INTEGRATED AI-DRIVEN PREDICTIVE AIRCRAFT MAINTENANCE SYSTEM

Abstract Modern aircraft maintenance has requirements for smart systems, which are capable of predicting failures, detecting structural defects and estimating the lifespan of components in real time. This paper introduces an integrated AI embedded-architecture for aircraft maintenance using all-in-one predictive analysis, real-time crack detection and battery RUL estimation. The proposed methodology is based on Random Forest and Support Vector Regression models to predict the degradation of components from sensor data, and a YOLO-based deep learning model for detection of cracks in real-time from high-resolution images of inspection of the aircraft. Also, a Long Short-Term Memory (LSTM) network is used to model the battery degradation and accurately predict RUL with the help of time series parameters of battery operation such as voltage, temperature, and charge cycle. Experimental evaluation shows improvement of fitness, detection of fault, cases of maintenance failures and downtime, and operational safety in comparison with traditional maintenance strategies. The proposed framework provides a way of proactive maintenance planning and also a scalable solution for the next generation intelligent health monitoring systems for aircrafts. Keywords Artificial Intelligence; predictive maintenance; aircraft structural health monitoring; crack detection; harmless remaining useful life of batteries; deep learning; YOLO model; LSTM; Random Forest