基于Optuna优化的遗传算法智能组卷模型  

Intelligent Exam Paper Generation Model Based on Optuna Optimized Genetic Algorithm

作　　者：常宸 胡安波 高鹏 CHANG Chen;HU Anbo;GAO Peng(Military Human Resource Support Center,Political Work Department of Central Military Commission,Beijing 100034,China)

机构地区：[1]中央军委政治工作部军事人力资源保障中心,北京100034

出　　处：《中北大学学报(自然科学版)》2025年第2期208-218,共11页Journal of North University of China(Natural Science Edition)

基　　金：国家社会科学基金重点项目(2022-SKJJ-B-072);国防科技战略先导计划(21-ZLXD-02-00-02-006-38)。

摘　　要：为了提升智能组卷性能,同时解决常规的基于遗传算法优化的智能组卷模型在实际应用时参数难以确定,在面对不同规模和特征分布的题库时性能不稳定的问题,提出了基于Optuna优化的遗传算法智能组卷模型。通过设计分层格雷编码来克服传统二进制、十进制编码引发的汉明悬崖问题,通过Optuna优化自反馈确定遗传算法的种群规模、迭代次数及其他参数,动态调整遗传算法的交叉、变异速率,实现对组卷搜索空间的自适应调整。实验结果表明,所提模型能够有效确定参数并实现动态调整,组卷质量优于其他基于随机和启发式算法的智能组卷模型。In order to improve the performance of intelligent exam paper generation,and solve the prob-lem that the parameters of the genetic algorithm based model are difficult to determine in practice,and the performance is unstable in the face of question banks with different sizes and feature distributions,an intel-ligent exam paper generation model based on Optuna optimized genetic algorithm was proposed.By designing hierarchical gray coding,the Hamming cliff problem caused by traditional binary and decimal encoding methods was overcome.The population size,number of iterations,and other parameters of the genetic algorithm were determined by Optuna optimization self-feedback model,and the crossover and mutation rate of the genetic algorithm were dynamically adjusted to achieve adaptive adjustment of the exam paper generation search space.Experimental results show that the proposed algorithm can effectively determine the parameters,and realize dynamic adjustment.Final exam paper generation quality is better than other models based on random and heuristic algorithms.

关 键 词：智能组卷 遗传算法 Optuna 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]