应用化学 ›› 2025, Vol. 42 ›› Issue (5): 675-683.DOI: 10.19894/j.issn.1000-0518.240326

• 研究论文 • 上一篇    下一篇

基于改进遗传算法优化化学反应条件

田梦阳1, 刘建闽2()   

  1. 1.广西民族大学人工智能学院,南宁 530000
    2.广西财经学院广西财经大数据重点实验室,南宁 530000
  • 收稿日期:2024-10-18 接受日期:2025-04-03 出版日期:2025-05-01 发布日期:2025-06-05
  • 通讯作者: 刘建闽
  • 基金资助:
    广西财经大数据重点实验室项目(202405);广西研究生教育创新计划项目(JGY2024351)

Optimizing Chemical Reaction Conditions Based on Improved Genetic Algorithms

Meng-Yang TIAN1, Jian-Min LIU2()   

  1. 1.School of Artificial Intelligence,Guangxi Minzu University,Nanning 530000,China
    Guangxi Key Laboratory of Big Data in Finance and Economics,Guangxi University of Finance and Economics,Nanning 530000,China
  • Received:2024-10-18 Accepted:2025-04-03 Published:2025-05-01 Online:2025-06-05
  • Contact: Jian-Min LIU
  • About author:jmliu@gxufe.edu.cn
  • Supported by:
    Guangxi Key Laboratory of Big Data in Finance and Economics Project(202405);Guangxi Graduate Education Innovation Program(JGY2024351)

摘要:

有机化学反应条件的优化一直是化学领域的重要研究课题。 然而,由于反应条件的多样性和复杂性,传统优化方法通常依赖大量实验,面临高昂的成本和长时间反应的挑战。 本研究针对有机化学反应优化的特点及传统遗传算法在收敛速度和局部最优解问题上的不足,提出了一种改进的遗传算法优化模型。 该模型结合了保留精英策略、自适应多次变异策略和随机选择策略,显著提高了算法的全局搜索能力和收敛速度。 研究首先在一个包含混合类型条件的直接芳基化反应数据集上对模型进行评估,实验结果表明,相较于传统的遗传算法和随机搜索算法,改进的遗传算法在不同实验环境下表现出更强的寻优能力和更高的稳定性。 随后,研究以包含3696个反应条件的Suzuki-Miyaura反应数据集作为待优化的反应。 实验表明,在产率≥96.20%的反应条件仅占整个数据集1%的搜索空间的情况下,改进遗传算法平均仅需搜索35个样本即可找到符合该条件的最优反应条件,充分展示了改进遗传算法在化学反应条件优化上的巨大潜力。

关键词: 反应条件, 优化, 改进, 遗传算法, 变异

Abstract:

The optimization of organic chemical reaction conditions has always been an important research topic in the field of chemistry. However, due to the diversity and complexity of reaction conditions, traditional optimization methods often rely on a large number of experiments, facing challenges such as high costs and long reaction times. This study proposes an improved genetic algorithm optimization model addressing the characteristics of organic chemical reaction optimization and the shortcomings of traditional genetic algorithms in terms of convergence speed and local optima issues. The model combines an elitism strategy, adaptive multiple mutation strategy, and random selection strategy, significantly improving the algorithm's global search capability and convergence speed. The study first evaluates the model on a dataset of direct arylation reactions containing mixed-type conditions, showing that the improved genetic algorithm exhibits stronger optimization ability and higher stability compared to traditional genetic algorithms and random search algorithms. Subsequently, the study optimizes the Suzuki-Miyaura reaction using a dataset containing 3696 reaction conditions. The experimental results demonstrate that when reaction conditions with a yield of at least 96.20% account for only 1% of the entire search space, the improved genetic algorithm requires an average of only 35 samples to find the optimal reaction conditions that meet this criterion, fully demonstrating the immense potential of the improved genetic algorithm in optimizing chemical reaction conditions.

Key words: Reaction conditions, Optimization, Improvement, Genetic algorithm, Mutation

中图分类号: