复杂时空网络冲突消解群组角色指派研究

Conflict Resoluting in Complex Spatiotemporal Networks via Group Role Assignment

为了解决因孤立时空约束而导致的多项任务指派的协同失效和全局优化性能急剧下降问题,使用角色协同理论(role-based collaboration)及其通用模型E-CARGO的子模型群组角色指派(group role assignment),以机场登机口调度为例,对问题进行指派时空约束形式化建模;分析不同代理承担不同角色、不同代理承担同个角色的协作情况,从而建立量化评估矩阵与协作矩阵;继而对时空约束进行解耦与消解,采用整数规划在追求协作空间利用率最大化的同时,考虑平衡旅客偏好,对问题进行多目标求解。大规模仿真实验论证了模型与方法的一般性、有效性和可靠性。此外,与传统GRA模型相比,主体利益指标提升6.21%,客体偏好指标提升9.72%,实现秒级求解,满足了复杂时空网络下的任务分配快速指派响应要求。

Tasks are often executed in a highly parallel and concurrent mode in spatiotemporal networks.Specified task distribution benefits the decreases of the complexity of cooperation among system components.As one of the important and pivotal issues in collaborating,spatiotemporal constraints must be taken into consideration in task allocation.Otherwise,the assigned tasks will be in conflict frequently during the execution step,which will lead to a sharp decline of the overall system performance.In order to avoid conflicts and optimize the team performance and the interoperability,the following three aspects are mainly investigated:1)taking airport parking space scheduling as an example,modeling the assignment problem under time and space constraints via group role assignment(GRA),which is a sub-model of Role-Based Collaboration(RBC)and its general model E-CARGO;2)analyzing situation of different agents undertaking different roles,and different agents undertaking the same role,so as to establish the qualification matrix and the collaboration matrix;3)decoupling and dissolving spatiotemporal constraints,and furthermore,pursuing an integer programming,which is used to solve the multi-objective balance between the passengers’preference and the utilization of airport parking space.Large-scale simulation experiments and results indicate that,this method is general,valid and reliable.In addition,compared with the traditional GRA model,the subject performance is increased by 6.21%,the object preference is increased by 9.72%,and the second level running speed can meet the requirements of rapid assignment response in complex spatiotemporal networks.