This work aims to predict airfoil's performance characteristics from a binary image of an airfoil at a specific AoA. This implementation uses a rudimentary CNN that has about 700000 parameters.
The dataset for this project is from 'ziliHarvey's' project. The '1_300.mat' file was used for training.
The model was trained for 35 epochs and the Validation loss fell to 0.0037.
- Absolute normalized Loss of 50 samples from the training dataset.
\
- Training and Validation Loss
\
- cl/cd comparison of predicted values and real values from the train dataset.
\
This repo and the project contained herein is purely for educational purposes. Under no circumstances are you authorized to use any form of this repo for any type of commercial purposes. The Dataset Used for this project is not mine and hence I have no authority over it.