多Agent系统协作求解的粒子模型方法

摘要

The relation between cooperative problem solving of distribution in MAS and partical collaboration are discussed, and a partical model for cooperative problem solving is proposed in MAS, which transforms the process of cooperative problem solving into co-optimization of particles. A parameter of collaboration extent is introduced, formula of demand intensity and effectiveness of target function of benefits are established, and particle swarm optimization algorithm to solve such problem is developed. Through evolutionary computation, an optimal solution for task allocations and resource assignments can be found. The proposed approach can describe and process the self-organization phenomena of Agent as well as the randomness and simultaneity of social interaction behaviors to complicated problem solving. The simulation experiments demonstrate the effectiveness and convergence of the method.%讨论了多Agent系统分布协作求解和粒子协作之间的关系,提出了一种多Agent系统协作求解粒子模型方法,将任务资源规划协作求解过程转化为多粒子共同寻优的过程.引入了协作程度变化参数,建立了需求强度计算公式和效益目标函数,并构造了适合求解的粒子群算法.通过算法的寻优计算,得到了任务资源规划协作求解的最优解.仿真实验结果表明,对于复杂的任务资源规划问题,该方法能描述和处理Agent本身自组织现象和社会交互行为的随机性和并发性,并具有良好的收敛性和有效性.