动态进化环境在组卷中的建模与应用  

Modeling and application of dynamic evolutionary environment in test paper generating problem

作　　者：肖桂霞[1] 

机构地区：[1]常德职业技术学院现代教育技术中心,湖南常德415000

出　　处：《微型机与应用》2015年第2期77-79,82,共4页Microcomputer & Its Applications

摘　　要：针对遗传算法组卷易陷入早熟、难以收敛的问题进行研究,结合进化环境对进化过程的影响和引导,对动态进化环境进行建模,提出了一种基于动态变异池的策略。该策略的种群不共享变异池,在每次变异前,根据每个个体的弱点动态生成该个体的变异基因库,以此改善当前变异环境,实施引导性变异,提高解质量。该策略能加速收敛,并在很大程度上提高收敛精度。实验数据表明,采用了该策略的组卷算法能快速生成各项指标都与约束条件十分贴近的试卷,具有很好的实用价值。Considering the effects and guidance of environment on the evolutionary process, the dynamic evolutionary environment is modeled and a Dynamic Mutation Pool （DMP） method is proposed to overcome premature and slow convergence defects of genetic algorithms when generating a test paper intelligently. In DMP, the mutation pool of the population is not shareable. Every individual has its own mutation pool, and it is generated dynamically according to the performance before each mutation operation. DMP can improve the mutation environment, thus it significantly improves the quality of solutions, and further enhances the convergence of the algorithm. The comparative experiment shows that an algorithm which adopts this dynamic mutation strategy can generate a satisfying test paper quickly.

关 键 词：组卷 进化环境 遗传算法 动态变异池 收敛精度 题库系统 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术] TP301.6[自动化与计算机技术—计算机科学与技术]