Concerns over global climate change and decreasing reserves of traditional, petroleum-derived fuels has spurred interest in finding cleaner, more efficient alternative fuels. To this end, computational models have played an important role in estimating key physical and chemical properties of alternative fuels and fuel additives before experimental testing occurs. Artificial neural networks have been shown to accurately predict a variety of properties relevant to alternative fuel development, however the vast majority of these models are trained to predict one<br/>property. The present work introduces a neural network architecture capable of predicting research and motor octane number (RON, MON) simultaneously and is shown to be very competitive in accuracy relative to modern alternatives designed to predict RON/MON individually.<br/>Additionally, key relationships between compound structure and RON/MON are illustrated and investigated further.

Predicting research and motor octane number using a single artificial neural network

ACS Fall 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.