Airfoil Self Noise Prediction Using Machine Learning and Explainable AI

A major environmental challenge in the aerospace sector is presented by aircraft noise, particularly airfoil noise. Traditional methods for predicting airfoil noise involve complex and computationally intensive numerical simulations. This research explores the use of machine learning to enhance the efficiency and accuracy of airfoil noise prediction. By utilizing extensive datasets from wind tunnel experiments, encompassing a variety of airfoil geometries and flow conditions, supervised learning algorithms are employed to construct a predictive model. Diverse regression models, including random forests, support vector regression, CatBoost, XgBoost, Gradient Boosting and decision trees, are trained on the NACA 0012 airfoil dataset to uncover correlations between airfoil characteristics and noise emissions. Initial findings underscore the potential of machine learning, with Gradient Boosting regression achieving a noteworthy R2 score of 0.96625 and RMSE as 1.2622. This study not only equips engineers with a valuable tool for optimizing airfoil shapes to reduce noise but also establishes the groundwork for real-time noise prediction systems in aircraft operations.