An adaptive large neighbourhood search as a matheuristic for the multi-mode resource-constrained project scheduling problem

摘要

>The multi-mode resource-constrained project scheduling problem is a well-known problem in the field of scheduling. The objective of the problem is to find the minimum makespan for the project. Hence, each activity has to be assigned to a mode and a starting time. At the same time, precedence and resource constraints must not be violated. We present a hybrid approach which combines an adaptive large neighbourhood search with mixed integer programming. We applied the procedure to datasets from the MMLIB library with up to 100 activities and nine modes. The computational results show that the approach is competitive with other state-of-the-art heuristics. Moreover, it found 294 new best known solutions and outperformed all other published methods on the MMLIB+ dataset. [Received 30 September 2016; Revised 6 April 2017; Accepted 26 April 2017]