AI-driven intelligent vibration monitoring for predictive maintenance in chemical process plants: Signal processing, diagnostics, and industrial integration
Intelligent vibration monitoring is increasingly important for predictive maintenance (PdM) in chemical process plants, where machinery faults may appear as weak, mixed, nonstationary, or process-dependent signatures. This review examines how vibration measurements can be converted into reliable maintenance intelligence using signal processing, artificial intelligence (AI), and plant-level integration. A structured critical narrative review was conducted using literature identified from Scopus, Web of Science, ScienceDirect, SpringerLink, and IEEE Xplore, focusing mainly on studies published from 2016 to September 2026. It compares fault signature complexity across pumps, compressors, agitators, turbines, fans, gear-driven systems, and rolling-element bearings and evaluates time- and frequency-domain analysis, wavelet methods, empirical mode decomposition, cepstral enhancement, and time–frequency approaches. AI-based classification, anomaly detection, deep learning, and prognostic modelling were assessed in relation to interpretability, generalisation, benchmark data limitations, uncertainty, and industrial trust. Particular emphasis is placed on process-aware interpretation because vibration responses can be affected by the flow, pressure, hydraulic instability, operating duty, structural transmission, and maintenance constraints. This review proposes a deployment-oriented, tiered PdM framework that aligns analytical depth, AI support, asset criticality, process context, and maintenance workflow integration. A standards-informed distinction between condition monitoring, diagnostic advisories, and machinery protection functions was also introduced. Future studies should prioritise explainability, field robustness, cross-condition validation, and traceable linkages between diagnostic outputs, plant information systems, and maintenance decisions.
Authors
- Jan Lean Tai (ORCID: https://orcid.org/0000-0003-0841-4446)
- Mohammad Sohel (ORCID: https://orcid.org/0009-0005-5738-1144)
Institutions
- Universiti Putra Malaysia (MY)
- Central Institute of Technology (AU)
Publication Details
- Journal
- Journal of AI-Driven Engineering Innovation and Applications
- Published
- 2026-11-13
- DOI
- https://doi.org/10.67014/jadeia.v1i1.0049
- Primary Topic
- Machine Fault Diagnosis Techniques
- Type
- article
- Field-Weighted Citation Impact
- 0.00