Publications / 2024 Proceedings of the 41st ISARC, Lille, France

Acoustic Signal-Based Excavator Fault Detection Using Deep Learning Method

Yuhan Zhou, Zhen Lin Foo, Yong Zhi Koh, Ker-Wei, Justin Yeoh
Pages 1136-1143 (2024 Proceedings of the 41st ISARC, Lille, France, ISBN 978-0-6458322-1-1, ISSN 2413-5844)
Abstract:

As machinery assumes a critical role in modern construction, particularly in Singapore’s development initiatives, maintaining excavators becomes paramount. Despite the prevalence of faults within this equipment, the scarcity of skilled mechanics compounds the challenge of timely diagnoses and maintenance. Leveraging deep learning methodologies, this research endeavors to analyze audio signals from excavators, aiming to identify distinctive patterns indicative of faults. Unlike existing studies primarily relying on vibration signals, this research focuses on audio signals for excavator fault prediction. Challenges involving ambient noise in construction sites and limitations in dataset size and imbalance compel the need for robust machine learning models capable of accurate fault diagnoses. The proposed methodology involves dataset collection, audio signal processing, feature extraction, and neural network training to differentiate normal operation from faulty conditions. This study delves into the application of machine learning and signal processing techniques to discern excavator conditions, aiming to classify their operational state as either faulty or operational. With an achieved 89.33% accuracy, 94.74% precision, and 85.71% recall, the method demonstrates promising performance. This research offers the potential to fortify excavator maintenance practices, potentially mitigating the impact of faults on construction productivity and costs.

Keywords: Excavator Maintenance, Deep Learning, Fault Detection, Audio Signal Processing