论文部分内容阅读
流体的pVT性质在化工理论基础和应用研究中都具有重要的意义,通常采用经验关系式或基于一定理论模型的状态方程来描述.而流体性质与其结构存在着内在的、本质的联系,因此用于描述分子结构的分子拓扑指数可进行结构/性质的定量关联.本文基于分子拓扑指数,采用人工神经网络技术来研究烷烃的pVT性质.并假定物质的pVT性质与分子拓扑指数呈以下函数关系V=f(p,T,~mX_t)(1)函数f无统一的解析表达式且为一非线性函数.将分子拓扑指数对pVT性质的影响视为一个由拓扑指数空间到pVT性质空间的一个映射,即V=N(p,T,~mX_t)(2)在此映射关系中,N为神经网络,其权重参数可由部分物质的已知实验数据经神经网络训练后抽提得到,进而可对其他物质的pVT性质进行预测.1 分子拓扑指数的选择和计算
The pVT properties of fluids are of great significance both in theory and application of chemical engineering, and are usually described by empirical equations or state equations based on theoretical models. However, fluid properties are intrinsically and intrinsically linked to their structure, The molecular topological index describing the molecular structure can be quantitatively related to the structure / property.In this paper, based on the molecular topological index, artificial neural network technology is used to study the pVT properties of alkanes, and the pVT property of the material is assumed to have the following functional relationship with the molecular topological index V = f (p, T, ~ mX_t) (1) The function f does not have a uniform analytical expression and is a nonlinear function. The effect of the molecular topological index on the pVT property is treated as one from the topological index space to the pVT nature space (2) In this mapping, N is a neural network whose weight parameters can be extracted from the known experimental data of some materials by neural network training, and then can be extracted Predict the pVT properties of other substances.1 Selection and calculation of molecular topological indices