An integrated approach for lot sizing and scheduling problems using meta-heuristics: Genetic algorithms (GA) and simulated annealing (SA)

This work presents the application of genetic algorithms (GA) and simulated annealing (SA) to solve an important and difficult problem in production management: the integration of lot-sizing and scheduling problems for a capacitated parallel machine production system. The problem is defined as obtaining the order quantities and the production schedule for a multi-item, time-varying demand, and sequence dependent setup time environment in order to minimize the total setups and inventory holding costs. A novel representation scheme is proposed to manage both problems simultaneously. Experimentation is performed to evaluate how different aspects of the genetic algorithm affect the results. In addition, solutions are compared with a well-known heuristic so called integrated approach method (IA). Results show that the proposed meta-heuristics provide better solutions than sequential and integrated approaches. Simulated annealing shows better efficiency and effectiveness than genetic algorithms when the initial solution is generated by the integrated method.