摘要
提出一种改进的遗传算法作为组卷的策略。染色体采用符号编码设计,解决了遗传运算过程中满足约束条件的问题。采用"非优超排序法"对染色体进行评价,在选择算子的设计上,既能够复制一部分较好的个体,又体现了选择的概率性。变异概率和交叉概率能随个体的不同适应度自适应改变,同时变异概率随种群多样性自适应变化。采用基于数据仓库的最优解保存策略,使搜索结果呈现出丰富的Pareto解集。
Proposes a new strategy of composing test paper based on the improved genetic algorithm. The symbol coding of chromosome makes sure the process of genetic operation goes under the constraints. The individuals are evaluated by nondominated sorting method. Both certainty of good individuals reproducing and probability are concerned in designing the selection operator. Probability of mutation and crossover probability would vary by the different individual's fitness automatically, and the mutation probability would change by the diversity of the population adaptively. Adopts a preservation strategy of Pareto optimum solutions based on a date warehouse, by which can obtain a collection of extensive Pareto optimum solutions.
出处
《现代计算机》
2007年第8期8-10,共3页
Modern Computer
关键词
组卷
多目标优化
遗传算法
随机联赛
符号编码
Composing Test Paper
Multiobjective Optimization
Genetic Algorithm
Stochastic Tournament Model
Symbol Coding