2010 Volume 61 Issue 4 Pages 223-233
The vehicle routing problem has been focused on as a significant problem in recent production environments. This problem involves decision-making in regard to the number of vehicles and combination of locations for each vehicle. When multiple measures are used to evaluate results and some measures have a trade-off relationship, the problem should be treated as a multi-objective problem. Furthermore, incorporating different search procedures is useful for optimizing different measures and obtaining better Pareto solutions. In this study, a hybrid multi-objective genetic algorithm is proposed to generate Pareto solutions in a short calculation time for the bi-objective vehicle routing problem with a time window. Total distance traveled and average completion time are adopted as two objective functions in the problem. In the proposed method, different search methods are introduced to optimize different objective functions and the methods are incorporated into a multi-objective genetic algorithm framework. The proposed method is evaluated through numeric experiment and bench-mark problems. The experimental results show that introducing a combination of different search methods is effective in improving values of the different objective functions in Pareto solutions. Furthermore, the proposed method generates high-quality Pareto solutions in short calculation times.
