Predicting PC-SAFT pure-component parameters by machine learning using a molecular fingerprint as key input
Fluid Phase Equilibr.: RESOLV PI Prof. G. Sadowski's work introduces a Machine learning (ML) approach using extended-connectivity fingerprints (ECFPs) that can accurately predict PC-SAFT pure-component parameters
In this work, a machine learning (ML)-approach was developed to predict pure-component parameters for the Perturbed-Chain Statistical Associating Fluid Theory (PC-SAFT) for non-associating molecules using a deep neural network. Extended-connectivity fingerprints (ECFP) were used as key inputs for the neural network to achieve flexible and easily available non-experimental representations of a chemical molecule. A detailed analysis of bit collisions during the ECFP generation was performed to obtain the optimal ECFP initial bit lengths creating possible inputs for neural networks to be trained.
