基于原子类型电性拓扑状态指数的最小点火能预测模型研究  

Research on Minimum Igniting Energy Prediction Model Basedon Atomic Type Electrotopological-State Indices

作　　者：王贝贝 张以晨 刘新[1] 李丽[1] 关忠慧 WANG Beibei;ZHANG Yichen;LIU Xin;LI Li;GUAN Zhonghui(College of Jilin Emergency Management,Changchun Institute of Technology,Changchun 130012,China)

机构地区：[1]长春工程学院吉林应急管理学院,长春130012

出　　处：《长春工程学院学报（自然科学版）》2024年第1期112-120,共9页Journal of Changchun Institute of Technology：Natural Sciences Edition

基　　金：吉林省教育厅项目(12021Z016);吉林省科技厅项目(120220057)。

摘　　要：针对危险化学品最小点火能的试验测量存在一定的困难且具有很大不确定性的问题,收集了61种化学品的最小点火能试验值,并仅根据其分子结构计算了原子类型电性拓扑状态指数(ETSI),最终选择了23种ETSI作为分子描述符进行建模。首先,尝试基于61种危险化学品的最小点火能值,通过多元线性回归和支持向量机进行QSPR建模,但所建模型的内部稳定性很差,不满足基本的要求。经分析可能是因为描述符太多而样本相对较少造成的。通过从数据集中删除13个最小点火能值,ETSI的种类减少到14个,并采用相同的方法进行QSPR建模。经验证,新建模型在拟合能力、内部稳定性和外部预测能力3个方面均具有优异的性能,可以在不需要购买任何昂贵的软件和硬件的前提下,方便、快捷地预测化学品的最小点火能。There are certain difficulties and uncertainties in the experimental measurement of the minimum ignition energy of hazardous chemicals.The minimum ignition energy test values of 61 chemicals are collected,and the atomic type electrical topological state index(ETSI)is calculated only based on their molecular structure.Finally,23 ETSI are selected as molecular descriptors for modeling.Firstly,an attempt is made to model QSPR using multiple linear regression and support vector machine based on the minimum ignition energy values of 61 hazardous chemicals.However,the internal stability of the constructed model is poor and does not meet the basic requirements.After analysis,it may be due to too many descriptors and relatively few samples.By removing 13 minimum ignition energy values from the dataset,the number of ETSI types is reduced to 14,and the same method is used for QSPR modeling.After verification,the new model has excellent performance in three aspects:fitting ability,internal stability,and external prediction ability.It can conveniently and quickly predict the minimum ignition energy of chemicals without the need to purchase any expensive software and hardware.

关 键 词：最小点火能 定量结构—性质关系模型 ETSI 多元线性回归 支持向量机 

分 类 号：O6[理学—化学] X932[环境科学与工程—安全科学]