Quantitative structure-property relationship (QSPR) study for predicting gas-liquid critical temperatures of organic compounds

Gas-liquid critical temperature is an important parameter of critical state. Organic compounds are under rapid phase changes leading to explosions when conditions are changed at their critical states. Therefore, for safety purposes it is important to study the gas-liquid critical properties for different organic compounds, especially their critical temperatures. In this work, critical temperatures of 692 organic compounds were collected and applied to build quantitative structure-property relationship (QSPR) models. Dragon software was used to obtain their molecular structure information. Methods of multiple linear regression (MLR) and support vector machine (SVM) were applied to build the models, combined with genetic algorithm method. Between these two models, the MLR model has better internal robustness and the SVM model has better goodness-of-fit predictive ability. The results show the developed models have great performance in predicting the gas-liquid critical temperatures. With these models, critical temperatures of organic compounds can be predicted solely based on their molecular structures.