Full-Line Idler Fault Monitoring in Belt Conveyors via UWFBG-DAS and Characteristic Energy Feature Analysis.

Reliable full-line monitoring of belt-conveyor idlers remains challenging because large numbers of idlers operate under spatially varying structural stiffness and strong industrial vibration. This study develops an ultra-weak fiber Bragg grating distributed acoustic sensing (UWFBG-DAS) method combined with characteristic energy feature analysis for long-distance idler monitoring. The method makes three main contributions. First, a simplified finite-element model identifies the middle crossbeam as an effective vibration-transmission path and guides the deployment of the sensing array. Second, envelope demodulation and variational mode decomposition (VMD) are employed to isolate the fault-sensitive IMF2 component, whose energy is temporally accumulated and evaluated using a zone-specific self-referencing threshold derived from normal-operation data. Third, the method is validated through field deployment and fault-type classification. Approximately 1.2 km of a sensing cable was deployed in a coal-fired power plant, and identifiable characteristic-energy increases were observed in 9 of 10 idler-replacement tests. For three representative fault types, stratified five-fold cross-validation of 300 samples achieved an overall classification accuracy of 90.3%, with a 95% Wilson confidence interval of 86.5-93.2%. These results demonstrate the feasibility of UWFBG-DAS combined with zone-specific characteristic energy analysis for long-distance idler monitoring under spatially heterogeneous industrial conditions.