A Model and Genetic Algorithm to Solve Complex PS (Project Scheduling) Problems

This paper solves a complex PS problem by employing a reproduction mechanism which is closer to natural than the common GA methods. The PS problem consists of: operations sequence constraints, alternative modes per operation, multi renewable resources, resource setups, and other properties. The resulting model is very complex; hence, common GA methods produce unfeasible solutions very frequently. By mimicking the multilevel Chromosome’s structure in natural reproduction, we reduce the rate of unfeasibility produced by the crossover. The GA levels we consider are: DNA, Chromosome, Gene, and a Nucleic. The main idea is that crossover is conducted only between chromosomes, and mutations within genes. The proposed GA method is much more efficient, especially for large problems, and reaches its objective close to optimal versus a customized B&B algorithm.