Authors:
Alireza Etminaniesfahani
1
;
Hanyu Gu
1
;
Leila Naeni
2
and
Amir Salehipour
3
Affiliations:
1
School of Mathematical and Physical Sciences, University of Technology Sydney, Sydney, Australia
;
2
School of the Built Environment, University of Technology Sydney, Sydney, Australia
;
3
The University of Sydney Business School, The University of Sydney, Sydney, Australia
Keyword(s):
Approximate Dynamic Programming, RCPSP, Priority Rule, Uncertainty.
Abstract:
The resource-constrained project scheduling problems (RCPSP) with uncertainties have been widely studied. The existing approaches focus on open-loop task scheduling, and only a few research studies develop a dynamic and adaptive closed-loop policy as it is regarded as computationally time-consuming. In this paper, an approximate dynamic programming (ADP) approach is developed to solve the RCPSPs with stochastic task duration (SRCPSP). The solution from a deterministic average project is utilised to reduce the computational burden associated with the roll-out policy, and a parameter is introduced in the roll-out policy to control the search strength. We test the proposed approach on 960 benchmark instances from the well-known library PSPLIB with 30 and 60 tasks and compare the results with the state-of-the-art algorithms for solving the SRCPSPs. The results show that our average-project-based ADP (A-ADP) approach provides competitive solutions in a short computational time. The invest
igation of the characteristics of the instances also discloses that when resources are tight, it is more important to intensify the search in the roll-out policy.
(More)