Combining metaheuristics and ILP for solving the car sequencing problem

Abstract

In this paper we propose two forms of hybridization between a metaheuristic (Ant Colony Optimization or Genetic Algorithm) and an exact method (ILP) for the solution of the car sequencing problem. We examine whether such hybridizations can improve solution quality over what can be obtained from the alternative approach of using local search within one of these metaheuristics. We present a computational experiment based on the CSPLib benchmarks. The results presented show that the use of hybrid approaches provides a significant improvement and that the performance of the heuristic or metaheuristic that the hybridization is based on greatly influences the quality of the results, however, computations time are long. We conclude that diversification in a population-based metaheuristic plays an important role in the hybridization.