Resilience-Oriented Rail Traffic Management Optimization Methods

Urban rail transit is a kind of fast and large-capacity public transportation,which greatly meets the travel demands of residents.However,unexpected incidents,such as signal failure and vehicle failure,are usually inevitable in the operations.At present,the traffic management task under emergencies mainly relies on human labors,and the emergencies are typically handled in separate stages.Since the disposal process of each stage does not consider the influence of other stages and lacks rigorous decision support,it may produce serious consequences of "small failure,large impact".Therefore,this paper introduces resilience as the evaluation metric of rail traffic management,which guides the train rescheduling optimization model constructed for the whole process of emergency development.Finally,from the perspective of traveling passengers,the resilience of the traffic dispatch command is further validated.The main work is as follows:(1)Resilience metric of urban rail transit during emergency: In order to scientifically and quantitatively evaluate the ability of rail traffic management system to respond to emergencies,a Bayesian network-based system resilience definition method is proposed.Specifically,a local search algorithm based on K2 score is constructed to extract the Bayesian network structure.Then,a method for correcting the initial search order of Bayesian network structure based on expert knowledge is proposed,and then three Bayesian network structures are proposed.Finally,the performance of the three proposed Bayesian networks is compared through the historical fault data of Beijing Metro,and the superiority of Bayesian network and expert knowledge is verified in terms of training time and prediction accuracy.(2)Urban rail transit rescheduling optimization method for system resilience improvement: Considering the emergency stages,involving the emergency handling process,fault handling process and recovery process,a train rescheduling optimization model is proposed based on the event-activity network for the improvement of system resilience.First,using the planned timetable and the real-time train position as input,an event activity network model for emergencies is constructed which defines the system recovery time(ie,the time recovery to the planned timetable)as the objective function.Then,binary decision variables are introduced,and the constraints of the model are defined according to the safety headway,section running time,etc.,so as to generate a rescheduling train timetable.Based on the big M method and the extended cutting plane method,the constraints in the model are linearized and reconstructed,and a branch and cut method is introduced to accelerate the optimization of the model.Finally,through a simulation example,it is shown that this method can reduce the system recovery time by adjusting the train operation under emergencies,and it is found that redundant resources such as station turnaround lines and reserved trains can effectively improve the system resilience under emergencies.(3)Evaluation and verification of passenger-oriented resilience metric of urban rail traffic dispatching: The primary task of urban rail transit is to provide travel services for passengers.Therefore,a simulation model is constructed based on passenger travel history data to verify the effectiveness of the event activity model and the branch-and-cut method.First,in order to verify the impact of different traffic management strategies(including reversed trains,short-turning,etc.)on the spatial and temporal distribution of passenger flow,a passenger travel-based simulation optimization model is constructed,which is based on the dynamic passenger demand and passenger boarding process.Then,the evacuation time of passengers on the platform under emergencies is used as an evaluation index to further analyze and adjust the train rescheduling decision to reduce the impact of emergencies on traveling passengers.Finally,simulation experiments are conducted based on the passenger demand data of Beijing Metro Line 1.The results show that the resilience-based optimization model proposed in this paper can effectively reduce the impact of emergencies on traveling passengers.