Traffic Management And Control With Passenger Service Quality In Urban Rail Transit

With the increase of China’s urban population and the rapid development of urbanization,the urban rail transit(URT)system has an impressive expansion on network scale for the large traffic volume,rapidness,safety and high efficiency.In large and medium-sized cities,passenger traveling process is more dispersed and frequent,and shows dynamic and random characteristics with increasing the number of transfer activities in URT.While attracting a large number of passengers,URT undertakes the great transport pressure with the raised proportion of total passenger volume in the public transportation system year by year.For the increased passenger traveling demands and wide attention of service quality in the train operation,how to design the passenger-oriented traffic management and control has become the active research issue in recent years.According to passenger traveling demands and train operation characteristics under different scenarios(i.e.,normal operation and service disruption),we clarify the traffic management and control into three levels to improve service quality,including train scheduling of URT network,train rescheduling with fixed passenger route and train rescheduling with passenger route selection.Aimed to guide the normal operation in URT network,demand-driven timetable synchronization is optimized for the feeder and connecting trains to improve passenger satisfaction.Besides,caused by signaling system,rolling stock and track failures,etc.,service disruption induces service trains delay and stranded passengers at stations unable to board on trains,which propagates rapidly to the nearby transit lines through the transfer stations and affects operation reliability and following passenger travels in the whole network.Hence,train rescheduling optimization is proposed to provide efficient strategies for train recovery and passenger transport of the metro company.In comparison with previous studies,train capacity constraint is considered to model the passenger boarding,alighting and waiting processes,by integrating requirements of service quality under different scenarios.The linearized reformulation method is proposed for the large-scale rescheduling nonlinear problem with the shortened computation time.The main contributions of this thesis are summarized as the following four aspects:1.With current constraints of infrastructure condition,fleet size,etc.,assessment criteria of the service quality are analyzed for normal operation and service disruption to reflect the passenger satisfaction.For different passenger traveling characteristics,the fundamental model is firstly established for train management and control problems to meet the requirement of service quality under different scenarios.Then,we propose the closed-loop control framework and solving approaches to guide train normal and recovery operations based on passenger origin-destination demands.2.For time-varying passenger demands,we formulate a multi-objective programming model to jointly optimize network timetable,through considering the total passenger travel time and operational energy consumption.With the limitation of train capacity and pre-selection of passenger planned route,passenger waiting,boarding and transferring processes are simulated to analyze the correlation with train timetable,passenger traveling time and train energy consumption.3.Under the disruption scenario with partial blockage of URT,nonlinear mixed-integer programming models with Alternate strategy and Short-turn strategy are developed for the train rescheduling problem to alleviate the inconvenience for passengers and regain the nominal train regularity.To satisfy the real-time computing requirement,a novel method with coupling time-indexed and big-M formulations is firstly proposed to linearize nonlinear parts of the proposed model.During the incident with real-time information(like passenger demands and the duration time),a two-stage approach is also designed by solving the proposed linearized model: in Stage 1,the best strategy is selected with a series of historical data to obtain the initial timetable;the initial timetable is iteratively updated for the real-time rescheduling in Stage 2.4.During the large service disruption,we investigate affected passengers’ behavior and develop the passenger route selection model to provide feasible new routes with minimal traveling time.Then,a bi-level programming model is developed with the upper train rescheduling model and lower passenger route selection model and the iterative algorithm is designed to find the near-optimal solution.In the passenger route selection model,k-means clustering algorithm with modified criterion is adopted for a large number of passenger origin-destination pairs to cluster passenger demands and decrease decision variables,which shorten computation time and satisfy online optimization requirements.