基于改进自适应遗传算法的组卷研究

针对遗传算法容易出现早熟和收敛速度慢的问题，根据群体适应值分布的变化特点，启发性地提出了一种新的基于小生境技术的自适应遗传算法（ANGA）。根据群体中各个个体的适应值分布情况加以启发，引入了一个自适应的常数Cmin，通过白适应调整Cmin以适时改变群体适应值的分布，优化了各个个体被选择的概率。详细介绍了ANGA应用于组卷问题的步骤。涵盖了其中的各项关键技术：组卷策略、编码方案、适应值函数的确定、选择交叉变异箅子的实现。并以目前的计算机等级考试三级信息管理技术的组卷为例，采用ANGA算法进行了仿真计算。仿真结果表明，ANGA算法能够成功地应用于自动组卷。算法能够以100％的概率在较短的时问内完成组卷，组卷效率高、成功率高；且算法对初值不敏感．具有较好的鲁棒性。

Investigation of Test Paper Auto-generation Based on Improved Adaptive Genetic Algorithm

To deal with the prematurity and the low convergence speed of genetic algorithm, a new adaptive genetic algorithm based on niches (ANGA) was developed according to the variety of population fitness distribution. Inspired by the variety of population fitness distribution, a self-adaptive constant Cmin was introduced. By adjusting Cmin according to the population fitness distribution, the selection probability of each population was optimized. The procedure to solve the question of test paper auto-generation with ANGA is presented in detail. The test paper auto-generation of the present "National computer rank examination(3):information management technical" was taken as an example. The tests results indicate that: ANGA can be successfully applied in test paper auto-generation system; the algorithm can finish a calculation so as to 100% of the probability within a short time, and the speed is quite fast, furthermore, the success rate is high, the sensitivity to initial value is dull.