Solution Evaluation-Oriented Multi-objective Differential Evolution Algorithm for MOVRPTW

Multi-objective vehicle routing problem with time windows (MOVRPTW) is a canonical logistics problem widely existing in supply chain. It is challenging to obtain the feasible solutions with fast convergence and well diversity due to the constraint of time windows. To address this issue, a solution evaluation-oriented multi-objective differential evolution (SE-MODE) algorithm is presented in this paper. First, a solution evaluation mechanism based on constraint dominance principle is developed to evaluate the dominance degree of feasible solutions and infeasible solutions quantitatively. Second, infeasible solutions with less dominance degree are utilized to generate solutions in the early stage of evolution adopting a memetic algorithm framework. Third, a feasible solution-oriented differential mutation strategy is developed to increase the probability of generating feasible solutions and improve the convergence of the population. Finally, the proposed SE-MODE algorithm is evaluated on the RC instances from Solomon, experimental results show that SE-MODE algorithm is promising in solving MOVRPTW.