考虑多技能人力资源的分布式多项目调度问题

在分布式多项目管理中，当考虑共享资源为多技能人力资源时，共享资源的指派涉及到"活动-技能-资源"的匹配关系，进一步考虑人力资源多技能异质的特性，还会导致活动实际执行工期会随着指派方案的不同而变化，因此，考虑共享资源为多技能人力资源的分布式多项目调度问题是一类具有很强现实背景和理论研究价值的NP-hard问题.基于多Agent系统，建立以单项目完工时间为优化目标的局部调度模型，以及多项目总延期成本为优化目标的全局协调决策模型；考虑人力资源多技能异质的特点，设计基于贪婪指派策略的序贯博弈谈判机制指派共享的人力资源；基于Ran Gen随机生成的多项目算例开展实验研究.研究表明：基于贪婪指派策略的序贯博弈谈判机制对于程度不同的资源冲突和不同规模问题的求解均具有较好的适应性，可有效降低多项目总延期成本，且求解效果优于随机分配策略；设计的优先指派技能水平高、掌握技能数少、工作时间短的全局资源指派策略，可获得较优的人力资源指派方案和多项目调度计划.

Distributed multi-project scheduling problem with multi-skilled staff

In the distributed multi-project management, assigning shared resources as multi-skilled staff involves the "activity-skill-resource" matching relationship. The heterogeneous characteristic of multi-skilled staff causes the actual duration of activities to change with different assignment schemes. Therefore, distributed multi-project scheduling problem sharing multi-skilled staff is a type of NP-hard problem with a strong realistic background and theoretical research value. Based on the multi-agent system, a local scheduling model with single-project completion time as the optimization goal is established. The global coordination decision model is established to optimize multi-project total tardiness costs. The sequential game-based negotiation mechanism with greedy assign strategy is designed to assign the sharing staff. The research results of multi-project studies randomly generated by Ran Gen show that the approaches we proposed can effectively reduce total tardiness costs of multi-project, and solutions are better than the random allocation strategy. Resources with high skill level, small number of skills and short working hours are assigned first, and better staff assignment schemes and multi-project scheduling plans can be obtained.