Research On Restaurant-to-customer Delivery Routing Problem Based On A Multi-objective Particle Swarm Optimization Algorithm

With the rapid development of e-commerce technology,people can enjoy customized distribution services through fast logistics transportation networks,which greatly facilitates their lives and works.Among the previous researches of the delivery optimization problem,most application scenarios mainly adopted the shortest path or the minimum investment cost as the single objective to optimize.This decision led to the one-sided pursuit of e-commerce that only served the interests of themselves instead of the customers.Meanwhile,the customer feedback information in the process of complex delivery should be equally important as the cost of the merchants,so these idealistic and one-sided thinking theoretical models are not suitable for the study of delivery routing problems which are related to complex reality.In actual operations,the cost invested by the merchants should not be regarded as a figure of revenue.In fact,the comprehensive revenue that the merchants need to optimize includes not only the short-term revenue in the current delivery service,but also the quality of service which is related to user comments.Positive user comment will create a good brand for the merchants,and become an important indicator that will affect potential revenue in the future.The motivation of this thesis is based on comprehensive consideration of the delivery service scenario which considers with both the delivery side and the customer side.The structure of this model is based on the research of the traditional vehicle routing problems,and aims at overcoming the defects of the traditional models.Moreover,a modification of the optimization objectives and a mechanism of service scoring for both sides are proposed.These contributions of this research lead to innovative design.The main work of this thesis is shown as follows:(1)An formulation model based on restaurant-to-customer delivery routing problem is proposed.Compared with the traditional delivery routing problem and its variants,this model redesigns a set of satisfaction scores mechanisms based on user comments,and regards them as the same as the cost in traditional delivery routing problems.At the same time,in addition to receive the user comments in the form of satisfaction scores,an additional set of parallel scoring mechanisms is used to evaluate the experience of the side of delivery workers in each delivery service.Therefore,the quality of the delivery service can be re-evaluated from the perspective of a non-user.This improvement makes this model comprehensively consider the direction of optimization process.(2)A particle swarm optimization algorithm with high result accuracy is proposed.The optimization algorithm mainly uses a set-based discrete representation schem,so that the particle swarm optimization algorithm can be enabled in a discrete problem space,and the redefined updating rules can quicken the search efficiency for the optimal solution.After the end of each iteration of optimization,the supplementary neighborhood generation scheme can further refine the current feasible solution to improve the lack of diversity in the optimization process.This algorithm combines the strategy which is named as”multiple populations with multiple objectives”,and adopts an archive mechanism that stores the optimal solutions from previous generations,ranking the high-quality feasible solutions to ensure that the algorithm works well.