摘要本文采用差示扫描量热（DSC）、绝热量热仪（ARC）、热筛选仪（TSu）和数字熔点仪（WRS-1）以及定量构效关系（QSPR）基本原理对偶氮二异丁腈（AIBN）和偶氮二环己基甲腈（ACCN）的熔点进行了实验研究和理论预测。64162

通过动态DSC、ARC、TSu对AIBN和ACCN热分解特性进行研究，发现这两种物质均在熔化前开始分解放热反应，并得到了其在上述三种实验条件下出现相变现象的温度。通过等温DSC对其在等温条件下的热行为进行研究，发现AIBN在其实验条件（80℃，82℃，86℃）下出现了相变现象，且其分解反应定律随温度的变化而变化，而ACCN在其实验条件（98℃，100℃，103℃）下表现为两步反应。通过基于毛细管法的WRS-1得到了AIBN，ACCN熔点值分别为98.77℃和112.57℃。

运用线性建模方法（遗传算法GA和最佳多元线性回归MLR）建立了125种含氮类化合物的熔点预测模型，采用R2、RMSE、Q2loo、Q2ext进行了模型的评价。筛选出的与含氮类化合物熔点密切相关的结构参数为ZM2V和H-050，其中训练集和测试集的R2分别为0.6338和0.7031，均方根误差RMSE分别为58.1028和38.4679，交互验证参数Q2loo 为0.6388，Q2ext为0.7769。分析可知，该模型具有一定稳健性和预测能力；影响含氮类物质熔点的主要因素可能是原子质量分布与连接信息、化学键的分支度、特定结构碎片形成氢键的能力等方面的结构特点；AIBN，ACCN基于该模型的理论预测值分别为51.39℃，84.37℃。

毕业论文关键词  含氮类化合物 熔点 差示扫描量热 绝热量热仪 热筛选仪 等温法 定量构效关系 遗传算法

毕业设计说明书（论文）外文摘要

Title    The experimental research and prediction analysis   of the melting point of two kinds of azo compound                                                

Abstract Differential scanning calorimeter（DSC）, accelerating rate calorimeter (ARC), thermal screening unit (TSu), digital melting point apparatus（WRS-1）, and the quantitative structure-property relationship(QSPR) study was used to make experimental research and theoretical forecast on the melting point of azodiisobutyronitrile (AIBN) and 1,1'-Azobis(cyclohexane-1-carbonitrile) (ACCN).

Thermal decomposition characteristics of AIBN and ACCN was studied by non-isothermal DSC，ARC and TSu, and the temperature of its phase transition occurs in the a conditions of above three kinds of experiment was obtained.

Studying its thermal behavior under isothermal conditions by isothermal DSC, the phase change phenomenon AIBN was found under the experimental condition (80 ℃, 82 ℃, 86 ℃), and the law of decomposition reaction varies with temperature, while the ACCN in the experimental condition (98 ℃, 100 ℃, 98 ℃) is characterized by two steps reaction. The melting point value of AIBN and ACCN was obtained by the operation of melting point apparatus based on the capillary method, which is 98.77℃ and 112.57 ℃ respectively.

by employing linear modeling method(the variable selection method of genetic algorithm(GA) and multiple linear regression(MLR)),the melting point prediction model of 125 kinds of nitrogenous compounds was set up, and the evaluation of the model was carried on by using R2、RMSE、Q2loo、Q2ext.Descriptors which have significant contribution to the melting point selected are ZM2V and H-050．The multiple correlation coefficient(R2) of training set and test set are 0.6338 and 0.7031 respectively, RMSE are 58.1028 and 38.4679 respectively, and the interactive authentication parameters Q2loo is 0.6388 while Q2ext is 0.7769. The results show a certain robustness and prediction ability. The main factors affecting the melting point of nitrogenous compounds could be structure characteristics such as atomic mass distribution and connection information, the structure of the branching degree of chemical bond, specific structure fragments' ability to form hydrogen bonds, etc. The predicted melting point value of AIBN and ACCN based on the theory of the model were 51.39 ℃ and 84.37 ℃ respectively.