Machine Learning-Augmented Far-Field Noise Prediction for Distributed Electric Propulsion Aircraft

The growing adoption of Distributed Electric Propulsion (DEP) aircraft presents new challenges in aeroacoustic modeling, as multi-propeller interactions, airframe effects, and noise certification requirements become critical barriers to large-scale deployment. Current low-fidelity noise models fail to capture complex noise sources, while high-fidelity CFD-based simulations are computationally prohibitive for rapid design iteration. To address this, we propose a machine learning-driven surrogate modeling framework that enables real-time, high-accuracy noise prediction for DEP aircraft. This framework integrates Graph Neural Networks (GNNs) and Fourier Neural Operators (FNOs) to predict full flow-fields and far-field noise signatures, rather than relying solely on empirical regression of noise metrics. Compared to traditional scale-resolving CFD, this approach achieves 1,000x faster computations while maintaining high accuracy (5% MSE with a few hundred simulations). By leveraging multi-fidelity aeroacoustic data sources (FW-H, VPM, CFD) and embedding physics constraints, the model can generalize across various DEP configurations, enabling faster aircraft design, optimization, and certification. Phase I funding will be used to develop and validate the FNO-GNN prediction methodology, starting with single-propeller and wing interactions before expanding to multi-propeller DEP configurations in Phase II. The technology targets NASA aeronautics programs, OEMs (Airbus, Boeing, Joby, Archer, Lilium), defense contractors (Lockheed Martin, Northrop Grumman), and regulatory agencies (FAA, ICAO, EASA), supporting urban air mobility (UAM), hybrid-electric regional aircraft, and UAV applications. By accelerating DEP aircraft noise prediction and mitigation, this innovation directly supports NASA’s Sustainable Aviation and Advanced Air Mobility (AAM) initiatives, providing a scalable, high-impact solution for future electric aviation.