BROWSE
Detail
Conference Paper
An End-to-End Predictive Maintenance System for Autonomous Inspection based on Health Index
Cited - time in 
Share 
Share
- Authors
- Ho-Min Park, Dongkoo Shon, Yeonggwang Oh, Tae Hyun Yoon, Woo-Sung Jung, Jeong-Ho Park, Dae Seung Yoo
- Issue Date
- 2026-02
- Citation
- International Conference on Advanced Communications Technology (ICACT) 2026, pp.141-149
- ISSN
- 1738-9445
- Publisher
- IEEE
- Language
- English
- Type
- Conference Paper
- Abstract
- This paper proposes WFSPM, an end-to-end predictive maintenance system for assessing generator stator wedge fastener strength, centered on a Mahalanobis distance (MD) based health index (HI). The system implements a standards-compliant pipeline—data flow, feature extraction, decision rules, and UI/reporting—and maps the MD-HI to a 0–100 health scale for field interpretability. It exhibits strong static classification performance and robustness under injected synthetic noise. Scenario-based prognostics and health management (PHM) key performance indicators (KPI) and lead-time analyses further demonstrate a practical operate cycle of early warning, action and recovery. Overall, WFSPM provides an explainable and robust solution, with operating policies that can be tuned to field constraints.
- Keyword
- Predictive Maintenance, Mahalanobis Distance, Health Index, Acoustic Signal Processing
- KSP Keywords
- Acoustic signal processing, Classification Performance, Data Flow, Decision rules, Early Warning, End to End(E2E), Feature extractioN, Health Index, Key Performance Indicators(KPI), Lead Time, Predictive maintenance