摘要
临界温度是一种非常关键的热物理性质,对其进行理论预测一直是热物性研究的热点。然而,早期预测模型往往不能有效区分工质同分异构体。本文借助机器学习算法,采用“分子指纹+拓扑指数”的新型分子结构描述方法表达工质的分子结构并建立临界温度模型,在测试集预测中的绝对平均偏差为3.99%,表明本文模型具有良好的预测能力。本文模型与文献对比的结果表明,新模型不仅可以有效区分工质同分异构体,在计算精度方面也超越了现有其他模型。
The critical temperature is a very critical thermophysical property,and its theoretical prediction has always been a hot topic in thermophysical property research.However,the previously reported models cannot effectively distinguish the isomers in working fluids.To solve this problem,the new structure description‘molecular fingerprints+topological index’was considered in the predicted model,the average absolute deviation between the predicted and experimental values is 3.99%,which proves the reasonable prediction performance of the proposed model.Then the estimation accuracy of the model was compared with previous models,the results indicated that the proposed model can not only effectively distinguish the isomers of working fluids,but also surpass other existing models with respect to accuracy.
作者
任嘉辉
刘豫
刘朝
刘浪
李莹
REN Jiahui;LIU Yu;LIU Chao;LIU Lang;LI Ying(Key Laboratory of Low-grade Energy Utilization Technologies and Systems,Ministry of Education,School of Energy and Power Engineering,Chongqing University,Chongqing 400030,China;State Key Laboratory of Pollution Control and Resource Reuse,Nanjing 210023,Jiangsu,China;CNNC Key Laboratory on Nuclear Reactor ThermoHydraulics Technology,Nuclear Power Institute of China,Chengdu 610213,Sichuan,China)
出处
《化工学报》
EI
CAS
CSCD
北大核心
2022年第4期1493-1500,共8页
CIESC Journal
基金
国家自然科学基金项目(51876015,52106218)
污染控制与资源化研究国家重点实验室开放基金(PCRRF19038)。
关键词
工质
热力学性质
临界温度
拓扑指数
机器学习
神经网络
预测
working fluids
thermodynamic properties
critical temperature
topological index
machine learning
neural networks
prediction