中文摘要：

使用两个热力学模型NRTL-SAC和COSMO-RS预测胆甾醇和链甾醇在混合溶剂中的溶解度,并以实验数据为计算基准,通过平均相对偏差（ARD）评价模型的预测效果.为填补溶解度数据空白,采用静态法结合高效液相色谱法（HPLC）测定了298.2 K下两种甾醇分别随混合溶剂组成变化的溶解度.NRTL-SAC对两种甾醇的溶解度预测在满意范围,最大ARD为50.58％.COSMO-RS预测结果非常不理想.通过改进,输入一个298.2 K时的实验值预测同一混合溶剂组成下随温度变化的溶解度,结果预测值与实验值吻合得较好,胆甾醇和链甾醇在（正己烷＋乙醇）混合溶剂中,总的ARD分别从80.57％和82.30％降为12.77％和21.32％.上述结果表明,只需少量实验数据的NRTL-SAC和只需一个实验数据的COSMO-RS （ref）可用于两种甾醇溶解度的预测,为工业分离纯化过程中的溶剂快速筛选提供重要的理论依据.

英文摘要：

Solubility of cholesterol and desmosterol in mixed solvents was predicted via two thermodynamic models, NRTL-SAC and COSMO-RS. With experimental data as the basis, average relative deviation （ARD） was used to evaluate the prediction effects of the models. To fill the blank of solubility data, the solubilities of two sterols at 298.2 K in mixed solvent with different compositions were determined using the static method combined with high performance liquid chromatography （HPLC）. The ARDs by NRTL-SAC for two sterols in mixed solvents are less than 50.58%. The prediction results by COSMO-RS are extremely dissatisfied, but the predicted values satisfactorily agree with the experimental values by using an experimental datum. The total ARDs for cholesterol and desmosterol in n-hexane ＋ ethanol mixed solvents decrease to 12.77% and 21.32% from 80.57% and 82.30%, respectively. The results show that NRTL-SAC from limited experimental data and COSMO-RS with only one experimental datum can be used for solubility prediction {or two sterols, and are useful in rapid selection of suitable solvents for industrial separation and purification processes.