Abstract:

The global resource optimization model typically uses a mathematical model represented by a linear programming model combined with a quadratic nonlinear module to represent the refining process of a refinery, which can solve the optimization problem of super multi-variable in the whole plant. The model has limited expressive power, and the optimization capability for the production of asphalt products from residue components is relatively poor, making it difficult to accurately execute the measurement plan and achieve optimal global resource calculation. This study proposes a modeling optimization method for asphalt production: using asphalt production data to train a linear regression model to represent the key index properties of asphalt products, and integrating it into the resource optimization model as a quadratic nonlinear constraint, to achieve global optimization. The research results can help enterprises to calculate the best crude oil blending formula to produce qualified asphalt products more accurately, and improve the benefit comparison calculation method for residue components. The modeling optimization method proposed in this study was applied to a certain refining enterprise, and the results show that using the asphalt blending modeling method in this study can further optimize the enterprise's crude oil resource allocation and ensure the quality of asphalt products.

Key words: Asphalt Product Property, Prediction, Global Resource Optimization Model, Atmospheric Distillation Unit

摘要：

全局资源优化模型通常采用线性规划模型搭配二次非线性模块的数学模型表征炼油厂加工流程，能解决全厂超多变量优化问题。优化模型中的常减压模块通常采用线性收率并固定原油品种，以约束各加工线路。模型表达能力有限，对于渣油组分生成沥青产品的优化能力较差，测算方案难以准确执行，无法做到全局资源最优计算。该研究提出了一种针对沥青生产的建模优化方式：使用沥青生产数据训练线性回归模型表征沥青产品关键指标性质，并集成到资源优化模型中，转化为二次非线性约束项，实现全局优化。研究结果能帮助企业更精准地测算最佳原油调配配方以生产合格沥青产品，并完善渣油组分效益比较测算方式。采用该建模优化方式对某炼油企业进行了测算，结果表明，采用该研究中的沥青调和建模方法能进一步优化企业的原油资源配置，保证沥青产品质量。

关键词: 沥青产品性质, 预测, 全局资源优化模型, 常减压装置