Parallelized Population-Based Multi-heuristic Approach for Solving RCPSP and MRCPSP Instances

Project scheduling with resource constraints are of significant importance in various application domains, including logistics, production, management, health care, and computer science. Two generic problems including the Resource-Constrained Project Scheduling Problem (RCPSP) and the Multi-Mode Resource-Constrained Project Scheduling Problem (MRCPSP) have attracted a lot of research effort. Since both are computationally difficult, finding satisfactory solutions for instances of even moderate size is not an easy task. In the literature there have been proposed numerous approaches based on using heuristic or metaheuristic algorithms. An effective way of addressing discussed problems include parallelizing computations and using multiple heuristic or metaheuristics. This paper introduces a set of heuristic algorithms and a parallelized population-based multi-heuristic system designed for the Apache Spark environment as an efficient method for solving instances of project scheduling problems. The approach has been validated in an extensive computational experiment based on datasets from PSPLIB library.