Ridesharing Network Equilibrium Model And Its Applications

Shared mobility is one of the most promising applications in the modern urban transportation network.Since shared mobility allows travelers to travel at a low cost,its users increase rapidly around the world.Moreover,since many researchers suggest that shared mobility can effectively reduce traffic congestion and environmental costs,it is supported by many governments and scientific institutions.Ridesharing,as one representative of the shared mobility in private vehicle trips,has been greatly concerned.Along with the increasing of its users,ridesharing is deeply changing the urban transportation network.This study deal with traffic assignment problems(TAP)in the urban transportation network with ridesharing which is referred as to the ridesharing network.We investigate the impacts of ridesharing on the urban transportation network,and propose new methods for the TAPs in ridesharing network.Specifically,our contributions are given as follows:First of all,we introduce the theory of urban transportation network,and then extend traditional network equilibrium models.Based on the characteristics of modern urban transportation networks,we propose some representative problems in the modern networks,as well as the models and algorithms to solve them.By doing these,we lay a theoretical ground and prepare some useful tools for solving TAPs in the ridesharing network.Besides,according to the characteristics of ridesharing,we extend the user equilibrium(UE)in traditional urban transportation networks to the ridesharing user equilibrium(RUE),and propose the generalized travel costs,ride-matching constraints,and variational inequality(VI)models for the RUE problem.Different from existing RUE models,the proposed model removes stringent assumptions in the literature,and is able to describe different ridesharing services,which leads to a better applicability.Moreover,to meet the appeals of transportation networked companies(TNCs),i.e.,the companies providing ridesharing services,we provide a generalized travel cost function to describe the ridesharing pricing strategies that is based on the ridesharing supply and demand between the origin and destination(OD)of trips.An RUE problem under OD-based surge pricing strategy is proposed.In this problem,the generalized travel cost function is non-additive,which leads to the failure of existing RUE models in the literature.To solve this problem,we propose a novel VI model and demonstrate the existence and uniqueness of its solution.Moreover,we also design a parallel selfadaptive projection method to find the solution of the RUE problem.The designed algorithm incorporates the techniques of column generation and parallel computing so that it greatly reduces the computational times and in-memory requirement.Finally,we take the stochasticity of travelers' perception and the elasticity of travel demand into account,and propose a novel VI model for the stochastic ridesharing user equilibrium(SRUE)problem with elastic demand(SRUE-ED).The proposed model is more suitable for reality.Since existing algorithms cannot solve the proposed problem,we propose a method of ridesharing successive averages(MRSA)and a method of ridesharing successive weighted averages(MRSWA)to solve this problem.The MRSA extends traditional method of successive average(MSA)for solving the SRUE-ED problem,while the MRSWA further improves the computational efficiency by proposing a super-linearly step size sequence.