摘要
合同网是用于分布式问题求解环境下各Agent进行通信和协作的一种协议,各个Agent内部的知识、能力和策略在问题求解过程中总是动态变化的。为了能够得到实用性强且易于实现的协作模型,分析了传统经典合同网的缺点并进行了改进,利用熟人模型对投标者进行了合理的筛选。采用Agent的主动感知特性,改进了任务Agent接受任务的被动性。在投标过程中引入信任度、感知系数和活跃度等智能参数,防止了投标者在投标过程中的自私性,平衡了各个合同承担者的负载。通过实验证明,改进的合同网降低了网络通信量,提高了多Agent系统协作的效率。
Contract net is a communication and collaboration protocol for each of Agent in the distributed problem solving environment. Each Agent with the knowledge, abilities and strategies is always dynamic in the process of solving problems. In order to obtain a collabo- rative model that is practical and easy to implement, this paper analyzes the shortcomings of the traditional classical contract net and improves them . We had a reasonable selection to the bidders using an acquaintance model . Agent model＇s proactive perception characteristic improved the passive acceptance of the task Agent. In order to prevent the selfishness in the bidding process and balance the load of each contract holder, introduced the intelligent parameters such as reliance, perception and activity. Proved through expe.rience, the improved contract net reduced the network traffic and improved collaboration of the multi -- Agent system efficiency.
出处
《计算机技术与发展》
2010年第3期71-75,共5页
Computer Technology and Development
关键词
多AGENT
合同网
熟人模型
感知
协作
multi - agent
contract net
acquaintance model
perception
cooperation
