基于合作博弈协商机制的分布式资源受限多项目调度

分布式资源受限多项目调度问题涉及多个项目的独立调度以及项目之间共享资源的协调分配,因此,解决该问题的关键是设计有效的机制获取协调全局资源所需的调度信息并据此配置全局资源.考虑每个单项目延期的影响程度不同,以最小化多项目总拖期成本为全局目标建立分层决策模型:考虑分布式多项目调度问题的信息不对称性和各单项目决策者的自利性,设计基于合作博弈协商机制的分阶段进化算法对问题求解.以MPSPLIB问题集开展实验研究,结果表明:采用合作博弈协商机制可以有效降低多项目的总延期成本,并且该求解机制对问题规模和资源冲突强度的变化有很好的适应性;自主Agent参与合作博弈并选择诚信策略可获得更高的单项目收益,同时多项目获得高质量的调度方案.

Distributed resource constrained multi-project scheduling problem with cooperative-game based negotiation mechanism

The distributed resource constrained multi-project scheduling problem (DRCMPSP) involves the individual scheduling of multiple projects and the coordination of shared resources among the projects. It is the crux of solving the DRCMPSP that designing an effective mechanism to obtain the necessary projects' information in order to coordinate the global resources. We constructed the hierarchical model with the global objective of optimizing the total delay cost of the multi-projects under the consideration of different unit project. And we designed a cooperative-game based negotiation mechanism to allocate the global resources and the hierarchical model was solved by the proposed phased evolution algorithm under the account of the information asymmetry in DRCMPSP and the self-interested project decision makers. According to the computational results of instances in MPSPLIB, it is effectively to decrease the total delay cost of the multi-projects that using the cooperative-game based negotiation mechanism and the approach is available to various problems with different sizes and utilization factors. In addition, in order to get low delay cost the autonomous agents should participate in the cooperative game and behave honestly and meanwhile a high-quality solution will be obtained.