[ Special Issue : Engineering Design of ADBL ]

Journal of the Korean Society of Manufacturing Technology Engineers - Vol. 29, No. 3, pp.210-215

ISSN: 2508-5107 (Online)

Print publication date 15 Jun 2020

Received 13 Apr 2020 Revised 11 May 2020 Accepted 20 May 2020

DOI: https://doi.org/10.7735/ksmte.2020.29.3.210

신경망을 이용한 음향 측정 기반의 고장진단 시스템

남지인^a ; 박희재^a^{, *}

A Neural Network based Fault Detection and Classification System Using Acoustic Measurement

Jiin Nam^a ; Hee Jae Park^a^{, *}

aDepartment of Mechanical Design and Robot Engineering, Seoul National University of Science and Technology

Correspondence to: ^*Tel.: +82-2-970-6341 E-mail address: looki@seoultech.ac.kr (Hee Jae Park).

Abstract

In this study, a fault detection and classification method using neural network-based acoustic measurement is proposed. In this method, a measured acoustic signal of the target equipment undergoes Fast Fourier transformation. The magnitude, for a range of frequencies, is accumulated and normalized to train predefined neural network model. To validate the proposed method, an experimental setup for cooling fan is established. The faults of the device are classified into five categories. A series of experiments for the experimental setup are conducted to validate the performance of the fault detection and classification of the proposed method. An accuracy of up to 98.6% is obtained for the test data. Thus, the experimental results show the effectiveness of the proposed fault detection algorithm.

Keywords:

Artificial intelligence, Machine learning, Neural network, Fault detection, Fast fourier transform

Acknowledgments

이 연구는 서울과학기술대학교 교내연구비의 지원으로 수행되었습니다.

References

Baek, S. J., Kim, D. Y., 2016, Fault Symptom Extraction of Automotive Engines by the Event Occurrence Analysis Over Time interval, Proceedings of The Korean Institute of Industrial Engineers Fall Conference, 1324-1328.
Jang, J. H., Ha, C. K., Chu, B. S., Park, J. Y., 2016, Development of Fault Diagnosis Technology Based on Spectrum Analysis of Acceleration Signal for Paper Cup Forming Machine, Journal of the Korean Society of Manufacturing Process Engineers, 15:6 1-8. [https://doi.org/10.14775/ksmpe.2016.15.6.001]
Ko, J. U., Jung, J. H., Kim, M. Y., Kong, H. B., Youn, B. D., 2018, Noise Robust Fault Diagnosis Technique to Simultaneously Learn Classification and Denoising, Proceedings of The Korean Society of Mechanical Engineers Conference, 165-167.
Ham, S. J., Han, S. Y., Kim, S. G., Park, H. J., Park, K. J., Choi, J. H., 2019, Fault Diagnosis Study of Train Door Using Motor Current Signal, Journal of Applied Reliability, 19:4 353-361. [https://doi.org/10.33162/JAR.2019.12.19.4.353]
Chollet, F., 2017, Deep Learning with Python, Manning Publications, New York.
Vasilev, I., 2019, Advanced Deep Learning With Python: Design and Implement Advanced Next-generation AI Solutions Using TensorFlow and PyTorch, Packt Publishing Ltd, Birmingham.
Potdar, K., Pardawala, T. S., Pai, C. D., 2017, A Comparative Study of Categorical Variable Encoding Techniques for Neural Network Classifiers, International Journal of Computer Applications, 175:4 7-9. [https://doi.org/10.5120/ijca2017915495]
Rahmani, M., Atia, G. K., 2018, Data Dropout in Arbitrary Basis for Deep Network Regularization, Proceedings of 52nd Asilomar Conference on Signals, Systems, and Computers, 66-70. [https://doi.org/10.1109/ACSSC.2018.8645094]

Jiin Nam

Student in the Department of Mechanical Design and Robot Engineering, Seoul National University of Science and Technology.Her research interest is Machine Learning.

E-mail: jiin@seoultech.ac.kr

Hee Jae Park

Professor in the Department of Mechanical Design and Robot Engineering, Seoul National University of Science and Technology. His research interest is Mechatronics and Machine Learning.

E-mail: looki@seoultech.ac.kr