A Hyper-Heuristic Method To The Integrated Bicycle Sharing Network Design

In recent years,with the enhancement of people’s environmental awareness,a large number of shared bicycles have flooded into the city.However,due to the imperfection of the supporting facilities,the lack of bicycle lanes and the random parking of shared bicycles not only make the urban road traffic face great pressure,but also make people’s travel safety difficult to be guaranteed.In order to improve the efficiency and safety of public transportation system and promote more travelers to choose bicycle transportation mode,this paper studies the design of shared bicycle network.The innovation of this paper lies in the establishment of a bi-level model which considers the selection of bicycle stations and bicycle lanes together and the use of a hyper-heuristic method.The model has two decision variables: whether to build bicycle station near the origination and destination,and whether to build bicycle lane on a alternative road section.The upper model aims to minimize the total cost,which comprehensively considers the time cost and construction cost.The lower model expands the muti-mode selection model and route selection model,so as to divide the traffic mode and allocate the traffic volume.In order to solve the bi-level model,this paper adopts the sequence based selective hyper heuristic method and combines it with Frank Wolfe algorithm.When the upper model determines the combination of bicycle lane and bicycle station,the lower model will finish the division of traffic mode and the selection of bicycle route.Frank Wolfe algorithm is used to allocate the traffic flow under different modes.After that,the upper model receives the time cost.The code is written in Python programming language.After several iterations,the optimal combination of site and lane and the lowest cost can be obtained.A small network and a medium-sized network are used for instance verification.The small network is used to test the correctness of the algorithm.The optimal combination scheme of bicycle lane and station is calculated and verified by exhaustive method.The result shows that the scheme can minimize the total cost.Taking the time cost and the fixed facility cost in the objective function as the abscissa and ordinate axes respectively,the Pareto optimal frontier is drawn.The points on the curve can provide a better reference space for decision-makers as they have less target conflicts.Taking Nguyen-Dupius network as amedium-sized example,the optimal combination scheme of bicycle lanes and bicycle stops under various scenarios is calculated.It can be concluded that the setting of bicycle lanes and bicycle sharing stations is helpful to improve the operation efficiency of the traffic network and save the total cost.But it’s not that the higher the construction cost,the better.Setting up useless stations will only increase the cost of fixed facilities,and too many bicycle lanes will affect the efficiency of motor vehicles.Therefore,we should fully consider the contradiction between fixed cost and time cost,and use the solution on Pareto’s optimal frontier to get the construction plan with the minimum cost.