复杂自适应多Agent系统的模型驱动开发方法  

Model-driven Development Methodology for Complex Self-adaptive Multi-agent Systems

作　　者：曹江[1] 毛新军[1] 王怀民[1] 卢锡城[1] 

机构地区：[1]国防科技大学计算机学院,长沙410073

出　　处：《计算机科学》2012年第2期126-131,共6页Computer Science

基　　金：国家自然科学基金(61070034);教育部博士点基金(20094307110007);国家973课题(2011CB302601)资助

摘　　要：自适应系统是一类复杂系统,如何有效地支持此类系统的工程化开发,一直是软件工程领域的关注焦点。提出了一个基于Agent的模型驱动软件开发方法,试图将主流软件工程中的MDA技术与软件Agent技术相结合,从而为高效、高质量地开发复杂自适应多Agent系统提供方法学指导。该方法将基于组织抽象和ODAM+方法学所建立起来的模型视为平台无关模型,将基于SADE平台的实现模型视为平台相关模型,通过建立这两个不同抽象层次元模型间的映射关系,来实现从平台无关模型到平台相关模型以及最终代码框架的转换。介绍了集成MDA和Agent技术的软件方法学ODAM+,阐述了复杂自适应系统模型驱动开发的一组关键技术,包括不同层次的元模型以及它们之间的映射关系、模型转换规则和模型转换算法,最后分析了相应的支撑软件工具和应用验证情况。Self-adaptive system is a complex system.How to develop such systems in an effective engineering way has become an important topic in the software engineering community.This paper presented an agent-based model-driven methodology that attempts to integrate MDA and agent technology to support the development of complex self-adaptive systems.Our approach takes organization-based models created by ODAM＋ as platform-independent models,and models based on SADE platform as platform-specific models.The transformation between these two kinds of models is performed by establishing the mapping relationships of meta-models in these two different levels.This paper introduced in details the software methodology ODAM＋ and the involving technologies such as meta-models and their mapping relationships between platform-independent models and platform-specific models,transformation rules and algorithms,lastly the software toolkits supporting the methodology was introduced and a case was studied.

关 键 词：自适应系统 AGENT 组织抽象 模型驱动架构 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]