05 - Propulsion and Energy
PD-STGNN: A PHYSICS-INFORMED DYNAMIC SPATIO-TEMPORAL GRAPH NEURAL NETWORK FOR AERO-ENGINE THRUST PREDICTION
C. Song¹, H. Shi¹, Z. Huang¹, H. Xiao¹; ¹Northwestern Polytechnical University, China
This study proposes a physics-informed dynamic spatio-temporal graph neural network framework that jointly exploits and integrates complex spatio-temporal features from multi-source sensor data to construct a high-accuracy engine thrust prediction model. Validation on a real engine test dataset demonstrates that the proposed method achieves high prediction accuracy and strong robustness.
DOI: 10.71945/icas2026_0314