結合機器學習方法與COSMO溶合計算以估算非締合化學物質之PC-SAFT狀態方程式參數

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：19

、訪客IP：3.141.7.152

姓名

詹士翰(Shi-Han Zhan) 查詢紙本館藏

畢業系所

化學工程與材料工程學系

論文名稱

結合機器學習方法與COSMO溶合計算以估算非締合化學物質之PC-SAFT狀態方程式參數
(Combining Machine Learning Methods and COSMO Solvation Calculation to Estimate Pure Component Parameters of PC-SAFT EOS for Non-association Chemicals)

相關論文

★ 預測固體溶質於超臨界二氧化碳添加共溶劑系統之溶解度	★ 碳酸二乙酯與低碳醇類於常壓下之汽液相平衡
★ 探討Peng-Robinson+COSMOSAC狀態方程式中分散項與溫度之關係	★ 探討分散項之溫度函數與體積參數之修正對PR+COSMOSAC於相平衡預測之影響
★ 預測有機物與二氧化碳雙成份系統之固液氣三相平衡	★ 常壓下乙酸酯類之雙成份混合物汽液相平衡
★ 以第一原理計算鋰嵌入與擴散於具氧空缺之二氧化鈦結構	★ 探討不同量子化學方法對PR+COSMOSAC狀態方程式應用於預測純物質及混合流體相行為之影響
★ 預測固體溶質於超臨界二氧化碳中的溶解度	★ 鋯金屬有機框架材料之碳氫氣體吸附與分離預測
★ 甲基水楊酸異構物於超臨界二氧化碳中之溶解度量測	★ 原料藥與水楊酸衍生物於超臨界二氧化碳中之溶解度量測
★ 以第一原理計算探討鋰於鈮摻雜二氧化鈦之嵌入與擴散路徑	★ 探討COSMO-SAC-dsp模型中分散項和組合項之效應
★ 第一原理計算探討藍磷烯異質結構用於鋰離子電池負極材料之特性	★ 以第一原理計算探討鋰離子於鐵摻雜磷酸鋰鈷之塊材與表面附近之擴散路徑

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

至系統瀏覽論文 (2033-6-30以後開放)

摘要(中)

本研究針對PC-SAFT狀態方程式純物質參數進行預測，PC-SAFT又分為有氫鍵系統與無氫鍵系統，有氫鍵系統需要五個純物質參數m、σ、ϵ、 ϵAB與кAB，無氫鍵系統則需要三個純物質參數m、σ、ϵ，本文只探討PC-SAFT無氫鍵系統的部份。我們對兩種機器學習方法進行討論與比較，分別為深度神經網絡(deep neural network, DNN)與隨機森林(random forest, RF)。我們提出機器學習與熱力學模型COSMO溶合計算結合，透過COSMO計算後產生的Sigma Profile作為機器學習的特徵進而對純物質參數進行預測。
本研究收集了1153個無氫鍵鍵結的分子Sigma Profile與其PC-SAFT的三個純物質參數，作為機器學習的數據集。我們將訓練模型的大小分成兩組，將數據集以70%作為訓練集30%測試集為第一組，80%作為訓練集20%測試集作為第二組，共兩組。在以分子特徵作為輸入除了Sigma Profile外，加上透過COSMO計算得到的面積與體積(Sigma Profile + area + volume)。在深度神經網絡與隨機森林超參數的優化，我們使用網格搜索(grid search, GS)與貝葉斯優化(bayesian optimization, Byn) 進行模型調整參數並做比較。在70%的訓練的系統下表現最好的為GS搭配DNN:GS+DNN-m的誤差(AARD%)為8.61%，GS+DNN-σ與GS+DNN- ϵ的誤差為3.23%與6.41%。而在80%的訓練系統下，整體誤差最低的也是GS搭配DNN:GS+DNN-m的誤差為8.68%，GS+DNN-σ與GS+DNN- ϵ的誤差為3.06%與6.41%，我們提出的COSMO溶合計算結合機器學習的模型非常穩定，而DNN比RF提供更準確的預測結果。

關鍵字: 機器學習、深度神經網絡、隨機森林、COSMO溶合計算、PC-SAFT狀態方程式、熱力學

摘要(英)

This study focuses on predicting pure component parameters of the PC-SAFT (perturbed chain statistical associating fluid theory) equation of state for the non-hydrogen bonding system. PC-SAFT has two variants, one for systems with hydrogen bonding that requires five pure component parameters (m, σ, ε, ϵAB, and кAB), and another for systems without hydrogen bonding that requires three pure component parameters (m, σ, and ε). This study specifically explores the non-hydrogen bonding system of PC-SAFT. This research discusses and compares two machine learning methods, deep neural network (DNN) and random forest (RF). We propose a combination of machine learning and the thermodynamic model COSMO-SAC, where the Sigma Profile calculated by COSMO-SAC is used as a feature for machine learning to predict the pure component parameters. The dataset for machine learning consists of 1153 Sigma Profiles of molecules without hydrogen bonding and their corresponding three pure component parameters for PC-SAFT. The dataset is divided into two groups for training the models: the first group uses 70% of the data as the training set and 30% as the test set, while the second group uses 80% as the training set and 20% as the test set. For the optimization of hyperparameters in DNN and RF, we employ grid search (GS) and bayesian optimization (Byn) methods to tune the model parameters and make comparisons. Under the 70% training system, the best-performing model is GS combined with DNN (GS+DNN-m) with the an AARD% (average absolute relative deviation) of 8.61%, GS+DNN-σ is 3.23% and for GS+DNN-ε is 6.41%. Under the 80% training system, the overall lowest error is also observed with GS and DNN, for GS+DNN-m is 8.68%, GS+DNN-σ is 3.06% and for GS+DNN-ε is 6.41%. The proposed model, combining COSMO solvation calculation result with machine learning, demonstrates great stability, and DNN provides more accurate predictions compared to RF.
Key words: Machine Learning、Deep Neural Network、Random Forest、COSMO Calculation、PC-SAFT Equation of State、Thermodynamic

關鍵字(中)

★ 機器學習
★ 深度神經網絡
★ 隨機森林
★ 熱力學
★ 網格搜索
★ 貝葉斯優化

關鍵字(英)

★ Machine learning
★ Deep neural network
★ Random forest
★ COSMO calculation
★ PC-SAFT equation of state
★ thermodynamic

論文目次

中文摘要 i
Abstract ii
致謝 iii
目錄 iv
圖目錄 vi
表目錄 ix
第一章緒論 1
1-1熱力學性質對工業的貢獻 1
1-2 回顧熱力學結合AI 2
1-3 機器學習簡介 3
1-4研究動機 5
第二章計算原理與細節 6
2-1 簡述PC-SAFT與數據集 6
2-2 Sigma Profile的產生 9
2-3 深度神經網絡 10
2-4 隨機森林 17
2-5 超參數調整 19
2-5-1 網格搜索 20
2-5-2 貝葉斯優化 20
2-6 PC-SAFT純物質參數計算流程 24
第三章結果與討論 25
3-1機器學習之計算結果 26
3-2 特徵重要性與判斷是否過擬合 49
3-3 與GC＋ANN計算PC-SAFT EOS純物質參數結果比較 57
第四章結論 59
參考文獻 60
附錄(一) 訓練規模0.6使用Sigma Profile + area + volume作為輸入 65
附錄(二) 訓練規模0.7根據特徵重要性僅使用area + volume作為輸入之計算結果 67
附錄(三) PC-SAFT EOS純物質參數真實值 68

參考文獻

[1]. S. Kim, et al., “PubChem in 2021: new data content and improved web interfaces”, Nucleic Acids Res, Vol 49, pp. D1388-D1395, 2021.
[2]. J.D. Van Der Waals and J.S. Rowlinson, On the continuity of the gaseous and liquid states, Courier Corporation, 2004.
[3]. D.-Y. Peng and D.B. Robinson, “A new two-constant equation of state”, Industrial Engineering Chemistry Fundamentals, Vol 15, pp. 59-64, 1976.
[4]. O. Redlich and J.N. Kwong, “On the thermodynamics of solutions. V. An equation of state. Fugacities of gaseous solutions”, Chemical Reviews, Vol 44, pp. 233-244, 1949.
[5]. G. Soave, “Equilibrium constants from a modified Redlich-Kwong equation of state”, Chemical Engineering Science, Vol 27, pp. 1197-1203, 1972.
[6]. J. Gross and G. Sadowski, “Perturbed-chain SAFT: An equation of state based on a perturbation theory for chain molecules”, Industrial & Engineering Chemistry Research, Vol 40, pp. 1244-1260, 2001.
[7]. J. Gross and G. Sadowski, “Application of the perturbed-chain SAFT equation of state to associating systems”, Industrial & Engineering Chemistry Research, Vol 41, pp. 5510-5515, 2002.
[8]. F. Tumakaka, et al., “Modeling of polymer phase equilibria using Perturbed-Chain SAFT”, Fluid Phase Equilibria, Vol 194, pp. 541-551, 2002.
[9]. I.A. Kouskoumvekaki, et al., “Novel method for estimating pure-component parameters for polymers: application to the PC-SAFT equation of state”, Industrial & Engineering Chemistry Research, Vol 43, pp. 2830-2838, 2004.
[10]. A. Tihic, et al., “A predictive group-contribution simplified PC-SAFT equation of state: application to polymer systems”, Industrial & Engineering Chemistry Research, Vol 47, pp. 5092-5101, 2008.
[11]. M.C. Kroon, et al., “Modeling of the carbon dioxide solubility in imidazolium-based ionic liquids with the tPC-PSAFT equation of state”, The Journal of Physical Chemistry B, Vol 110, pp. 9262-9269, 2006.
[12]. Y. Chen, et al., “Modeling the solubility of carbon dioxide in imidazolium-based ionic liquids with the PC-SAFT equation of state”, The Journal of Physical Chemistry B, Vol 116, pp. 14375-14388, 2012.
[13]. A.P. Carneiro, et al., “Solubility of sugars and sugar alcohols in ionic liquids: measurement and PC-SAFT modeling”, The Journal of Physical Chemistry B, Vol 117, pp. 9980-9995, 2013.
[14]. H. Renon and J.M. Prausnitz, “Local compositions in thermodynamic excess functions for liquid mixtures”, AIChE Journal, Vol 14, pp. 135-144, 1968.
[15]. D.S. Abrams and J.M. Prausnitz, “Statistical thermodynamics of liquid mixtures: a new expression for the excess Gibbs energy of partly or completely miscible systems”, AIChE Journal, Vol 21, pp. 116-128, 1975.
[16]. G.M. Wilson, “Vapor-liquid equilibrium. XI. A new expression for the excess free energy of mixing”, Journal of the American Chemical Society, Vol 86, pp. 127-130, 1964.
[17]. S.-T. Lin and S.I. Sandler, “A Priori Phase Equilibrium Prediction from a Segment Contribution Solvation Model”, Industrial & Engineering Chemistry Research, Vol 41, pp. 899-913, 2002.
[18]. K.G. Joback and R.C. Reid, “Estimation of pure-component properties from group-contributions”, Chemical Engineering Communications, Vol 57, pp. 233-243, 1987.
[19]. A. Fredenslund, et al., “Group‐contribution estimation of activity coefficients in nonideal liquid mixtures”, AIChE Journal, Vol 21, pp. 1086-1099, 1975.
[20]. J. Gmehling, et al., “A modified UNIFAC model. 2. Present parameter matrix and results for different thermodynamic properties”, Industrial & Engineering Chemistry Research, Vol 32, pp. 178-193, 1993.
[21]. A. Klamt, “Conductor-like screening model for real solvents: a new approach to the quantitative calculation of solvation phenomena”, The Journal of Physical Chemistry, Vol 99, pp. 2224-2235, 1995.
[22]. S. Gebreyohannes, et al., “A comparative study of QSPR generalized activity coefficient model parameters for vapor–liquid equilibrium mixtures”, Industrial & Engineering Chemistry Research, Vol 55, pp. 1102-1116, 2016.
[23]. S. Hu, et al., “A deep learning-based chemical system for QSAR prediction”, IEEE Journal of Biomedical Health Informatics, Vol 24, pp. 3020-3028, 2020.
[24]. H. Hayer, et al., “Support vector machine and CPA EoS for the prediction of high-pressure liquid densities of normal alkanols”, Journal of the Taiwan Institute of Chemical Engineers, Vol 45, pp. 2888-2898, 2014.
[25]. A.M. Abudour, et al., “Volume-translated Peng–Robinson equation of state for saturated and single-phase liquid densities”, Fluid Phase Equilibria, Vol 335, pp. 74-87, 2012.
[26]. S. Dick, “Artificial intelligence”, pp. 2019.
[27]. A. Abdallah el hadj, et al., “Novel approach for estimating solubility of solid drugs in supercritical carbon dioxide and critical properties using direct and inverse artificial neural network (ANN)”, Neural Computing and Applications, Vol 28, pp. 87-99, 2015.
[28]. F. Gharagheizi, et al., “Artificial neural network modeling of solubilities of 21 commonly used industrial solid compounds in supercritical carbon dioxide”, Industrial & Engineering Chemistry Research, Vol 50, pp. 221-226, 2011.
[29]. B. Mehdizadeh and K. Movagharnejad, “A comparison between neural network method and semi empirical equations to predict the solubility of different compounds in supercritical carbon dioxide”, Fluid Phase Equilibria, Vol 303, pp. 40-44, 2011.
[30]. M. Abdi-Khanghah, et al., “Prediction of solubility of N-alkanes in supercritical CO2 using RBF-ANN and MLP-ANN”, Journal of CO2 Utilization, Vol 25, pp. 108-119, 2018.
[31]. A. Aminian, “Estimating the solubility of different solutes in supercritical CO2 covering a wide range of operating conditions by using neural network models”, The Journal of Supercritical Fluids, Vol 125, pp. 79-87, 2017.
[32]. P. Santak and G. Conduit, “Predicting physical properties of alkanes with neural networks”, Fluid Phase Equilibria, Vol 501, pp. 2019.
[33]. Y.-L. Yang, et al., “Molecular structure incorporated deep learning approach for the accurate interfacial tension predictions”, Journal of Molecular Liquids, Vol 323, pp. 114571, 2021.
[34]. Z. Zhang, et al., “Machine learning predictive framework for CO2 thermodynamic properties in solution”, Journal of CO2 Utilization, Vol 26, pp. 152-159, 2018.
[35]. D.S. Palmer, et al., “Random forest models to predict aqueous solubility”, Journal of Chemical Information Modeling Vol 47, pp. 150-158, 2007.
[36]. R. Rajappan, et al., “Quantitative Structure− Property Relationship (QSPR) Prediction of Liquid Viscosities of Pure Organic Compounds Employing Random Forest Regression”, Industrial & Engineering Chemistry Research, Vol 48, pp. 9708-9712, 2009.
[37]. J. Wang, et al., “Prediction of CO2 solubility in deep eutectic solvents using random forest model based on COSMO-RS-derived descriptors”, Green Chemical Engineering, Vol 2, pp. 431-440, 2021.
[38]. J.P.S. Aniceto, et al., “Machine learning models for the prediction of diffusivities in supercritical CO2 systems”, Journal of Molecular Liquids, Vol 326, pp. 2021.
[39]. H. Matsukawa, et al., “Estimation of pure component parameters of PC-SAFT EoS by an artificial neural network based on a group contribution method”, Fluid Phase Equilibria, Vol 548, pp. 2021.
[40]. J.P. Wolbach and S.I. Sandler, “Using molecular orbital calculations to describe the phase behavior of cross-associating mixtures”, Industrial & Engineering Chemistry Research, Vol 37, pp. 2917-2928, 1998.
[41]. N. Ramírez-Vélez, et al., “Parameterization of SAFT models: analysis of different parameter estimation strategies and application to the development of a comprehensive database of PC-SAFT molecular parameters”, Journal of Chemical Engineering Data, Vol 65, pp. 5920-5932, 2020.
[42]. A. Klamt and G. Schüürmann, “COSMO: a new approach to dielectric screening in solvents with explicit expressions for the screening energy and its gradient”, Journal of the Chemical Society, Perkin Transactions 2, pp. 799-805, 1993.
[43]. E. Mullins, et al., “Sigma-profile database for using COSMO-based thermodynamic methods”, Industrial & Engineering Chemistry Research, Vol 45, pp. 4389-4415, 2006.
[44]. B. Delley, “From molecules to solids with the DMol 3 approach”, The Journal of Chemical Physics, Vol 113, pp. 7756-7764, 2000.
[45]. S. Wang, et al., “Use of GAMESS/COSMO program in support of COSMO-SAC model applications in phase equilibrium prediction calculations”, Fluid Phase Equilibria, Vol 276, pp. 37-45, 2009.
[46]. F. Ferrarini, et al., “An open and extensible sigma‐profile database for COSMO‐based models”, AIChE Journal, Vol 64, pp. 3443-3455, 2018.
[47]. M. Frisch, et al., Gaussian 03, revision C. 02. 2004, Gaussian, Inc., Wallingford, CT.
[48]. T. Mu, et al., “Performance of COSMO-RS with sigma profiles from different model chemistries”, Industrial & Engineering Chemistry Research, Vol 46, pp. 6612-6629, 2007.
[49]. E. Paulechka, et al., “Reparameterization of COSMO-SAC for phase equilibrium properties based on critically evaluated data”, Journal of Chemical Engineering Data, Vol 60, pp. 3554-3561, 2015.
[50]. W.-L. Chen, et al., “A critical evaluation on the performance of COSMO-SAC models for vapor–liquid and liquid–liquid equilibrium predictions based on different quantum chemical calculations”, Industrial & Engineering Chemistry Research, Vol 55, pp. 9312-9322, 2016.
[51]. Y. Bengio and Y. LeCun, “Scaling learning algorithms towards AI”, Large-scale kernel machines, Vol 34, pp. 1-41, 2007.
[52]. K. Koh, et al., “An interior-point method for large-scale l1-regularized logistic regression”, Journal of Machine Learning Research, Vol 8, pp. 1519-1555, 2007.
[53]. D.E. Rumelhart, et al., “Learning representations by back-propagating errors”, Nature, Vol 323, pp. 533-536, 1986.
[54]. M.A. Nielsen, Neural networks and deep learning, Vol. 25. Determination press San Francisco, CA, USA, 2015.
[55]. Y. Bengio, et al., “Learning long-term dependencies with gradient descent is difficult”, IEEE Transactions on Neural Networks, Vol 5, pp. 157-166, 1994.
[56]. L. Breiman, “Random forests”, Machine Learning, Vol 45, pp. 5-32, 2001.
[57]. F. Pedregosa, et al., “Scikit-learn: Machine learning in Python”, The Journal of Machine Learning Research, Vol 12, pp. 2825-2830, 2011.
[58]. P.I. Frazier, Bayesian optimization, in Recent advances in optimization and modeling of contemporary problems. 2018, Informs. p. 255-278.
[59]. L. Constantinou and R. Gani, “New group contribution method for estimating properties of pure compounds”, AIChE Journal, Vol 40, pp. 1697-1710, 1994.
[60]. J. Marrero and R. Gani, “Group-contribution based estimation of pure component properties”, Fluid Phase Equilibria, Vol 183, pp. 183-208, 2001.
[61]. Y. Nannoolal, et al., “Estimation of pure component properties: Part 1. Estimation of the normal boiling point of non-electrolyte organic compounds via group contributions and group interactions”, Fluid Phase Equilibria, Vol 226, pp. 45-63, 2004.
[62]. Y. Nannoolal, et al., “Estimation of pure component properties: Part 2. Estimation of critical property data by group contribution”, Fluid Phase Equilibria, Vol 252, pp. 1-27, 2007.

指導教授

謝介銘(Chieh-Ming Hsieh)

審核日期

2023-7-24

推文