<ul><li>A comparative evaluation of surrogate modeling approaches for predicting drag on a real-world car aerodynamics dataset was conducted.</li><li>The evaluation compared a Convolutional Neural Network (CNN) model using a signed distance field as input and a commercial tool based on Graph Neural Networks (GNN) processing a surface mesh.</li><li>The CNN-based method achieved a mean absolute error of 2.3 drag counts, while the GNN-based method achieved 3.8.</li><li>Both methods achieved approximately 77% accuracy in predicting the direction of drag change relative to the baseline geometry.</li></ul>

Benchmarking Convolutional Neural Network and Graph Neural Network based Surrogate Models on a Real-World Car External Aerodynamics Dataset

Discover more