Abstract:

This paper mainly discusses two implementation methods of non-linear blending in the refined oil blending module of the enterprise planning optimization model: the property mixing index model and the “linear+interactive influence” model. This paper introduces the conversion relationship between strong nonlinear properties research octane number and octane number mixing index, and proposes a method for regression calculation of “linear+interactive influence” model parameters based on historical formula data. Based on real blending results data, a comparative analysis is conducted on the calculation results of the linear blending model, the property mixing index model, and the “linear+interactive influence” model. The results indicate that the method of using historical formula data regression to calculate the required interaction factor parameters for establishing a linear+interaction influence model in dealing with nonlinear blending problems can effectively reduce the calculation error of nonlinear properties in the blending process and reduce the workload of blending experiments.

Key words: refinery enterprise, gasoline, research octane number, nonlinear blending, linear addition, property index, interactive factor

摘要：

主要讨论了企业计划优化模型中成品油调合模块非线性调合的两种实现方式：性质混合指数模型和“线性+交互影响”模型。介绍了强非线性性质研究法辛烷值转换为辛烷值混合指数的转换关系式，提出了一种基于历史配方数据回归计算“线性+交互影响”模型参数的方法。基于真实的调合结果数据，对比分析了线性调合模型计算结果、性质混合指数模型计算结果、“线性+交互影响”模型计算结果。结果表明：在处理非线性调合问题时，采用通过历史配方数据回归计算出建立“线性+交互影响”模型所需的交互因子参数的方法，可以有效减少调合过程中非线性性质的计算误差，同时减少调合实验工作量。

关键词: 炼化企业, 汽油, 研究法辛烷值, 非线性调合, 线性加和, 性质指数, 交互因子