Mechanism Knowledge-enhanced Graph Attention Neural Network for Aviation Engine Nacelle Structural Damage Prediction

It is crucial to conduct damage prediction on the aeroengine nacelle to ensure the safety of aircraft operation. Using maintenance data for nacelle damage prediction faces challenges such as long-time spans, sparse data, and an ignored impact of meteorological factors on nacelle structural damage. Common deep learning models ignore the coupling effects of meteorological factors on structural damage when processing such data. To address this issue, this paper proposes a mechanism-enhanced graph attention neural network (MKEGAT) for nacelle structural damage prediction. First, the correlations between temperature and relative humidity are derived based on the interaction mechanism and Spearman correlation coefficients between meteorological parameters to construct the graph structure representing these correlations. Second, a graph attention neural network (GAT) is introduced to learn the effects of meteorological parameters on nacelle structural delamination damage. Finally, global features are extracted from the graph structure and mapped into structural damage predictions using a regressor composed of fully connected layers (FCLs). Experimental results indicate that the proposed MKEGAT model significantly improves prediction accuracy compared to existing methods, validating its effectiveness and superiority.