Genetic Algorithm for Resource Leveling Problems under Various Objective Functions

In this paper some resource leveling problems for project scheduling under resource constraints are considered. In this problem activities of a project are to be scheduled in a way that all precedence and resource constraints are satisfied, and a given objective function describing the fluctuations of resource usage is minimized. Activities are nonpreemptable, and resources are renewable. Several various objective functions for the problem are analyzed. A genetic algorithm is proposed to solve this strongly NP-hard problem. The performance of the algorithm is examined and compared to other simple heuristics on a basis of a computational experiment performed on a set of standard benchmark instances.