A Monte Carlo-Based Probabilistic Framework for Decision Risk in ISO 13374-Based CBM Systems under Information Uncertainty

Condition-Based Maintenance (CBM) has been widely adopted in railway systems to improve maintenance efficiency and operational reliability. The ISO 13374 standard provides a structured information-processing architecture for CBM implementation; however, uncertainty originating from data acquisition and prognostic inference can propagate through the decision process and affect maintenance outcomes. This study proposes a probabilistic framework for quantitatively evaluating maintenance decision risk under information uncertainty within ISO 13374-based CBM systems. The framework interprets the ISO 13374 architecture as a stochastic information-processing framework while explicitly modeling uncertainty originating from the Data Acquisition (DA) and Prognostics Assessment (PA) layers. A Gamma process is employed to represent stochastic degradation, and a particle filter is used for Bayesian state estimation and Remaining Useful Life (RUL) prediction. Maintenance decision performance is evaluated using decision accuracy and a cost-based risk metric through Monte Carlo simulation. The results show that uncertainty introduced at the DA layer propagates through prognostic inference and degrades maintenance decision performance, while Bayesian state updating progressively reduces RUL uncertainty as information accumulates. A regime-based analysis further identifies fragile, adaptive, and robust decision performance regions, demonstrating that maintenance performance depends on the interaction between information uncertainty and decision policy. The proposed methodology provides a generalized framework for quantitatively analyzing uncertainty propagation and its influence on maintenance decision risk within the ISO 13374 architecture, while the quantitative uncertainty analysis is intentionally focused on the DA and PA layers.