Enhanced Prediction of Airfoil’s Drag Coefficient Using Curve Fitting and Artificial Neural Network Preprint Version Information

EasyChair Preprint 14537

Abstract

This study explores the application of Artificial Neural Networks (ANNs) for predicting the aerodynamic coefficients of airfoils, with a focus on the drag coefficient (C_D), as the literature has not predicted it as precisely as other aerodynamic coefficients. A novel quadratic fitting function is introduced to improve the accuracy of C_D predictions. Two datasets, DI and DII, with varying ranges of Mach numbers, were prepared, and the performance of the ANNs was evaluated. Model I was trained on Dataset I (Mach 0.1 to 0.3), while Model II was trained on Dataset II (Mach 0.1 to 0.8). The results indicate that a larger and more diverse dataset significantly enhances the predictive capabilities of the model. Additionally, the model's ability to generalize to airfoils and flight conditions outside the training data was tested, revealing the generalization power of the model.

Keyphrases: Aerodynamic Prediction, Airfoil, Artificial Neural Networks (ANNs), Drag coefficient, Optimization, data-driven models, polynomial regression

@booklet{EasyChair:14537,
  author    = {Mohssen Elshaar and Naef Qasem},
  title     = {Enhanced Prediction of Airfoil’s Drag Coefficient Using Curve Fitting and Artificial Neural Network  Preprint Version Information},
  howpublished = {EasyChair Preprint 14537},
  year      = {EasyChair, 2024}}