Performance Comparison of One-Class Classifiers in Faulty Sanding Machine Detection Using Sound
Haque, MD. Zubairul (2023)
Haque, MD. Zubairul
2023
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2023060923347
https://urn.fi/URN:NBN:fi:amk-2023060923347
Tiivistelmä
Nowadays, product quality is a paramount concern in the modern machine manufacturing industry. Millions of dollars are being spent by modern machine manufacturing industries to maintain product quality. This expenditure on quality control can be reduced, and the efficiency of quality control can be enhanced, by reducing human contact and utilizing technology in quality control. Furthermore, automatic mechanical failure detection is an important technology in the fourth industrial revolution. Sound from the machine is used as the primary way to detect mechanical faults by quality control technicians. The concept behind this is that faulty machines must sound different from normal machines. After converting the sound data into a suitable digital format, machine learning algorithms can be used to identify abnormal sounds.
In this thesis, we have tried several machine learning algorithms to find the most suitable one for anomalous sound detection. Additionally, we have discussed the working principle and optimization process of these machine learning algorithms. To provide a reliable comparison between these machine learning algorithms, the anomaly scores for the same data were shown. Our aim was to provide a comprehensive understanding of the performance of these anomaly detection algorithms by analyzing the accuracy of detecting faulty machines.
In this thesis, we have tried several machine learning algorithms to find the most suitable one for anomalous sound detection. Additionally, we have discussed the working principle and optimization process of these machine learning algorithms. To provide a reliable comparison between these machine learning algorithms, the anomaly scores for the same data were shown. Our aim was to provide a comprehensive understanding of the performance of these anomaly detection algorithms by analyzing the accuracy of detecting faulty machines.