项目名称: 混合物定量结构性质关系的化学计量学方法研究
项目编号: No.21305108
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 数理科学和化学
项目作者: 焦龙
作者单位: 西安石油大学
项目金额: 25万元
中文摘要: 混合物体系因组分间相互作用导致其性质并非各组分性质的简单加和。进行混合物的定量结构性质关系(QSPR)研究,对于完善QSPR理论及扩大QSPR方法的应用范围有重要的意义。本项目拟研究用于建立混合物QSPR模型的化学计量学方法。计划在量子化学参数、拓扑指数及分子结构片段描述符基础上,结合红外、紫外可见、荧光、核磁、质谱等波谱分析手段,构建新型定量结构描述符用于混合物体系中结构、组成和相互作用等化学信息的表达。探索以数据融合方法为基础的混合物描述符使用策略,以更有效的利用描述符中包含的化学信息。研究随机森林方法在混合物QSPR研究中的应用,以发展一种有效的混合物QSPR数学建模方法。通过研究烃类混合物的物理性质QSPR模型以及氯代烃混合物、多氯联苯混合物和多溴联苯醚混合物的毒性QSPR模型,构建较系统的用于建立混合物QSPR模型的方法,为混合物QSPR研究提供一种新的途径。
中文关键词: 定量结构-性质关系;混合物;拓扑指数;人工神经网络;波谱区间选择
英文摘要: Because of the interaction between the components, the properties of a mixture may not be the direct sum of each component's properties. Studying the quantitative structure property relationship (QSPR) of mixtures is of great importance for developing the QSPR theory and promoting the application of QSPR method. Presently, it has become one of the most important tasks of chemometrics. Thus, this project intends to investigate the methods for developing the QSPR model of mixtures. The method which combines quantum chemistry descriptor, topological index, molecular fragmental descriptor, IR, UV-Vis, fluorescence, NMR and MS analysis is proposed to quantitatively describe the chemical information of mixtures. A strategy based on data fusion methods is proposed in order to acquire more chemical information from the descriptors of mixtures. Moreover, the application of random forest method to modeling the quantitative relationship between the descriptors and properties of mixtures will be investigated in order to find an effective mathematical modeling method. The QSPR model for the physical property of hydrocarbon mixtures and the QSPR models for the toxicity of chlorinated hydrocarbon mixtures, PCBs mixtures and PBDEs mixtures will be investigated. In summary, the general objective of this project is providing a co
英文关键词: quantitative structure-property relationship;mixture;topological index;artificial neural network;spectral interval selection