an integrated fault tree analysis and bayesian network fta-bn framework for predictive maintenance of ice-powered drilling machines

The frequent breakdown of internal combustion engine (ice)–powered drilling machines and other laboratory equipment in our higher institutions and field engineering environments is largely due to aging equipment, lack of spare parts, and absence of documentation for discontinued machines. these challenges reduce maintenance effectiveness and extend downtime. this study presents an integrated predictive maintenance framework that combines fault tree analysis (fta), analytic hierarchy process (ahp), and bayesian network (bn) inference to address these issues. unlike traditional fta-based reliability approaches, the proposed framework supports dynamic updating of component failure probabilities using structured expert judgement and real-world diagnostic evidence. expert assessments were weighted using ahp to construct unbiased prior failure probabilities mapped into a bn structure. diagnostic field data—vibration, sound, and exhaust emissions—were collected to validate and update the model. results show that carburetor faults, piston ring wear, and fuel line blockages are the dominant contributors to failure. dynamic bn inference improved diagnostic accuracy, while the maintenance strategy derived from model outputs increased mean time between failures (mtbf) by approximately 25% and reduced unplanned downtime by about 30%. the proposed framework offers a practical, low-cost predictive maintenance solution for legacy equipment in resource‑constrained environments.