Research On Optimization Of Ride-Hailing Bus System

Recently,the rapid development in mobile internet technology has greatly enabled a new generation of demand-responsive public transit.Ride-hailing bus is a kind of demand-responsive service with flexible routes,non-transfer convenience and high comfort.This type of public transit service holds the promise to satisfy multilevel travel demands.The paper focuses on potential demand,optimal routes,and paths of ridehailing bus,as well as ride-hailing service considering the multi-service level.It is expected that these works can provide the theoretical basis and decision-making support for solving practical problems.The main work and contribution of this paper are as follows:(1)This paper studies the potential demands of ride-hailing bus,and proposes a methodology for extracting the potential users of ride-hailing bus based on Smart Card Data(SCD).The methodology proposed here consists of three process: trip chain generation,origin-destination(OD)recognition and travel mode comparison.Based on the case study,the SCD of Beijing is processed for analyzing the spatial-temporal properties of passenger travel behavior and exploring potential users of ride-hailing bus.The results indicate that using ride-hailing bus for long distance trips of more than 10 km has a strong utility advantage,this advantage becomes stronger with the increase of travel distance.(2)Based on the passenger travel demands,a model is developed to optimize the ride-hailing bus routes.The model aims at minimizing passenger travel costs and operating costs,with suggestions for determining bus stop locations,routes and passenger-to-vehicle assignment based on a series of constraints,like operation standard,vehicle capacity and number of visiting stations.To solve the model,a method is designed based on the genetic algorithm(GA),whose optimization performance and operating efficiency are verified in the numerical example.Finally,a case study is addressed based on the potential demands of ride-hailing bus.The case study analyzes the impacts of speed,vehicle quantity and upper limit of route length on related parameters,and evaluates the optimal routes.(3)Considering the passengers' expected travel time windows,a ride-hailing bus route optimization model with temporal-spatial travel restriction is put forward.In addition,the penalty caused by partial unsatisfied service(unassigned passengers)is considered in the objective function.In terms of algorithm design,two solution methods based on GA and tabu search(TS)are designed respectively to solve small-scale examples to determine the parameter values of algorithms.With the comparison of numerical results,it is found that the TS is an efficient way in solving the problem.A case study based on the travel demands in Beijing is conducted to analyze the impacts of speed and vehicle capacity on related costs,travel time and distance in the partial service and complete service scenarios.(4)For the ride-hailing bus routing optimization problem,the paper analyzes the impact of traffic congestion on unpunctuality,and establishes a time-dependent ridehailing bus service optimization model that explicitly considers path flexibility between nodes to be visited.It is noting that the decision-making considerations in bus route planning,timetable and passenger assignment are concurrently integrated into the model.Meanwhile,the path choice between nodes is determined to further optimize the bus routes.Furthermore,the objective function considers the penalty of service delay caused by actual road conditions.Then,a hybrid metaheuristic that combines TS and variable neighborhood search(VNS)is developed to solve the model,where a dynamic programming method is used to deal with the fixed sequence arc selection problem(FSASP).A numerical example experiment is conducted to verify the effectiveness of TSVNS.Finally,based on the case study with the potential demands and actual road network of Beijing,the paper considers the multiple paths and passenger distribution,and investigates the effects of time-window and traffic congestion,as well as the benefits from path flexibility on results.(5)Introducing a multi-service level ride-hailing public transit with the automatic driving technique.In the model formulation,the objective function aims to maximize the profit of operation,while the model considers two types of constraints,namely,the features of different levels of reservation service and operating characteristics of autonomous electric vehicles.It is noting that the model can optimize the route,timetable,passenger assignment and charging plan collaboratively.According to the travel demands in each cycle,the solution method is designed based on the adaptive large neighborhood search(ALSN)framework,in which the greedy algorithm is used to obtain the initial solution,and the destroy and repair operators are proposed for demand nodes and charging stations.In order to verify the model and algorithm,numerical examples and simulation experiments based on the actual road network are conducted,with analysis of the algorithm effectiveness,service levels,profits,and costs of the combinations for different passenger distribution in the fast charging and battery swapping scenarios,as well as the impacts of electric power and charging rate on related costs.