A Hybrid Bayesian–Weibull–Logistic (HBWL) Risk Framework for Decision-Oriented Predictive Maintenance in Critical Infrastructure Systems

Critical infrastructure systems rely on interconnected systems such as HVAC, fire safety, elevators, and energy management units, where failures can significantly impact safety, efficiency, and operational continuity. Traditional maintenance strategies, including reactive and time-based preventive approaches, are often inadequate in addressing the dynamic nature of these environments. This study presents a predictive modeling framework for a Hybrid, by integrating Weibull, Bayesian, and Logistic (HBWL) approaches to support critical infrastructure systems for next-generation maintenance strategies. Weibull analysis is employed to characterize time dependent failure behavior of building systems, while logistic regression models the probability of fault occurrence based on operational and environmental factors. Bayesian inference is used to continuously update reliability estimates as new data becomes available, enabling adaptive and data-driven decision-making. Results indicate that while standalone logistic regression exhibits poor predictive performance (AUC ≈ 0.506), the integration of probabilistic and reliability-based models significantly enhances decision support. The hybridization of these methods enhances prediction accuracy, reduces uncertainty, and improves maintenance planning compared to standalone models. The proposed approach supports proactive maintenance scheduling, minimizes system downtime, and optimizes resource allocation, making it highly suitable for critical infrastructure systems and intelligent, sensor-driven smart building environments.