遗传模拟退火智能组卷策略研究

摘要

智能组卷是一个包含多重约束条件的最优组合问题,设计有效算法求解优化问题对获得高满意度试卷至关重要,为此,提出了一种基于模拟退火与标准遗传算法的混合智能组卷算法.约束模型采用了基于期望平均分的难度分布函数计算试卷难度系数,该算法模型采用分组实数编码,单点交叉算子,单个和随机两种基因变异算子,并对每代个体进行模拟退火操作.给出了详细的实验对比结果,表明了该算法的有效性.%Auto-generating test paper which satisfies multiple requirements is a combinatorial optimization problem. An effective algorithm is quite important for obtaining a satisfying test paper. So a hybrid algorithm based on standard genetic and simulated annealing algorithm is proposed. Test paper' s difficult coefficient is computed through the difficulty distribution function based on the expectation of average score. Group real code is adopted, one point crossover operator, and single point and random point mutation operator are used as genetic operators in the algorithm. After genetic operation simulated annealing operation is applied to each individuals for every generation. Finally, the detail algorithm' s steps and test results are given, and the algorithm is proved effective, feasible and practical.