Abstract: The existing studies on distributed resource constrained multi-project scheduling problem have assumed that the global resource limit cannot be broken during the multi-project duration and taken the makespan as the optimization objective. Considering global resources that can be obtained from the outside, this study constructs an integer programming model of distributed flexible resource-constrained multi-project scheduling problem with the objective of max-NPV and designs an effective algorithm to solve the problem. Firstly, the problem is defined and project cash flow calculation is determined. Then, due to NP-hard of the problem, a genetic-simulated annealing hybrid algorithm (GA_SA) is designed to solve the model. Finally, other algorithms used to compare with GA_SA algorithm are designed through multiple numerical experiments, and the impact of key parameters on the net present value of multi-projects is analyzed. The results demonstrate that the GA_SA algorithm has good solution effects, and global resource with flexible usage status can improve the performance of distributed multi-project profit significantly.

Key words: distributed multi-project scheduling, resource flexibility constraints, net present value, genetic-simulated annealing hybrid algorithm

摘要： 现有的分布式资源约束多项目调度问题研究中,假定全局资源限量在多项目工期内不可突破且多以工期为优化目标。针对此问题,考虑全局资源可从外部获取,以净现值为目标,构建带有全局资源柔性约束的分布式多项目调度问题的整数规划模型并设计有效的求解算法。首先,界定问题并确定项目现金流的计算方法;然后,针对求解问题的NP-hard属性,设计了遗传-模拟退火混合算法(GA_SA)求解此模型。最后,通过多组数值实验,设计不同算法与GA_SA算法进行比较,并分析了关键参数对多项目净现值的影响。结果表明,GA_SA算法具有较好的求解效果;与传统的全局资源刚性约束条件相比,全局资源柔性使用状态可以显著改善分布式多项目的收益绩效。

关键词: 分布式多项目调度, 资源柔性约束, 净现值, 遗传-模拟退火混合算法