基于随机森林算法的绿色丙二醇工艺过程模拟和预测

Random Forest Algorithm-Based Modeling for Green Propylene Glycol Synthesis

中文关键词: 绿色丙二醇  智能优化  绿氢绿氧  机器学习  随机森林算法

英文关键词: Propylene glycol  Intelligent optimization  Green hydrogen and oxygen  Machine learning  Random forest algorithm

中文摘要:

      为实现“双碳”目标，应对现有丙二醇生产工艺对化石原料依赖度高、碳排放量大的问题，本研究提出一种可再生能源驱动的绿色合成新工艺。针对新能源电力波动特性，该研究开发了基于数据驱动的智能优化系统，以克服传统稳态模拟在动态响应方面的局限。 通过构建涵盖电解水制氢、二氧化碳催化转化与丙二醇合成的全流程集成模型，并运用随机森林算法建立工艺参数间的智能关联网络，实现了从原料到产品的全流程动态模拟与推演。在此基础上，研究将一氧化碳制备、3-羟基丙酸甲酯合成与丙二醇生产三个工段进行模块化建模，构建了氢气投加量与丙二醇产量的非线性预测模型。结果表明，该方法可有效实现可再生能源与化工过程的动态匹配，为传统化工绿色化改造提供智能化解决方案，对推动流程工业低碳转型具有重要参考价值。

英文摘要:

      To support the dual carbon goals and address the high reliance on fossil fuels and significant carbon emissions in conventional propanediol production, this study proposes a novel green synthesis process driven by renewable energy. In response to the fluctuating nature of new energy power sources, a data-driven intelligent optimization system is developed to overcome the limitations of traditional steady-state simulations in handling dynamic responses. An integrated process model is constructed, covering hydrogen production via water electrolysis, catalytic conversion of carbon dioxide, and propanediol synthesis. By applying a random forest algorithm, an intelligent correlation network among process parameters is established, enabling dynamic simulation from raw materials to final products. Furthermore, the three key stages—carbon monoxide preparation, methyl 3-hydroxypropionate synthesis, and propanediol production—are modularly modeled, resulting in a nonlinear predictive model linking hydrogen input to propanediol yield. The results demonstrate that this approach effectively achieves dynamic matching between renewable energy and chemical production, providing an intelligent solution for greening traditional chemical processes and offering valuable insights for promoting low-carbon transformation in the process industry.