New Machine Learning Model for Predicting AIrcraft Emissions Based on Engine, Operational Parameters, and Fuel Properties

With growing environmental concerns surrounding it, all efforts in the aviation sector are moving toward reducing the ecological footprint of this sector. One of the promising solutions is sustainable aviation fuel, which offers an alternative to traditional jet fuel. This study investigates the possibility of using a developed machine learning model designed to forecast aircraft emissions using a set of 11 inputs related to engine specifications, fuel properties, and ambient air conditions. The model is trained based on version 30 of the International Civil Aviation Organization’s engine emissions databank. An artificial neural network was created after data cleaning and preparation for its strength in modeling intricate, nonlinear interactions between inputs and predicted emissions. The model generates estimates of the emission index for carbon dioxide, nitrogen oxides, carbon monoxide, and fuel flow. These predictions help assess how adjustments in operational parameters influence emissions. Additionally, the model can support more refined analysis across different flight scenarios by incorporating data from the Automatic Dependent Surveillance–Broadcast and weather information.