摘要： 在基于网络的分布式系统应用基础上，分析了大型复杂问题复合模型协作求解的过程特征描述，提出基于多Agent 的领域问题协作求解的主动控制策略，探讨了用户交互Agent、系统主控Agent、协作Agent以及模型Agent和数据Agent等复合模型协作求解的4种Agent类型。应用多Agent层次结构，提出一种复合模型协作求解的自适应QoS体系结构，通过实现复合模型协作求解的主动调度规划算法对其进行了验证，支持分布式网络环境下实现模型资源和数据资源的共享，以提高协同计算环境分布式问题协作求解的运行效率和服务质量。

关键词: 多Agent, 复合模型, 主动控制, QoS, 协作

Abstract: On the basis of network-based distributed application system, this paper discusses the process description of composite model collaborative solving for large scale complexity problem. An multi-agent based active control strategy is put forward to support distributed domain problem solving, including four kinds of agent named user agent(UA), system agent(SA), collaborative agent(CA), model agent and data agent(MA&DA) separately. By using the multi-agent architecture, an adaptive QoS adjusting mechanism for composite model collaborative solving is presented to enhance the capability of network resource sharing both in the scope of model-base system and database system. An algorithm to schedule and plan the composite model sequence for distributed problem solving is implemented so as to improve the running efficiency and quality of service under distributed computing environment.

Key words: multi-agent, composite model, active control, QoS, collaboration

中图分类号: 

• TP391