Comparative analysis of machine learning models for Ammonia Capture of Ionic Liquids
This work addresses the need for accurate solubility predictions in eco-friendly solvents for industrial applications, but it is incremental as it applies existing methods to a specific domain.
The study compared machine learning models and equations of state for predicting ammonia solubility in ionic liquids, finding that AI methods like MLP and PSO-ANFIS produced promising results while traditional equations were inaccurate.
Industry uses various solvents in the processes of refrigeration and ventilation. Among them, the Ionic liquids (ILs) as the relatively new solvents, are known for their proven eco-friendly characteristics. In this research, a comprehensive literature review was carried out to deliver an insight into the ILs and the prediction models used for estimating the ammonia solubility in ILs. Furthermore, a number of advanced machine learning methods, i.e. multilayer perceptron (MLP) and a combination of particle swarm optimization (PSO) and adaptive neuro-fuzzy inference system (ANFIS) models are used to estimate the solubility of ammonia in various ionic liquids. Affecting parameters were molecular weight, critical temperature and pressure of ILs. Furthermore, the salability is also predicted using the two-equation of states. Down the line, some comparisons were drawn between experimental and modeling results which is rarely done. The study shows that the equations of states are not able estimate the solubility of ammonia accurately, by contrast, artificial intelligence methods have produced promising results.