一种基于BDI Agent的复杂系统设计建模方法

Research on Design and Modeling Technique for Complicated Systems Based BDI Agents

提出一种通过信念愿望意图(BDI)结构实现多Agent系统设计建模的方法.多Agent系统设计建模的目的是具体地模型化多Agent组织中承担不同组织职能的每一类Agent的结构,使其易于在现有的程序设计语言环境中实现.在该方法中,多Agent系统的设计建模需要建立三个模型:Agent模型、相互作用模型和相识者模型.Agent模型由信念、目标、计划三个基本子模型构成,这些子模型是根据分析阶段所获得的职能模型、协同工作过程模型以及领域本体来建立的.MAS系统中的每一Agent都是Agent模型中某一类Agent的一个实例.相互作用模型主要说明Agent之间的交互细节,如交互协议、交互语言、交互约束条件等.相识者模型说明每一类Agent的所有相识者及其属性.目前通过这一建模方法已在多智能体协同工作平台(MBOS)上开发出实际的应用系统“多智能体物资调配决策支持系统MAEDSS”.

This paper presents an approach to model Multi-Agent Systems (MASs) for design by Belief-Desire-Intention (BDI) Structure. The objective of modeling for design is to model concretely the structure of every type of agent, which takes on the roles of a Multi-Agent Organization, so as to realize the MASs in existing programming environment easily. In this modeling method, each type agent needs three models: Agent Model, Interaction Model and Acquaintance Model. Agent Model consists of three sub-models: Belief Model, Goal Model and Planning Model, which is built according to the role models and cooperative working process models obtained from agent-oriented analysis phase. Each agent in MASs is an instance of some type of agent in Agent Model. Interaction Model specifies the detail of interaction between agents, such as interactive protocols, interactive language, interactive constrains, and so on. Acquaintance Model specifies all acquaintance and their properties for each type of agent. Now through this modeling approach, an effective application system, called 'Multiply Agent Equipment preparing Decision Support System (MAEDSS) ', has been developed on a cooperative working platform MBOS.