Theoretical and computational investigations on estimation of viscosity of ionic liquids for green adsorbent: Effect of temperature and composition
Ionic liquids can be recognized as green adsorbent for water purification due to their superior characteristics compared to conventional organic solvents. However, their physiochemical properties are difficult to estimate due to their wide range and complex structure. The viscosity of ionic liquids...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-01-01
|
| Series: | Case Studies in Thermal Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2214157X24017349 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Ionic liquids can be recognized as green adsorbent for water purification due to their superior characteristics compared to conventional organic solvents. However, their physiochemical properties are difficult to estimate due to their wide range and complex structure. The viscosity of ionic liquids versus temperature and composition are estimated in this study via machine learning. The models used are Gaussian Process Regression (GPR), Multilayer Perceptron (MLP), and Support Vector Regression (SVR). The dataset comprises categorical information on Cation and Anion, along with numeric variables T(K) and xIL (mol%), serving as inputs for the models, while Viscosity (Pa.s) represents the output variable. The models are optimized using the Whale Optimization Algorithm (WOA) to fine-tune hyperparameters, enhancing their predictive performance. Subsequently, a comprehensive analysis of the results reveals the predictive capabilities of each model. GPR exhibits exceptional accuracy, yielding a high R2 of 0.99765, minimal MAE of 1.73330E-03 and RMSE of 2.2317E-03. This suggests that GPR effectively captures the underlying patterns in the dataset, demonstrating superior predictive capabilities. MLP also showcases commendable performance, with an R2 of 0.88827. While MLP may not match the precision of GPR, its neural network architecture proves effective in capturing non-linear relationships within the data. SVR emerges as an accurate model with a score of 0.99526 in terms of R2. SVR's ability to handle complex data distributions and provide accurate predictions positions it as a valuable tool for viscosity prediction in ionic liquids. |
|---|---|
| ISSN: | 2214-157X |