HANNA: hard-constraint neural network for consistent activity coefficient prediction
We introduce HANNA, the first hybrid neural network model that strictly complies with all thermodynamic consistency criteria for predicting activity coefficients and outperforms current benchmark methods in terms of accuracy and applicability.
Chemistry↗