QSPR Study On Flash Point、 Boiling Point And Viscosity Of Organic Compounds

Quantitative structure-property relationship (QSPR) study is an advanced method for prediction of properties, which is newly developed along with the formation of chemoinformatics subject in recent years. QSPR is based on the basic law in the chemistry that molecular properties are determined by molecular structures, and studies the intrinsic quantitative relation between molecular structures and properties of organic compounds. By relating the properties of interest to the structure parameters and developing corresponding quantitative models, QSPR can be used to predict various physicochemical properties of organic compounds from their molecular structures alone and make up for the disadvantages of experiments.In this thesis, basic QSPR principle was used to build the models for predicting the flash point、boiling point and viscosity of organic compounds. Meanwhile, find out the structure factors which have significant contribution to these three properties. The main works carried out in this thesis are summarized as follows:1. Three models were successfully established by Heuristic method(HM)、Best multilinear regression method(B-MLR) and Support vector machine(SVM) for predicting the flash point and boiling point of hydrocarbon compounds. All the models had good performance in goodness-of-fit, stability and generalization ablity, and the SVM model was more excellent compared to the other models.2. A series of derivatives were joined in the above sample of hydrocarbons. Then the B-MLR and SVM models were built for predicting the flash point and boiling point of hydrocarbon and its derivatives. The predicted results indicated that the two models had great prediction effect, which can achieve the requirement of prediction. As compared to the models of hydrocarbons, the performance of the models which added derivatives has been decreased.3. Three QSPR models were successfully established by HM、B-MLR and SVM for predicting the viscosity of organic compounds. The predicted results indicated that the models had robustness, strong generative ability and small prediction error. The predicted results were satisfactory. It can be used to predict the viscosity of the organic which is new or has not been synthesized yet.4. The mechanism of each model was interpreted. The results showed that the molecular size and branching, as well as the molecular electrostatic effects are the dominant structural factors that were relevant to the flash point of organic compounds. The flash point increases with increasing molecular bulk and branching, as well as increasing molecular electrostatic effects.5. The results showed that the molecular size and component, as well as the molecular hydrogen bonding effects and electrostatic effects are the dominant structural factors that were relevant to the boiling point of organic compounds. The boiling point increases with increasing molecular size and bulk, as well as increasing possibility of formation of hydrogen bonding and electrostatic effects.6. The results showed that the molecular size and component, as well as the molecular hydrogen bonding effects and polarity effects are the dominant structural factors that were relevant to the viscosity of organic compounds. The viscosity increases with increasing molecular bulk and the number of rings, decreasing number of F atoms which represents atomic electronegativities, as well as increasing possibility of formation of hydrogen bonding and polarity effects.