中文摘要：

多Agent系统中,Agent形成联盟来完成任务,是Agents间的一种重要合作方式.遗传算法在求解单任务Agent联盟时存在稳定性较差、收敛速度慢、寻优能力不强等问题,对此,提出一种基于改进遗传算法的单任务联盟形成策略.该方法通过定义衡量遗传算法种群多样性参数,根据该参数值使用不同的配对策略在潜在交叉集合中选择个体进行配对交叉,以减少无效的交叉操作,从而提高交叉操作的效率;针对传统变异算子缺乏一定的方向性,通过个体Agent能力大小确定变异基因位,以提高算法搜索性能.对比实验结果表明,该算法可以快速、高效地找出合适的Agent联盟.

英文摘要：

In multi-agent systems,agents form a coalition to compete the task which is the important cooperation in agents.There are some problems such as slow convergence,low stability and poor optimization when the genetic algorithm is used to solve agent coalition for single task.The paper proposes a single task coalition formation tactic based on improved genetic algorithm.The method defines a parameter to measure the diversity of population about genetic algorithm.The individual is selected to crossover according to matching strategy based on this parameter in the potential crossover set.Ineffective crossover operations are decreased greatly and the efficiency of crossover operation is increased.As the traditional mutation operation lacks a certain direction,the paper presents to find gene mutation based on individual＇s ability to the agent,to improve the searching performance of the algorithm.The comparative experimental results show that the improved algorithm can find out the coalition quickly and efficiently.