Optimization Of Matching And Vehicle Scheduling In Car-based Shared Travel System

Rideshairng and one-way carsharing,as two main car-based shared travel models,are still under exploration and development in China.With the process of commercial operation,a series of operational problems are highlighted,which results in a low service level,and thus leads to the loss of existing users.In such a context,faced with operation problems,how to design a scientific shared travel system and formulate an efficient management plan to improve the service level and ensure a positive and sustainable development of the system is one of the most concerned research issues to be solved.The paper targets at Optimization of matching and vehicle scheduling in car-based shared travel system.Based on the system optimization theory,together with multiple disciplines such as Operational Research and Probability Theory,the research mainly conducts a deep research in inaccurate assessment of the ridesharing trip cost,ignoring user experience in dynamic ridesharing systems,and vehicle imbalance problem(VIP)in one-way carsharing systems(OWCSs),through systematic analysis and simulation as well as quantitative and qualitative analysis.The main researches and achievements are as follows:(1)A ridesharing model with travel time uncertainty is proposed to improve the effectiveness of ridesharing trips.Ride-sharing involves a joint trip of at least two participants who share a vehicle and must coordinate their itineraries,it can reduce car traffic,and thus is an effective way to mitigate traffic congestion.Because travel times are stochastic in reality road network,it may lead to ineffective ridesharing matches given by traditional deterministic matching models,which reduces the user experience and effect of ridesharing systems.In view of this,we assume that travel time is stochastic and follow a general distribution that has a positive lower bound,and propose a ride-sharing problem in road networks with travel time uncertainty.To solve this problem,this study first introduces the generalized trip cost functions for both driving-alone and ridesharing trips,and analyzes their mathematical properties.Then,the optimal departure times for both two trips are given,and a bi-objective ride-sharing matching model is proposed to maximize both the total generalized trip cost saving and the number of matches.Numerical results show that the ridesharing model with the consideration of travel time uncertainty can achieve a better matching scheme than the traditional deterministic model.(2)A dynamic ridesharing system based on user experience is proposed.Dynamic ridesharing system is a newly-rising intelligent travel system,which could effectively integrate transportation resources,alleviate congestion and reduce pollution via real-time matching among riders and drivers.User experience plays an important role in the sustainable development of dynamic ridesharing system.However,the traditional dynamic ride-sharing system only cares operational efficiency,ignoring user experience.In view of this,this paper develops a new dynamic ridesharing system considering user experience,by introducing three mechanisms,that is,real-time feedback,priority allocation and fixed-match elimination.Numerical results show that,compared with the traditional system,the proposed system could not only reduce the users' waiting time sharply,improve the user experience,but also enhance the system efficiency.(3)A OWCS with ridesharing options is proposed.Carsharing can effectively reduce the total number of cars,thereby reducing the road occupation of cars,parking lot and other areas,and it is also an effective way to alleviate traffic congestion.Since travelers' demand is asymmetrically distributed across time and space,some carsharing stations can have excessive empty vehicles but no spared parking space,whereas some can be short of carsharing vehicles and cannot serve users' trip requests.This VIP directly affects the service quality and system performance of OWCSs.To solve the VIP in OWCSs,an innovative hybrid operator-user-based relocation scheme is proprosed by integrating the existing OWCSs with ridesharing options.Then,a biobjective mixed-integer linear programming(MILP)problem is formulated with the aims of maximizing system profit and minimizing total user cost.A solution method based on the hybrid genetic search with adaptive diversity control(HGSADC)is developed to solve this NP-hard problem.Numerical examples illustrate that the proposed solution method can produce highquality solutions within acceptable computing time.We also show that both carsharing operators and users can benefit from the proposed integrated carsharing and ridesharing scheme.Through theoretical proof and numerical experiment,this research conducts a deep research in inaccurate assessment of the ridesharing trip cost,ignoring user experience in dynamic ridesharing systems,and VIP in OWCS.The corresponding solutions are given and some novel research results are obtained.This study has important theoretical reference value for the establishment and improvement of the theoretical system of car-based shared travel system.