Constructing a physics-guided machine learning neural network to predict tonal noise emitted by a propeller

Artificial neural networks offer a highly nonlinear and adaptive model for predicting complex interactions between input-output parameters. However, these networks require large datasets which often exceed practical considerations in modeling experimental results. To alleviate the dataset size requirement, a method known as physics-guided machine learning has been applied to construct several neural networks for predicting propeller tonal noise in the time domain over a broad range of flight conditions. Three space-filling designs, namely, Latin-Hypercube, Sphere-Packing, and Grid-Space, were used to distribute points throughout the input parameter space encompassing nondimensional flight conditions and observer geometry. Each neural network's performance was validated by conditions outside of the training set and compared to the Propeller Analysis System tool from the NASA Aircraft Noise Prediction Program. The Latin-Hypercube and the Sphere-Packing designs provided a uniform representation of the domain, which improved the tonal noise prediction in comparison to the Grid design. The black-box nature of these neural networks was explored, and post-network functions were developed to remove discontinuities in the acoustic signal. Overall, the methods herein show a notable improvement in prediction performance in comparison to a multilayer perceptron. Additional loss functions are necessary for ensuring reasonable accuracy of predictive networks on small datasets and will be investigated for the final paper.