基于深度学习预测CO2在离子液体中溶解度的研究
Prediction of solubility of CO2 in ionic liquids based on deep learning
构建了深度神经网络(DNN)和随机森林(RF)技术用于预测CO2在ILs中的溶解度。收集5 000个不同温度和压力下ILs的CO2溶解度值,采用分子描述符(MD)充分获取到ILs的复杂结构信息。以MD作为DNN、RF机器学习模型的输入特征。其中,MD-RF模型以卓越的准确脱颖而出,展示了0.988 4的决定系数(R2)和0.000 5的均方误差(MSE)。
Deep neural networks (DNN) and random forest (RF) techniques were used to predict the solubility of CO2 in ILs in this work.5000 CO2 solubility values of ILs at different temperatures and pressures were collected and the complex structural information of ILs using molecular descriptors (MD) were fully obtained.Using MD as the input feature for DNN and RF machine learning models.Among them,the MD-RF model stands out for its excellent accuracy,displaying a coefficient of determination (R2) of 0.988 4 and a mean squared error (MSE) of 0.000 5.
离子液体 / 气体吸收 / 机器学习 / 溶解度 / CO2
ionic liquids / gas absorption / machine learning / solubility / CO2
| [1] |
|
| [2] |
|
| [3] |
|
| [4] |
|
| [5] |
|
| [6] |
|
| [7] |
|
| [8] |
|
| [9] |
|
| [10] |
|
| [11] |
|
| [12] |
张钰, 魏世丞, 董超芳, |
| [13] |
|
| [14] |
张锟滨, 陈玉明, 吴克寿, |
2019年宁夏青年拔尖人才培养工程(宁人社发〔2019〕90号)
2024年宁夏回族自治区重点研发计划项目(2024BEE02038)
2023年国家能源集团科技创新项目(GJNY-23-53)
/
| 〈 |
|
〉 |