| Urban rail transit has the characteristics of safety,efficiency,fast and punctuality,and environmental friendly,which is an important direction for the development of modern public transportation and has developed rapidly in recent years.With the gradual expansion of the urban rail transit network and the increasing passenger demand,the operating environment of urban rail transit has become more and more complex,which leads to the increasing difficulty of train operation control.During daily operations,disturbances will inevitably occur,such as equipment failures,passenger flow changes and other random factors that cause train delays.If effective measures are not taken to adjust the train operation in a timely manner,it will cause disorder in the operation of the urban rail transit line,and passengers stranded on the platform.Therefore,the study of train regulation based on passenger demand has important practical significance to ensure the operation efficiency and service quality of urban rail transit system.With the improvement of the automation level of urban rail transit systems,train dispatching and operation control are also developing in the direction of centralization and intelligence.How to combine train timetable and speed trajectory for integrated train regulation has become a hot research topic in recent years.This dissertation studies the integration-oriented train regulation methods from the three levels of passenger demand,train dispatching,and train control.Specifically,the main contributions of this dissertation are summarized as follows:1.In view of the impact of the unexpected disturbance on the train operation and waiting passengers during peak hour,an passenger-oriented train regulation method is proposed.A detailed passenger flow model which considers the capacity of the train and platform is developed to accurately describe the status of passengers waiting,getting on and off the train.Further,considering the impact of the passenger flow on the train dwell time,we develop a train rescheduling model under capacity limitation with the aim to reduce the train delay and the number of the stranded passengers.A cuckoo search based train regulation algorithm is designed to solve the problem.The simulation results show that the proposed method can obtain a high-quality rescheduled timetable in real time,and significantly reduce the influence of the disturbance on train operation and waiting passengers.2.With the consideration of the train energy consumption cost by regulation under the saturated passenger flow condition,an automatic train regulation method based on the preset recommended speed trajectory is proposed.We establish a train energy consumption calculation model based on the utilization of regenerative braking energy by considering the load variation during train operation.The binary variables are introduced to establish the mapping relationship between the train operation level and the preset recommended speed profile,running time and energy consumption.By adopting the time tuning strategy,we propose a mixed-integer nonlinear programming model with the goal of reducing train operation energy consumption,delay time and the number of stranded passengers.Based on the Big M method,the nonlinear constraints in the model are linearized and reconstructed,then the original model is transformed into a mixed integer linear programming model,and the mathematical programming software CPLEX is adopted to solve the proposed problem.The simulation results show that the proposed method can obtain an optimized regulation solution under in real-time,so that the train can resume the original schedule as soon as possible,and the number of stranded passengers and the energy consumption of the train can be reduced.3.In view of the limitation of the time tuning regulation strategy in reducing the train delay time and the number of stranded passengers in peak hour,an automatic train regulation method based on the combined strategy is proposed.By selecting train operation levels online,a combination strategy of train holding and time tuning is adopted to further balance the train operation interval.With the aim to reduce the number of stranded passengers and train delay time,an automatic train regulation model based on the train holding and time tuning strategies is developed.A simulation based optimization algorithm is proposed to solve the problem.A dynamic passenger flow simulation model is adopted to determine which train should be held and update the train holding constraints of the optimization model.The simulation results show that compared with a single regulation strategy,the regulation solution of the combined strategy can further reduce the train delay time and the number of passengers stranded on the platform,which significantly alleviates the pressure on the operation of urban rail transit.4.Aiming at the layered architecture of train dispatching and control of the rail transit system and considering the optimization of train speed profiles in real time,an integrated train regulation method based on deep learning and hybrid search is proposed.Considering the real-time requirements of train speed trajectory optimization,we develop a train speed trajectory optimization model and adopt the genetic algorithm to obtain labeled sample data,and the convolutional neural network is trained to fit the the mapping relationship between the input set(the train running time,the speed limit and gradient in each section)and the output set(the optimal train speed trajectory corresponding to the switch points and energy consumption).An integrated train regulation model combining the timetable and speed trajectory is developed with the goal of minimizing the train energy consumption and delay time.Then we design a train regulation algorithm based on the hybrid intelligent search to solve the problem.The simulation results show that the proposed method can simultaneously obtain the regulated and optimized train schedule and speed trajectory in a short time,and it reduces more energy consumption and delay time compared with the hierarchical optimization method. |
