CAJ | 학술논문

本文采用MINIFLASH FLPL全自动闪点测试仪对不同组成和配比的288组二元互溶有机混合液体的闪点进行了测定．根据气液平衡理论，计算了各混合液体的正常沸点、标准蒸发焓以及蒸气相中混合气体的化学计量浓度和平均碳原子数，并将它们作为输入参数，闪点实验数据作为输出参数，采用多元线性回归和多元非线性回归方法对二者之间的内在相关性进行研究，建立了二元互溶可燃混合液体闪点的预测模型．两种方法对训练集的预测平均绝对误差分别为1．6377 K和1．6458 K ，对测试集的预测平均绝对误差分别为2．5056 K和2．5373 K，均在实验允许误差范围之内．2种模型的预测性能基本相同，但考虑到模型的直观性和可解释性，线性模型则具有明显的优势，更适用于二元互溶可燃液体闪点的预测．
본문채용MINIFLASH FLPL전자동섬점측시의대불동조성화배비적288조이원호용유궤혼합액체적섬점진행료측정．근거기액평형이론，계산료각혼합액체적정상비점、표준증발함이급증기상중혼합기체적화학계량농도화평균탄원자수，병장타문작위수입삼수，섬점실험수거작위수출삼수，채용다원선성회귀화다원비선성회귀방법대이자지간적내재상관성진행연구，건립료이원호용가연혼합액체섬점적예측모형．량충방법대훈련집적예측평균절대오차분별위1．6377 K화1．6458 K ，대측시집적예측평균절대오차분별위2．5056 K화2．5373 K，균재실험윤허오차범위지내．2충모형적예측성능기본상동，단고필도모형적직관성화가해석성，선성모형칙구유명현적우세，경괄용우이원호용가연액체섬점적예측．
The flash points of 288 sets of binary organic mixtures composed of different components and the volume ratios were measured by the MINIFLASH FLPL fully automatic flash point tester in this re-search .According to the vapor-liquid equilibrium theory ,the normal boiling point and the standard en-thalpy of vaporization of liquid mixtures ,as well as the average number of carbon atoms and the stoi-chiometric concentration of the gas phase for the mixtures were calculated .And then ,the four affecting factors characterized the liquid and gas mixtures and the flash points of binary liquid mixtures were used as the input and output parameters respectively .The quantitative relationship between them was studied by the multiple linear regression analysis (MLR) and the multiple nonlinear regression method (MNR) , and the corresponding flash point prediction models of binary liquid mixtures were established .For the training set ,the average absolute deviations between the experimental and predicted values of flash point were 1 .6377 K by the linear regression analysis method and 1 .6458 K by the nonlinear regression meth-od ,while for the testing set ,they were 2 .5056K and 2 .5373K ,respectively .The results showed that the predicted flash points of the two models were in good agreement with the experimental data .The performances of both the MLR and MNR models were almost the same ,but the MLR model has obvious advantage in the directviewing and interpretability ability which is more appropriate to predict the flash points of binary liquid mixtures .