine investigation and inspection[J].arXiv：1909.09347，2019.
[13] SIMONYAN K，ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[J].arXiv：1409.1556，2014.
[14] IOFFE S，SZEGEDY C.Batch normalization：accelerating deep network training by reducing internal covariate shift[C]//International Conference on Machine Learning，2015：448-456.
[15] DOHI K，IMOTO K，HARADA N，et al.Description and discussion on DCASE 2022 challenge task 2：unsupervised anomalous sound detection for machine condition monitoring applying domain generalization techniques[J].arXiv：2206.05876，2022.
[16] SANDLER M，HOWARD A，ZHU M，et al.Mobilenetv2：inverted residuals and linear bottlenecks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2018：4510-4520.