基于模拟退火遗传算法的自动组卷系统研究

Research on autogenerating test paper based on simulating annealing genetic algorithm

从题库中抽出一组满足多项要求的试题是一个组合优化问题,针对该问题,比较了目前几种组卷算法的特点,提出把一种实数编码的模拟退火遗传算法应用在自动组卷问题中。为了对群体中每个个体进行调整并改善单一遗传算法的性能,该算法以遗传算法流程作为主体流程,在主流程中嵌入模拟退火算法。与现有遗传算法相比,该算法能较好地克服未成熟收敛现象,并且组卷的成功率和速度有明显的提高。

Autogenerating a test paper from a test database, which satisfies multiple requirements is a combinatorial optimization problem. The characteristics of some existing algorithms ofautogenerating a test paper is investigated and using a real code based simulating annealing genetic algorithm to address this problem is proposed. In order to adjust each individual of the group and improve the performance of a single genetic algorithm, the proposed algorithm combines the genetic algorithm with the simulating annealing method. Compared with the existing genetic algorithms, the proposed algorithm overcomes the premature convergence phenomenon more easily and it improves the success rate and the efficiency of autogenerating a test paper as well.