一种基于三维编码修正的重叠联盟形成方案

An overlapping alliance formation scheme based on three-dimensional coding modification

重叠联盟形成是人工智能与多Agent系统领域中的一个关键问题。由于每个Agent拥有的资源有限,不同任务求解联盟之间会存在潜在的资源冲突。本文将传统的差分进化扩充至三维整数编码,更直观、有效地完成智能资源体的快速分配,并针对初始化赋予任意值可能产生的联盟无效与资源冲突问题,设计了相应的编码修正策略,以确保任何一个无效编码都能够被修正为一个合法编码。

The formation of overlapping alliances is a key problem in the field of artificial intelligence and multi-agent systems.As each Agent has limited resources,there are potential resource conflicts between different task solving alliances. In this paper,the traditional differential evolution is extended to the three-dimensional integer coding,so as to realize the fast allocation of intelligent resources more intuitively and effectively. What’s more,in order to solve the problem of alliance invalidity and resource conflict caused by initialization assigning arbitrary values,the corresponding encoding correction strategy is designed to ensure that any invalid encoding can be modified into a valid encoding.