针对合同网协议协商机制缺乏问题求解质量与效率分析的情况,设定假设条件,构建马尔可夫链模型,得出了利用目前已有的合同类型无法保证全局收敛的结论.在此基础上,提出了变邻域合同系的概念,通过分析控制方式对收敛性的影响,得出了集中式控制可以保证全局收敛以及分布式控制以概率保证全局收敛的结论,并设计了概率的计算方法.采用Doebin理论,对应用变邻域合同系的收敛速率进行分析,得出了集中式控制收敛速率与分布式控制收敛速率的上下界估计.
Considering the deficiency of contract net protocol negotiation mechanism convergence and convergent rate research, the Markov chain model of contract net protocol is created under the assumption setting, and it is proven that using previous contract types cannot be guaranteed to get global convergence. Using this as a foundation, the concept of variable neighborhood contract family is proposed. By analyzing the influence of controlling to convergence, the conclusion is obtained that centralized control can guarantee the global convergence and distributed control can guarantee the global convergence with probability, and the calculation method of probability is presented. By taking the advantage of Doebin theory, the convergent rate of variable neighborhood contract family application is analyzed, and the estimates of the convergent rate under centralized control and the upper and lower bounds of convergent rate under distributed control are presented.