The Porsche 919 Hybrid Evo project applied machine learning-based dimensionality reduction to transform a large database of about 1,600 airfoil shapes into a compact set of 5 to 10 key parameters called "airfoil genes." This drastically reduced the high-dimensional design space, enabling efficient aerodynamic optimization of the rear wing. Techniques similar to Principal Component Analysis and autoencoders were used to extract these essential shape features.

With the reduced parameter set, Porsche engineers employed genetic algorithms to optimize wing designs, iteratively refining shapes to maximize downforce and aerodynamic efficiency. The optimized designs were validated through computational fluid dynamics simulations and wind tunnel testing, ensuring real-world performance gains.

This approach allowed Porsche to explore a vast design space quickly and effectively, contributing to a 53% increase in downforce and a 66% improvement in aerodynamic efficiency compared to the previous model. These aerodynamic advancements, combined with powertrain and weight improvements, helped the 919 Hybrid Evo achieve historic lap records at Spa-Francorchamps and the Nürburgring Nordschleife, outperforming contemporary Formula 1 cars on certain tracks.

The use of machine learning for dimensionality reduction and optimization exemplifies how advanced AI techniques can revolutionize engineering workflows, enabling faster innovation cycles and delivering a significant competitive advantage in motorsport.