A Two-Part Self-Adaptive Technique in Genetic Algorithms for Project Scheduling Problems

Abstract

The present paper introduces a novel two-part self-adaptive technique in designing the genetic algorithm for project scheduling problems. One part of the algorithm includes a self-adaptive mechanism for genetic operators like crossover and mutation. The second part contains another self-adaptive mechanism for genetic parameters such as crossover probability. The parts come in turn repeatedly within a loop feeding each other with the information regarding the performance of operators or parameters. The capability of the method is tested and confirmed in comparison to metaheuristic and exact algorithms based on well-known benchmarks.