摘要
研究一种基于交叉变异操作的连续域蚁群算法,该算法对解的每一分量的可能取值组成一个动态的候选组,并记录候选组中的每一个可能取值的信息量。在蚁群算法的每一次迭代中,首先根据信息量选择解分量的初值,然后使用交叉、变异操作来确定全局最优解的值,通过相应算法设计,对于来自相对适应度较大的解的分量值,其变异的区域较小,成为局部搜索,反之,变异的区域较大,则构成全局搜索。同时,随着迭代次数的增多,分量值的变异幅度逐渐变小,这样可使收敛过程在迭代次数较多时得到适当的控制,以加速收敛。最后通过仿真实验,把交叉变异操作的连续域蚁群算法与遗传算法性能进行比较,证明了交叉变异操作的连续域蚁群算法具有较高的搜索较优解的能力,大大节约了计算时间。
This paper studies a continual-domain ant colony algorithm based on the overlapping mutation operation, which forms a dynamic candidate group to the solution of each component possible value, and records each possibility value information content in the candidate group. In each iteration of ant colony algorithm, firstly, it should choose the starting value of solution component according to the information content, and then it should use overlapping and variation operation to determine the overall optimal solution value. Through the corresponding algorithm design, from the relative sufficiency big solution component value, its variation region is small and becomes the partial search. Otherwise, the variation region is big and then constitutes the overall situation search. At the same time, along with iterative number of times increase, the component value variation scope becomes gradually small, as this may enable the restraining process a lot when the iterative number of times is under the suitable control in accelerating convergence. Finally through the simulation experiment, compared the overlapping variation operation continual domain ant colony algorithm with the genetic algorithm performance. Conclusion has proven the overlapping variation operation continual domain ant colony algorithm has the high search superior solution ability and saves the computing time greatly.
出处
《重庆师范大学学报（自然科学版）》
CAS
2009年第2期87-89,共3页
Journal of Chongqing Normal University:Natural Science
基金
川北医学院苗圃基金(08基-17)
关键词
交叉变异
连续域
蚁群算法
overlapping mutation
continual domain
ant colony algorithm
