A Two-Layer Approach for the Decentralized Multi-Project Scheduling Problem Sharing Multi-Skilled Staff

With the rapid development of economic globalization and information technology, the projects undertaken by enterprises are gradually becoming larger and more complex. The multi-project management model has become the norm in enterprise management, and it shows a trend of decentralization in terms of geographical distribution and management environment. This decentralization mainly manifests in the fact that each project has its own private information and benefit objectives and shares limited global resources with other projects during the execution process, forming a decentralized decision-making environment. As an extension of the resource-constrained project scheduling problem, the decentralized resource-constrained multi-project scheduling problem integrates single-project scheduling and global resource coordination allocation in a decentralized decision-making environment with multiple independent decision-makers. Furthermore, when considering global resources as multi-skilled staff, the decentralized multi-project scheduling problem sharing multi-skilled staff studied in this paper is proposed. A two-layer model containing local scheduling and global coordination decision-making is established to describe this problem. A two-layer approach (TLA) is proposed to solve this problem. In the local scheduling layer, a bat algorithm based on forward-backward scheduling (BAFBS) is developed to generate local baseline schedules to minimize the single-project completion time. In the global coordination decision-making layer, a variable neighborhood tabu search algorithm with greedy assignment strategy is designed to resolve global resource conflicts to minimize the multi-project total tardiness cost. Computational experiments are conducted based on the Multi-Project Scheduling Problem LIBrary dataset. The results show that the BAFBS can obtain high-quality local baseline schedules. Compared to the existing decentralized and centralized methods, our proposed TLA can get better solutions on most problem subsets, which proves that our approach can effectively coordinate the allocation of multi-skilled staff among multiple projects.