Data-Driven Multi-Objective Vehicle Routing Algorithm With Soft Time Window And Application Simulation

Recently,China’s transportation industry has developed rapidly,and the vehicle routing problem is very crucial as the core link.Reasonable planning of vehicle paths can improve the refined transportation management level and efficiency of enterprises,which has become the development trend of related industries.Moreover,the vehicle routing problem with soft time window is more consistent with the demand of actual production.However,the influence of uncertain factors such as road conditions and weather on vehicle transportation is rarely taken into account in existing research,resulting in the lack of practical application value of the schemes obtained.Hence,the data-driven evolutionary algorithm is introduced in this paper based on the historical trajectory data,which adds new power to the related research of soft time window vehicle routing problem and provides more practical and feasible routing scheme for enterprises.The research is mainly carried out in the following parts:(1)Basic theory of the vehicle routing problem with soft time window.Firstly,the development of soft time window vehicle routing problem at home and abroad is analyzed from two aspects,namely problem model and solving algorithm.Then,the traditional optimization model and solving method are systematically introduced.Moreover,the generic definitions and approaches of multi-objective optimization and data-driven evolutionary optimization are expressed,which will provide a theoretical foundation for solving the data-driven multi-objective vehicle path problems.(2)Multi-depot multi-objective vehicle routing model with soft time window.Considering the actual situation of vehicle transportation,the time window constraint in the classical optimization model is further strengthened.A multi-depot multiobjective path optimization model with soft-time window was constructed to achieve a balance between transportation cost and fleet efficiency,which has dual objective functions of minimizing the penalty for violating the time window and the total distance traveled.(3)Data-driven multi-objective vehicle routing algorithm with soft time window.The impact of uncertain factors such as road condition and weather is introduced through data-driven optimization method.A dynamic random forest surrogate was constructed using trajectory data,which was used to evaluate the new schemes and guide the search of the algorithm.Meanwhile,an adaptive pheromone volatile factor and a multi-objective fitness mechanism are proposed to improve the quality of the solutions.(4)Application of multi-objective vehicle path optimization in an open-pit mine.The proposed model and algorithm are simulated in an open-pit mine in Central China to obtain the reasonable transportation plans between multiple loading and unloading points.Simulation analysis shows that the proposed model and algorithm have good applicability in the open-pit mine,which can save transportation costs while enhancing work efficiency,thereby improving corporate benefits.