| Summary: | 碩士 === 元智大學 === 工業工程與管理學系 === 92 === Rapid development of new product has shortened the life cycle of product tremendously over the decades. Therefore, enterprises have learned to be much stricter on project scheduling in order to make use of all resources, and this situation leads to focusing on the resources-constrained project scheduling problem (RCPSP) which is an NP-Hard problem.
This research develops an ant colony optimization algorithm (ACO-RCPSP) to solve resources-constrained project scheduling problem. Parallel-forward scheduling method is used, and the maximum ACTim value first (ACT) rule is applied as the local heuristic. To investigate the effects of parameters in the ACO-RCPSP algorithm, a comprehensive design of experiment is employed for both single-resource type and mixed-resource type problems. Four different sizes of test problems from PSPLIB are used as the benchmarks.
ACO-RCPSP performs comparatively to other best-known methods considering the average percentage deviation from the lower bound, outperforms the best-known algorithm (AS-RCPSP) in the minimum percentage deviation from the lower bound measure, and improves the computational efficiency significantly. In addition, when comparing with the best-reported simple heuristic ACT in RCPSP literature, ACO-RCPSP easily outperforms ACT in all types of problems.
|
|---|
