大学化学 >> 2025, Vol. 40 >> Issue (9): 19-24.doi: 10.12461/PKU.DXHX202408095

所属专题： AI赋能化学教育

专题 上一篇    下一篇

改进的模拟退火算法预测有机化合物分子式

陈晓东, 张玉敏   

  1. 吉林大学化学学院, 长春 130012
  • 收稿日期:2024-08-24 录用日期:2024-10-23 发布日期:2025-09-16
  • 通讯作者: 陈晓东 E-mail:cxd@jlu.edu.cn
  • 基金资助:
    2021年吉林大学实验技术项目(SYXM2021b001)；2021年吉林大学本科教学改革研究项目(2021XZC031)；2023年吉林大学本科教学改革研究重点项目(2023XZD039)

An Improved Simulated Annealing Algorithm for Predicting the Molecular Formulas of Organic Compounds

Xiaodong Chen, Yumin Zhang   

  1. College of Chemistry, Jilin University, Changchun 130012, China
  • Received:2024-08-24 Accepted:2024-10-23 Published:2025-09-16
  • Contact: Xiaodong Chen E-mail:cxd@jlu.edu.cn

PDF

152

赞

摘要： 模拟退火算法是人工智能组合优化算法，在此算法的基础上，提出了一种改进的模拟退火算法，用于预测有机化合物分子式。算法开始，设计使用遗传算法计算种群各个体的适应度函数值，从中选择最优个体作为模拟退火算法初始解。然后在这个初始解的基础上，随机扰动生成新解，并计算其适应度函数值。若适应度函数值的增量小于等于零，则接受新解，否则按Metropolis准则判断是否接受新解。随着退火温度的缓慢降低，依据算法终止条件判断是否搜索到全局最优解。实验证明，该算法提高了搜索到全局最优解的成功率。将其用于预测有机化合物分子式时，其适应度函数收敛性明显优于经典模拟退火算法。

关键词: 模拟退火算法, 遗传算法, 组合优化, 质量分数, 分子式

Abstract: Simulated annealing algorithm is an artificial intelligence combinatorial optimization algorithm. Building upon the classic simulated annealing algorithm, we propose an enhanced version for predicting the molecular formulas of organic compounds. The algorithm begins by using a genetic algorithm to calculate the fitness values of individuals in the population, selecting the optimal individual as the initial solution for the simulated annealing process. Based on this initial solution, new solutions are generated through random perturbation, and their fitness values are calculated. If the change in fitness is less than or equal to zero, the new solution is accepted. Otherwise, the Metropolis criterion is applied to determine whether the new solution should be accepted. As the annealing temperature gradually decreases, the algorithm’s termination condition is used to determine if the global optimal solution has been found. Experimental results show that this improved algorithm increases the success rate of finding the global optimal solution. When applied to predict the molecular formulas of organic compounds, it demonstrates significantly better convergence of the fitness function compared to the classical simulated annealing algorithm.

Key words: Simulated annealing algorithm, Genetic algorithm, Combination optimization, Mass fraction, Molecular formulas