Encoder Decoder RNN Model for Cold Start Emissions

This software was used to predict hazardous gaseous emissions such as CO, NOx and unburned hydrocarbons (UHC) from passenger cars. The predictions are based on seven measured features (2,500 time steps at a 5 Hz rate) as input. The three emissions (outputs) are predicted simultaneously within the same model. The measurements were conducted after the cars had been stationary for an entire "night," meaning the engine was in a cold state at the start of the measurements (hence, a cold-start). The corresponding scientific publication is: "Mangipudi, Manoj, et al. Prediction of Hazardous Gaseous Emissions from a Gasoline Engine during Cold Starts Using Machine Learning Methods. No. 2025-01-0321. SAE Technical Paper, 2025".