Estimation of distribution algorithms for the Multi-Mode Resource Constrained Project scheduling problem

Abstract

Multi-Mode Resource Constrained Project Problem (MRCPSP) is a multi-component problem which combines two interacting sub-problems; activity scheduling and mode assignment. Multi-component problems have been of research interest to the evolutionary computation community as they are more complex to solve. Estimation of Distribution Algorithms (EDAs) generate solutions by sampling a probabilistic model that captures key features of good solutions. Often they can significantly improve search efficiency and solution quality. Previous research has shown that the mode assignment subproblem can be more effectively solved with an EDA. Also, a competitive Random Key based EDA (RK-EDA) for permutation problems has recently been proposed. In this paper, activity and mode solutions are respectively generated using the RK-EDA and an integer based EDA. This approach is competitive with leading approaches of solving the MRCPSP.