基于渐进式的遗传算法在智能组卷中的应用

Application of a Progressive genetic algorithm in Generating Test Paper Intelligently

智能组卷是一个多约束目标的组合优化问题。文中首先分析了试卷的评价指标、各项指标的作用及几个重要指标间的关系。在此基础上采用各种指标的加权和构建了组卷的总体评估函数。然后针对传统遗传算法的后期搜索效率低并极易出现未成熟收敛的情况,提出了一种渐进式遗传算法。仿真实验表明,渐进式的遗传算法在全局搜索性能及收敛速度上较传统遗传算法有显著提高,系统测试结果表明渐进式遗传算法组卷速度快,组卷质量较好。

Automatic Test Paper is a combination of multi-objective constrained optimization problem. First, we analysis the evaluation indicators in automatic test paper system, the role of indicators and the relationship of them. On this basis we use a weighted indicators and building a test paper to assess the overall function. Traditional genetic algorithm is inefficient and vulnerable to premature convergence under some conditions, for this we bring forward a progressive genetic algorithm. Simulation results show that the performance and convergence speed of progressive genetic algorithm are better than the conventional genetic algorithm. The system testing results show that the progressive genetic algorithm has a better quality and speed.