考虑催化剂失活的环氧乙烷反应器机理与数据模型
Mechanism and data model of ethylene oxide reactors considering catalyst deactivation
黄纯西 1洪小东 1黄波 2严玲晴 2廖环宇 2阳永荣 1王靖岱 1廖祖维1
作者信息
- 1. 浙江大学 化学工程与生物工程学院,浙江 杭州 310058
- 2. 阿里云智能,浙江 杭州 311121
- 折叠
摘要
环氧乙烷是乙烯工业的重要衍生物之一,主要由乙烯在银催化剂作用下直接氧化生产.针对催化剂活性缓慢衰减,需定期调整操作条件以维持产量和高选择性的问题,提出了机理模型和数据模型耦合的方法,以确定催化剂的活性变化和各操作条件的影响.以某工业环氧乙烷反应器为对象,通过阿里云智能制造平台挖掘生产数据,分别使用反应动力学、失活动力学、反应器模型和支持向量回归,建立了各操作条件与反应器出口组成的模型.结果表明,在输入变量中添加机理模型所得的活性变量后,支持向量回归模型较原始模型更加准确,计算的反应器出口组成和选择性在测试集上的平均绝对百分比误差小于 3%,为优化选择性提供了基础.
Abstract
Ethylene oxide(EO)is one of the important derivatives in the ethylene industry and is primarily produced by direct oxidation of ethylene using silver catalysts.Due to slow deactivation of catalysts and requirements of maintaining yield and high selectivity by adjusting operating conditions periodically,a method of coupling mechanism model and data model was proposed to determine the changes in catalyst activity and the impact of various operating conditions.Using Alibaba Cloud's Intelligent manufacturing platform to mine production data of an industrial ethylene oxide reactor,models of various operating conditions and reactor outlet composition were established using reaction kinetics,deactivation kinetics,reactor model,and support vector regression,respectively.The results show that after adding the activity variables obtained from the mechanism model to the input variables,the support vector regression model is more accurate than the original model,and the mean absolute percentage error of calculated reactor outlet composition and selectivity on the test set is less than 3%,which provides a basis for optimizing selectivity.
关键词
环氧乙烷/催化剂失活/反应动力学/反应器模型/支持向量回归Key words
ethylene oxide/catalyst deactivation/reaction kinetics/reactor model/support vector regression引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据