Fault identification for intelligent real-time monitoring via mobile accessible networks: a critical review of MEMS microphone-based sensing for bearing diagnostics

The main objective of Condition Monitoring (CM) is to prevent unexpected downtime by monitoring machine conditions and reducing maintenance costs. To prevent failure, particularly in rotating machinery, CM is performed utilizing sensing devices, which have been the focus of extensive research in recent decades. This paper presents a critical review and provides comprehensive coverage of recent advances in CM for rotating machinery that utilize MEMS microphone-based sensing, with particular emphasis on bearing fault diagnosis relevant to the Fault Identification for Intelligent Real-time Monitoring via Mobile Accessible Network research programme. It synthesises recent journal articles focusing on MEMS (Micro-Electro-Mechanical Systems) microphone technologies, signal processing and machine learning techniques, methodological study design and experimental framework, and reported comparative analysis, advantages, limitations, and future research directions. Furthermore, the study reveals a dominant trend toward the use of Artificial Intelligence (AI) techniques, resulting in significant advances in CM and an emerging on non-contact-based sensing via mobile accessible system and self-powered sensing. Key gaps identified include the under-representation of MEMS microphone-based sensing, noise robustness and high sensitivity sensors, insufficient edge-deployment validation, the barriers of large datasets processing to achieve effectiveness, and the absence of multi-domain adaptive diagnostic frameworks. As a final point, options for scalable, low-cost, acoustic-based sensing and edge-intelligent computing are proposed, to deliver real-time monitoring across multiple domains.