Case studies on predictive maintenance of aircraft large-opening panels using integrated planar electromagnetic sensing and dynamic Bayesian networks

This study proposes an integrated framework for fatigue crack monitoring, propagation prognostics, and condition-based maintenance of aircraft aluminum alloy structures with large cutouts. A flexible planar eddy current sensor array is developed to achieve real-time, multi-channel monitoring of crack initiation and growth under laboratory fatigue testing conditions, with a minimum detectable crack length of 0.3 mm and a spatial resolution of 2 mm. A dynamic Bayesian network combined with a particle filter and the Walker crack growth model is employed to probabilistically forecast crack propagation by continuously updating model parameters using sensor data. Experimental results demonstrate that the proposed method significantly reduces prognostic uncertainty and improves accuracy compared with deterministic models. Based on the estimated crack evolution, a health index-driven condition-based maintenance strategy is established, enabling rational selection of repair methods and maintenance timing. The results show that the proposed approach effectively supports safe and economical maintenance decision-making for aircraft structures.