关键词: 分布式多项目调度, 柔性调度, 协调机制, 启发式算法

Abstract: With the development of economy, the projects of enterprises are generally large, diversified and decentralized, and the decentralized multi-project management mode is attracting more and more attention. In the process of decentralized multi-project scheduling, each project leader has more autonomy, and often has different decision-making goals from the senior manager. For sub-projects, how to get more global resources and complete the project as soon as possible is his decision focus, while senior managers need to take a holistic view to minimize the impact of limited resources on multiple projects. Therefore, how to coordinate global resources allocation and plan multiple project scheduling in a decentralized multi-project environment has become a key concern.
However, most of the existing studies on decentralized multi-project scheduling are based on the assumption that the task’s duration is known and constant and the tasks are scheduled at a uniform rate, which is not entirely consistent with the realistic background and also causes waste of resources, limiting the flexibility and practicality of tasks scheduling. For some projects that do not have a strict limit to the duration of tasks, the duration of tasks can be taken flexibly within a certain range and according to the supply of resources, and the execution rate of tasks can be adjusted according to the supply of resources at different time periods, to achieve the purpose of making full use of resources. Therefore, based on the existing research, the paper proposes a flexible scheduling method for the decentralized multi-project, which provides a new approach to solving global resources conflicts.
First, a bi-level optimization model is proposed in the paper. The upper-level model is a global resources scheduling model with the decision objective of minimizing the delay cost of multi-projects, and the lower-level model is a local sub-project scheduling model with the decision objective of minimizing the completion time of sub-projects. In the local project scheduling process, the duration and start time of tasks are jointly used as decision variables, and a nested heuristic algorithm is used. In the global resources scheduling process, the paper proposes a new coordination mechanism. That is, in addition to the method of “delaying the start time of newly arrived tasks”, the resources conflict can be solved by cutting the executing task into two subtasks at the moment of conflict, and rescheduling the incomplete part without allowing interruptions by changing its execution rate.
The paper conducts experimental tests on some randomly generated cases of the multi-project. The results show that the task-cutting coordination mechanism is effective in reducing both the multi-project costs and delay days. That is because it increases the ways to resolve global resources conflicts, and the duration and unit resources requirements of tasks are no longer strictly limited. The performance of this flexible scheduling approach is also influenced by the intensity of global resources conflicts and project size. The larger the multi-project size or the stronger the global resources conflict, the more difficult the project scheduling will be, and the advantage of the flexible scheduling approach decreases but tends to be stable. When the global resources conflict is more intense, the marginal utility of one more way to mitigate conflicts will be smaller, and it is better to solve by increasing the supply of resources. The flexible scheduling method better meets the needs of decentralized multi-project management and also improves the scientificity, flexibility and practicality of multi-project scheduling decisions. At the same time, the multi-skilled global resources also contribute to mitigating resources conflicts, which can be further considered in future research.

Key words: decentralized multi-project scheduling, flexible scheduling, coordination mechanism, heuristic algorithm