Solving large unconstrained multilevel lot-sizing problems using a hybrid genetic algorithm

We develop a genetic algorithm (GA) to solve the uncapacitated multilevel lot-sizing problem in material requirements planning (MRP) systems. The major drawback of existing approaches is undoubtedly their inability to provide cost-efficient solutions in a reasonable computation time for realistic size problems involving general product structures. By contrast, the proposed GA can easily handle large product structures (more than 500 items) with numerous common parts, a problem type for which standard optimization software memory becomes rapidly insufficient. Based upon several hybrid operators and an original way to build up the initial population, the resultant GA provides in a moderate execution time high cost-effectiveness solutions compared with other techniques, in the extensive tests we performed.