This research introduces and validates a new class of amplitude-correlation-based diagnostic technologies for the early detection of incipient mechanical faults in industrial conveyor systems. The methods developed the Cross-Correlation of Spectral Moduli (CCSM₄ and CCSM₃) and the Inverse Non-Linearity Feature (INLF) extend the capabilities of Motor Current Signature Analysis (MCSA) beyond traditional second order and phase-dependent higher-order spectral (HOS) techniques. By focusing on amplitude interactions rather than phase coherence, the proposed framework achieves superior sensitivity, robustness to noise, and stability under variable load and grid conditions. A comprehensive investigation combining theoretical modelling, numerical simulation, and experimental validation was conducted. The CCSM₄ method achieved Total Probability of Correct Diagnosis (TPCD) values above 95–100% in both simulated and experimental analyses, reliably identifying weak nonlinear amplitude couplings associated with early bearing degradation. The third-order CCSM₃ demonstrated 100% accuracy in detecting belt mis-tracking faults, outperforming bicoherence and tricoherence techniques by maintaining phase independence and resilience to grid-frequency instability. Complementing these, the INLF, an amplitude-domain metric derived from harmonic ratios, achieved complete diagnostic accuracy across all tested load conditions, confirming its effectiveness in quantifying friction-induced nonlinearity with minimal computational cost. Experimental testing on a purpose-built conveyor rig and full-scale airport conveyor installations validated the technologies under realistic operational scenarios, achieving Technology Readiness Level 7 (TRL 7). Collectively, these results confirm that amplitude-correlation-based diagnostics provide a noise-robust, load-insensitive, and scalable solution for real-time, non-intrusive incipient fault detection. The research establishes a foundation for advanced predictive-maintenance systems capable of enhancing reliability, energy efficiency, and operational safety in next-generation industrial monitoring frameworks.
| Date of Award | 11 Dec 2025 |
|---|
| Original language | English |
|---|
| Supervisor | Len Gelman (Main Supervisor) & Andrew Ball (Co-Supervisor) |
|---|