文献类型：期刊文献

英文题名：Artificial neural network-based optimization of medium-polar triterpenoids from the fruiting body of Ganoderma lucidum, kinetic modeling and thermodynamics

作者：Li, Chen[1,2] Meng, Xiaohui[1,2] Zhu, Lingqin[1] Cheng, Junwen[1] Wang, Yanbin[1] He, Liang[1] Huang, Qi[3] Zheng, Guangyao[4]

第一作者：Li, Chen

机构：[1] Key Laboratory of State Forest Food Resources Utilization and Quality Control, Zhejiang Academy of Forestry, Hangzhou, 310023, China; [2] Zhejiang Provincial Key Laboratory of Resources Protection and Innovation of Traditional Chinese Medicine, Zhejiang A&F University, Hangzhou, 311300, China; [3] School of Biological and Chemical Engineering, Zhejiang University of Science and Technology, Hangzhou, 310023, China; [4] Institute of Chemical Industry of Forestry Products, Chinese Academy of Forestry, Nanjing, 210042, China

年份：2026

卷号：393

外文期刊名：Separation and Purification Technology

收录：EI(收录号：20260820125410);Scopus(收录号：2-s2.0-105030440071)

语种：英文

外文关键词：Carbon dioxide - Entropy - Ethanol - Free energy - Genetic algorithms - Gibbs free energy - High temperature operations - Kinetics - Mass transfer - Mean square error - Supercritical fluid extraction

摘要：The medium-polar triterpenoids from the fruiting body of Ganoderma lucidum (MPTGL) were selectively released with polarity regulated supercritical CO2 extraction. Based on the well-designed response surface methodology (RSM), a genetic algorithm (GA) assisted artificial neural network (ANN) was applied to statistically optimize the polarity-palette process. The highest MPTGL yield was then achieved by GA-ANN at extraction pressure of 29.2 MPa, extraction temperature of 55 °C, duration time of 1.8 h, and cosolvent content of 3.5 mL/g. Under that scenario, the actual target reached 0.556% with 7 classical medium-polar compounds detected by HPLC covering Ganoderiols, Ganoderals and Ganoderones. The high coefficient of determination and low value of mean squared error illustrated an accuracy in the prediction of MPTGL output in GA-assisted ANN. Moreover, their kinetic and thermodynamic process were continuously conducted at the variation of extraction pressure (17.5–37.5 MPa) and temperature (313.15–353.15 K). The employed Sovová model could predict the extraction behaviors well with three typical periods and the increase of mass transfer coefficients apparently announced that certain operating pressure was conducive to improve the diffusion efficiency and mass transfer efficiency of MPTGL from the tissue to the ethanol-adjusted scCO2 solution. The lower Ea (6.82 kJ/mol) and negative value of Gibbs free energy revealed such extraction process for MPTGL was easily spontaneous and both positive enthalpy change (9.24 kJ/mol) and the entropy change (38.47 J/mol·K) elucidated an endothermic feature of the extraction, reinforcing the digestion of target from inner matrix under the optimized conditions. Those finding provided a possibility for high-valued MPTGL enrichments of G. lucidum involved with GA-ANN predicted extraction technique. ? 2026 Elsevier B.V.