Predicting Lap Times of Formula 1 Races using ANN

Motorsport performance analysis has become much more sophisticated with the inclusion of machine learning and predictive analytics. This paper proposes a Lap Time Prediction Model for Formula 1 (F1) races based on a Long Short-Term Memory (LSTM) neural network trained on past race data. The model includes tire degradation modeling, fuel burn rates, and track temperature changes to improve prediction accuracy. The main goal is to forecast lap times depending on tire compounds and racing conditions. The degradation model is formulated using logarithmic tire wear functions, motivated by motorsport research. The system can predict lap times for various tire compounds (Soft, Medium, Hard), combining real-time simulation knowledge. The model is also optimized with LeakyReLU activation to prevent the vanishing gradient problem when dealing with sequential data. The trained models are stored and can be utilized for real-time simulation of race strategies. The strategy enables race engineers to make well-informed strategic decisions about pit stops, fuel consumption, and tire usage.