多Agent系统中基于改进合同网模型的任务分配研究  

Study on the Task Allocation Based on Improved Contract Net in Multi-agent System

作　　者：裘杭萍[1] 覃垚[1] 胡汭[1] 管留[1] 

机构地区：[1]解放军理工大学指挥自动化学院,南京210007

出　　处：《计算机科学》2012年第B06期279-282,共4页Computer Science

基　　金：总装预研基金项目(9140A07020206JB8102)资助

摘　　要：任务分配是多Agent系统中研究的热点,合同网模型是关于多Agent系统中任务分配的经典策略,但传统的合同网模型存在很多不足。在引入基于信任度的招标策略和基于自适应的投标策略的基础上,主要针对传统合同网模型中标阶段存在的问题,从投标Agent的负载、能力和信任度3个方面进行综合考虑和权衡,提出了一种基于多属性评价中标策略的动态任务分配算法,从而有效地提高了任务分配和执行的效率。最后通过仿真实验验证了基于多属性评价中标策略的正确性和合理性。The research of task allocation in multi-agent system is becoming more and more.The contract net model is a classical method for task allocation in multi-agent system,but traditional contract net model has lots of deficiencies.With the base of bidding announcement based on the credit of agent and bidding based on self-adaptation,a synthetic evaluation strategy based on the load,capability and credit of bidding agent was also proposed,which can effectively resolve the problem during the procedure evaluating bidding.The simulation experiment proves that the performance of the extended contract net is effective.

关 键 词：多AGENT系统 任务分配 改进合同网模型 多属性评价 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]