Optimization Of Urban Distribution Path Based On Stochastic Travel Time

The development of society promotes the progress of the logistics industry,which manifests itself in the rapid development of the distribution industry in cities.Logistics distribution,as a source of power in the urban ecosystem,is a bridge for material exchange in various areas of the city,and has important strategic significance for the normal operation of people's daily lives.However,with the integration of the Internet and e-commerce into people's lives,consumption upgrades and accelerated product updates have made the original urban distribution model unable to adapt to more intensive logistics distribution needs,thereby restricting the improvement of urban logistics service efficiency and customer satisfaction.Therefore,considering the realistic factors to design a reasonable distribution scheme to improve the efficiency of the urban distribution network to meet the needs of customers has become a problem that scholars and enterprises pay close attention to and urgently need to solve.Emerging technologies such as big data and cloud computing provide new technological directions for optimizing existing urban logistics distribution systems.The traditional stochastic travel time vehicle routing problem(STVRP)stipulates that the travel time of a vehicle is a random travel time that obeys a certain probability distribution(usually normally distributed).Such a setting does not consider the impact of traffic congestion.Other studies have shown that it takes a certain minimum time for a vehicle to pass a certain distance,and then the speed probability increases rapidly to the maximum and then decreases rapidly,which is an obvious log-normal distribution law.Therefore,in this study,the travel time of a vehicle is set as a random travel time that obeys the log-normal distribution in consideration of traffic congestion,that is,the vehicle obeys the log-normal distribution of different expected variances under different traffic congestion situations.Set the situation closer to reality.This study discusses the following issues:(1)Consider the impact of weather,road conditions,accidents,and traffic congestion on vehicle travel time,and establish a time-varying random travel time vehicle routing problem model,where the random travel time is a log-normally distributed travel time considering traffic congestion.At the same time,a reliability evaluation model is introduced to evaluate the reliability of the vehicle journey,and the accuracy of vehicle arrival time can be improved by selecting a travel time with high reliability.A hybrid simulated annealing algorithm with genetic operators is designed to solve the model.Finally,combined with the case of Chongqing A supermarket,a detailed distribution scheme and route reliability are given,and the results prove the effectiveness of the model and algorithm.(2)Based on the time-varying random travel time vehicle routing problem,considering the characteristics of uncertain dynamic customer demand,a proactive forecasting method was designed to evaluate the dynamic customer's probable demand,and determine whether to provide dynamic customers with delivery services in advance.Reduce response time to demand.In order to meet the real-time needs of dynamic customers as much as possible,after the distribution route planning is completed,a strategy of vehicle waiting and customer buffering is designed,and the new dynamic customer needs are inserted into the routes that are being or will be served.This strategy can reduce the distribution cost of vehicles and increase the response rate to customer needs and improve customer satisfaction.(3)In this study,the Chongqing A supermarket's urban distribution network is taken as an example.Aiming at the problem of time-varying random travel time vehicle routing,a hybrid simulated annealing algorithm is used to solve the case and calculate the vehicle travel time reliability under different speed models.For the dynamic vehicle routing problem considering random travel time,the historical customer data of A supermarket is used to predict dynamic customers who may make requests,and to provide services in advance.For dynamic customers that fail to predict and newly added dynamic customers,wait and buffer strategies are used.Update vehicle distribution routes in real time to serve as many dynamic customers as possible.Experimental results prove that proactive prediction and waiting buffer strategies for dynamic customers can significantly reduce vehicle distribution costs and the number of dynamic customers who refuse service.Finally,the main content of this research is summarized,and the future research is prospected from the aspects of accurate reliability calculation and increasing the dimension of proactive prediction.