A Data-Driven Approach to Aircraft Engine MRO Using Enhanced ANNs Based on FMECA

Abstract

Aircraft engine MRO is essential for safe, reliable, and cost-effective aviation operations. Traditional maintenance methods, such as scheduled and condition-based maintenance, often result in excessive downtime, higher costs, and inefficient resource use. AI-driven predictive maintenance, combined with Reliability Engineering, enhances efficiency but typically lacks integration with systematic reliability assessment frameworks, limiting its ability to prioritize critical failures. This study introduces a hybrid predictive maintenance framework integrating artificial neural networks (ANN) with failure modes, effects, and criticality analysis (FMECA). Historical engine sensor data (temperature, pressure, vibration, and oil analysis) trains an ANN that predicts failure probabilities, repair durations, and costs. FMECA, utilizing the Risk Priority Number (RPN), ranks failures by severity, ensuring that the most critical issues are addressed first Weibull distribution analysis models component reliability, confirming wear-out failure modes, and supporting scheduled predictive maintenance. Validation with real aircraft engine data demonstrates the effectiveness of the ANN-FMECA model, achieving 94.3% accuracy in failure prediction and surpassing conventional methods. Maintenance prioritization efficiency improves by 15.7%, reducing maintenance costs by 35.3% and unplanned outages by 40.5%. This enhances fleet availability, improves flight safety, and reduces environmental impact.