催化学报 ›› 2025, Vol. 74: 211-227.DOI: 10.1016/S1872-2067(25)64733-4

• 论文 • 上一篇    下一篇

基于机器学习和优化算法的数据驱动框架预测用于合成气制烯烃的氧化物-沸石基复合催化体系和反应条件

Mansurbek Urol ugli Abdullaeva,b, Woosong Jeona, Yun Kangc, Juhwan Nohd, Jung Ho Shind, Hee-Joon Chune,*(), Hyun Woo Kimc,*(), Yong Tae Kima,b,*()   

  1. a韩国化学技术研究院氢能与C1气体研究中心, 大田, 韩国
    b科技大学(UST)先进材料与化学工程系, 大田, 韩国
    c光州科学技术院化学系, 光州, 韩国
    d韩国化学技术研究院化学数据驱动研究中心, 大田, 韩国
    e忠南大学化学系, 大田, 韩国

Data-driven framework based on machine learning and optimization algorithms to predict oxide-zeolite-based composite and reaction conditions for syngas-to-olefin conversion

Mansurbek Urol ugli Abdullaeva,b, Woosong Jeona, Yun Kangc, Juhwan Nohd, Jung Ho Shind, Hee-Joon Chune,*(), Hyun Woo Kimc,*(), Yong Tae Kima,b,*()   

  1. aHydrogen & C1 Gas Research Center, Korea Research Institute of Chemical Technology, Daejeon 34114, Republic of Korea
    bDepartment of Advanced Materials and Chemical Engineering, University of Science and Technology (UST), Daejeon 34113, Republic of Korea
    cDepartment of Chemistry, Gwangju Institute of Science and Technology (GIST), Gwangju 61005, Republic of Korea
    dChemical Data-Driven Research Center, Korea Research Institute of Chemical Technology, Daejeon 34114, Republic of Korea
    eDepartment of Chemistry, Chungnam National University, Daejeon 34134, Republic of Korea
  • Received:2024-12-30 Accepted:2025-04-15 Online:2025-07-18 Published:2025-07-20
  • Contact: *E-mail: hjchun@cnu.ac.kr (H.-J. Chun), hwk@gist.ac.kr (H. W. Kim), ytkim@krict.re.kr (Y. T. Kim).

摘要:

氧化物-沸石基双功能复合材料(OXZEO)已成为合成气直接制烯烃的潜在高效材料. 然而, 反应参数的实验筛选和优化过程耗时耗物. 为了应对该挑战, 本文提出了一种集成机器学习、贝叶斯优化与实验验证的三阶段数据驱动框架. 采用随机森林、梯度提升决策树和极限梯度提升等集成树模型(R² > 0.87), 成功预测出Zn-Zr和Cu-Mg双元混合氧化物为最优OXZEO体系. 实验验证表明, 该体系与HSAPO-34沸石物理混合后, 轻质烯烃时空收率达3.8-4.6 mmol·g-¹·h-¹. 密度泛函理论计算揭示了Zn-Zr与Cu-Mg氧化物的活性差异机制. 本研究开发的框架通过机器学习模型解析16个催化剂与反应条件描述符间的构效关系, 其中氧化物/沸石质量比、反应温度和压力是影响催化性能的关键参数. 结合贝叶斯优化实现多目标参数寻优, 为催化过程设计与优化提供了高效工具, 可推广至其他复杂催化体系的研究.

关键词: 合成气制烯烃, 氧化物-沸石基复合物, 机器学习, 贝叶斯优化, 催化剂和反应工程开发, 反应条件优化, 密度泛函理论

Abstract:

Bifunctional oxide-zeolite-based composites (OXZEO) have emerged as promising materials for the direct conversion of syngas to olefins. However, experimental screening and optimization of reaction parameters remain resource-intensive. To address this challenge, we implemented a three-stage framework integrating machine learning, Bayesian optimization, and experimental validation, utilizing a carefully curated dataset from the literature. Our ensemble-tree model (R2 > 0.87) identified Zn-Zr and Cu-Mg binary mixed oxides as the most effective OXZEO systems, with their light olefin space-time yields confirmed by physically mixing with HSAPO-34 through experimental validation. Density functional theory calculations further elucidated the activity trends between Zn-Zr and Cu-Mg mixed oxides. Among 16 catalyst and reaction condition descriptors, the oxide/zeolite ratio, reaction temperature, and pressure emerged as the most significant factors. This interpretable, data-driven framework offers a versatile approach that can be applied to other catalytic processes, providing a powerful tool for experiment design and optimization in catalysis.

Key words: Syngas-to-olefin, Oxide-zeolite-based composite, Machine learning, Bayesian optimization, Catalyst and reaction engineering discovery, Reaction condition optimization, Density functional theory