The Forecasting Of Passenger Demand And The Order Dispatch Method Under Hybrid Ride-hailing Service Modes

With the rapid development of the “Internet plus transportation”,ride-hailing service has become a significant part of urban transportation system with its innovative operation mode.In order to satisfy the diversified and personalized travel demands of residents,transportation network companies have gradually develope d a variety of service patterns.However,facing the rapid development of ride-hailing service,the existing researches only considered one type of service mode,and the demand forecasting framework and order dispatch method under the hybrid service modes were rarely investigated,which was not conducive to the improvement of vehicle operation efficiency and enterprise benefit.To this end,the dissertation studies the research entitled “The forecasting of passenger demand and the order dispatch method under hybrid ride-hailing service modes”,and establishes a research framework of “order analysis?demand forecast?order dispatch method?the efficiency improvement of order dispatch”,which is divided into the following aspects:(1)Identify the demand characteristics and key influential factors of express and ridesplitting ordersBased on the order data of Di Di Chuxing in Haikou,China,the spatial-temporal characteristics of express and ridespliting demands are compared.Then,the variables of historical order demand,the demand of neighbouring regions,travel time rate,point of interests(POIs)and weather are analyzed,so as to identify the influential factors of these two modes' passenger demands.(2)Establish an order forecasting model under hybrid ride-hailing demandsWavelet Transform(WT)is utilized to decompose and reconstruct the demand,and Fast Correlation-basd Filter(FCBF)algorithm is applied to identify the influential factors of the demand and determine the optimal input feature set.The n,a combined model based on WT-FCBF-LSTM(LSTM,Long Short-term Memory)is proposed to predict the hybrid passenger demand in different regions for different time intervals.Finally,the parameter tuning and validity analysis for the combined model are ca rried out.(3)Construct a ridespliting order dispatch model with condering ridespliting characteristicsBased on the characteristics of ridespliting service mode,e.g.the passengers' travel routes need to overlap partially or completely with each other,an order dispatch model of ridespliting service based on bipartite graph theory is established,with considering the scenarios of multiple passengers per group of order,three groups of orders at most for a driver,and the ridesplitting orders' response on the half-way.This dissertation designs corresponding accelerating strategies to speed up the selection of feasible matching,and a hierarchical approach(lexicographical goal programming)is adopted to solve the optimal matching model accurately.The case experiments are carried out to verify the effectiveness of the proposed algorithm.(4)Establish an order dispatch model of hybrid ride-hailing service modesConsidering the characteristics of hybrid modes,e.g.express and ridespliting services,in actual operation,such as the uncertainty of driver state(express or ridespliting driver),the heterogeneity of the maximum waiting time of different types of passengers,the priority of different types of passengers,an order dispatch model under hybrid ride-hailing service modes is established.Several unique characteristics of this problem are also examined,such as the proportion of different types of passengers,the priority of different types of services,and the proportion of initial vehicles with passengers.(5)Design an efficient order matching algorithm of hybrid ride-hailing service modesA fast and efficient hybrid greedy genetic heuristic algorithm is designed to deal with large-scale problems in actual operation for the order dispatch method under hybrid ride-hailing service modes.The initial solution is constructed by greedy algorithm,the crossover and mutation operations are designed according to the characteristics of hybrid ride-hailing service modes,and simulated annealing algorithm is applied to the optimal solution update of population.Three scenarios of express priority,ridesplitting priority,and random priority are designed to illustrate the practicality of the proposed algorithm.The accurate demand forecasting and order dispatch algorithm for ride-hailing service are particularly significant for improving the operational efficiency of ride-hailing vehicles and passengers' travel experience.The research results can provide theoretical guidance for the management,operation and sched uling of transportation network companies,and improve the real-time,accuracy and efficiency of urban ride-hailing management.And they also have important theoretical and practical significance for promoting the optimization and upgrading of the emerging service industry and the sustainable and healthy development of ride-hailing service.