Research On Recycling Route Optimization Of Broken Shared Bicycles Based On Stochastic Demands

In recent years,shared bicycles have developed rapidly.As a "green and healthy" mode of travel,they have been favored by people and have become an effective way to solve the "last mile" of urban traffic,all over the streets and alleys.However,there are still many problems in the actual operation process.Among them,the difficulty of recycling shared bicycles is particularly prominent.For many companies,they are negligent or unable to manage this aspect,resulting in few or no available vehicles when users use the car.Seriously affect the user experience.At the same time,most of the broken bicycles that are not recycled in time will become scrapped vehicles over time and pile up on the roadside.This is also the main reason for the emergence of "shared bicycle cemeteries" in many cities.Reasonable recovery of broken shared bicycles is of great significance to companies,individuals and society.Therefore,this article focuses on the recovery of broken shared bicycles.Therefore,this article focuses on the recovery of broken shared bicycles and provides an effective solution considering the risk preference and random demand of decision makers.This paper analyzes the spatial characteristics of shared bicycles based on the actual operating data of more than 3 million Mobike bicycles in Beijing in 2017.Focusing on the three characteristics of broken shared bicycle recovery: shared bicycle parking has the characteristics of no pile constraints,the recovery of broken shared bicycles needs to be timely,and the demand for recovery of broken shared bicycles is random,and a two-stage broken shared bicycle recovery network has been constructed.: The first-level network is the virtual recycling point clustering network;the second-level network is the virtual recycling point-recycling center line network.Based on the analysis of spatial characteristics,this paper takes the 4km×4km area of Haidian District as an example to obtain 25 virtual recovery points by performing K-means clustering on actual operating data.This method is more stable and more stable than the clustering of broken car data.In line with the actual use of users.This paper abstracts the actual problem of broken shared bicycles recycle into vehicle routing problem with stochastic demands,and establishes an optimization model for the recovery path of broken shared bicycles considering random demand.The objective function is the minimum total cost,including three parts: vehicle activation cost,vehicle path cost,and handling cost.The two major constraints of vehicle capacity and travel distance are mainly considered.Among them,the vehicle capacity constraint involves random variables.This paper introduces the parameter α reflecting the risk preference of decision-makers to construct random capacity opportunity constraints.In order to solve the vehicle routing problem with stochastic demands,this paper designs a variable neighborhood hybrid genetic algorithm.Constructed five different neighborhood structures,including within and between paths,to search for optimal solutions with variable neighborhoods to improve the local search capabilities of genetic algorithms.Finally,actual operation data is used as a calculation example to verify,and a prior sequence path plan under the risk preference of a certain decision maker is obtained.The algorithm has been run many times,and its convergence and stability are good,which shows the effectiveness of the algorithm.At the same time,the impact of different decision makers’ risk appetites on estimated costs and actual costs is analyzed through data experiments.