基于多优先规则启发式的分布式多项目随机调度  

A stochastic scheduling for distributed multi-project with multi-PR heuristic

作　　者：刘东宁 徐哲[1] LIU Dongning;XU Zhe(School of Economics and Management,Beihang University,Beijing 100191,China)

机构地区：[1]北京航空航天大学经济管理学院,北京100191

出　　处：《系统工程理论与实践》2021年第12期3294-3303,共10页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(71571005)。

摘　　要：在活动工期不确定的分布式多项目调度问题中,以最小化期望总拖期成本为目标,设计基于多优先规则启发式的动态调度算法,在每个决策点分别以不同的多优先规则启发式决策候选活动和协调候选活动间的全局资源冲突.此外,考虑到管理者信息披露的策略性行为,设计了支付协商过程,以对抗两个项目问题中管理者的投机行为.开展了详细的计算实验研究,结果表明:在五种活动工期分布下,本文算法可以有效降低总拖期成本的期望值;在工期常分布下,本文算法更新了相关研究中的最优结果.另外,实验还分析了两个项目问题中的管理者在协商过程中的策略性行为对多项目调度的影响,从成本的角度分析了管理者的占优行为策略及对应的多项目调度求解质量.In the distributed multi-project scheduling problem with uncertain duration,a dynamic scheduling algorithm based on multi-priority-rule heuristic is proposed to minimize the expected total tardiness cost.At each decision point,different multi-priority-rule heuristics are used to determine candidate activities and coordinate global resource conflicts among candidate activities,respectively.Moreover,considering the strategic behavior of information disclosure of managers,a payment negotiation process is designed to counter the opportunistic behavior of managers in the two-project problems.Based on a comprehensive computational study,the results show that the proposed algorithm can effectively reduce the expected total tardiness cost under five kinds of activity duration distribution,and update the optimal results in relevant studies under the constant duration distribution.Additionally,the experiment also analyzes the impact of the strategic behavior of managers in the negotiation process on the multi-project scheduling.From the perspective of cost,the dominant strategy of managers and the corresponding multi-project scheduling solution quality are also studied.

关 键 词：分布式多项目调度 优先规则 启发式算法 不确定工期 行为分析 

分 类 号：N945[自然科学总论—系统科学]