Current Issues

Korea Society for Naval Science and Technology - Vol. 6 , No. 4

[ Article ]
Journal of the KNST - Vol. 5, No. 1, pp. 16-19
Abbreviation: KNST
ISSN: 2635-4926 (Print)
Print publication date 31 Mar 2022
Received 26 Jan 2022 Revised 23 Feb 2022 Accepted 21 Mar 2022
DOI: https://doi.org/10.31818/JKNST.2022.03.5.1.16

소음기반 엔진고장 진단 기계학습 기법 연구
김성현*
한국선급 함정기술팀 책임검사원

A Study of Machine Learning Technique for Noise-based Engine Fault Diagnosis
Sung-Hyun Kim*
Senior Surveyor, Naval Vessel Technology Team, Korean Register
Correspondence to : *Sung-Hyun Kim Naval Technology Team, Korean Register, 36, Myeongji Ocean City 9-ro, Gangseo-gu, Busan 46762, Republic of Korea Tel: +82-70-8799-8854 Fax: +82-70-8799-8879 E-mail: kshine@krs.co.kr


© 2022 Korea Society for Naval Science & Technology

초록

본 연구는 소음기반 엔진고장 진단을 위한 기계학습 기법 연구로서 원본 학습데이터를 MFCC 변환 형상으로 대체 적용하는 경우 기계학습 모델 3종에서 고장감별 정확도가 크게 향상되는 것을 고찰하였다. 또한 고장 소음의 패턴추출에 집중할 수 있도록 수리 후 정상소음과 혼합 학습하는 전략이 유효하게 작동함을 확인하였으며, 향후 MFCC 형상변환 과정의 권장 보완을 통해 고장감별의 편향성의 개선과 심층강화 학습과 연계한 후속연구를 제안한다.

Abstract

As a study of machine learning technique for noise-based engine failure diagnosis, it is considered that the discrimination accuracy was greatly improved in the three machine learning models when the learning data was replaced with the transformed MFCC feature. In addition, it was confirmed that the strategy of mixing learning with normal noise after repair works effectively so that we can focus on the pattern extraction of faulty noise. In the following study, we propose to improve the bias of failure detection through the recommended supplementation of the MFCC feature extraction process, and to couple with deep reinforcement learning.


Keywords: Condition-based Maintenance, Noise-based, Engine Fault Diagnosis, Machine Learning, MFCC(Mel Frequency Cepstrum Coefficient), LR(Logistic Regression), SVM(Support Vector Machine), DTC(Decision Tree Classification)
키워드: 상태기반유지관리, 소음기반, 엔진 고장진단 기법, 기계학습

References
He X, Xu S. Artificial neural networks[J]. Process Neural Networks: Theory and Applications, 2010:20-42
McFee, Brian et al. “Librosa: Audio and music signal analysis in python.” In Proceedings of the 14th python in science conference, pp. 18-25. 2015.
Scikit-learn: Machine Learning in Python, Pedregosa et al., JMLR 12, pp. 2825-2830, 2011.
Y. S. Wang, Q. H. Ma, Q. Zhu, L. Zhao, “An intelligent approach for engine fault diagnosis based on hibert-huang transform and support vector machine,” Applied Acoustics, Vol. 75, pp. 1-9, 2014.
Dai W, Ji W. A mapreduce implementation of C4. 5 decision tree algorithm[J]. International journal of database theory and application, 2014, 7(1): 49-60