Optimal Strategy Based Passenger Simulation And Behavior Inference For Urban Rail Transit Network

The large-scale urban rail transit network played a significant role in city public transit.To take fully advantage of urban rail transit and improve its service,it is necessary to learn the passenger behavior and travel pattern and provide flexible operation plans such as long-short loop and skip-stop operation plan.Simulation based passenger behavior simulation is one of the key technologies to learn passenger pattern.In the previous research,passenger paths are determined in the simulation initialization and the path for each passenger is fixed.Meanwhile,the paths are calculated based on metro lines,which couldn't demonstrate the passenger behavior under the multi-operation plan situation.It is necessary to study the passenger simulation and behavior inference with the multi-operation plan situation.By analyzing the passenger behavior,this study tried to propose a new passenger simulation and behavior inference method based on optimal strategy.This study firstly analyzed the characteristics of passenger behavior and established the route-based urban rail transit network.The optimal strategy based passenger travel process was proposed.The depth-first path generation algorithm was proposed with the basic optimal strategy.To make the model more real,the capacity-constrained optimal strategy model and the simulation-mixed iteration algorithm,link cost updating algorithm and simulation model were designed.Finally,the feasibility and availability of the optimal stagey based passenger simulation was validated in the empirical study on Beijing metro.The main research contents are as follows:(1)The framework modeling of optimal strategy based passenger simulation.Firstly,this study analyzed the passenger behavior characteristics and proposed the route-based topology network.To overcome the shortcomings of the previous passenger simulation methods for the multi-operation plan situation,this study introduced the optimal strategy to demonstrate the flexible passenger behavior and determination process with the multi-operation plans.Finally,a simulation-mixed iteration algorithm was proposed to infer the passenger behavior.(2)The depth-first path generation for optimal strategy.Analyzing the characteristics of optimal strategy,this study proposed a new concept:strategy node.Comparing the breadth-first and depth-first searching algorithm and analyzing the characteristics of strategy node,the depth-first path generation for optimal strategy was proposed,which is the basic algorithm for optimizing the optimal strategy based passenger simulation.(3)The capacity-constrained optimal strategy model establishment and simulation.Considering the train capacity constrains in the real situation,the effective depart frequency was proposed to simplify the capacity-constrained optimal strategy model.This study proposed the optimal strategy based passenger simulation.Comparing the actual and simulated travel time,the under-over index algorithm and least-square link cost updating algorithm were proposed to update the optimal strategy.When the simulation results converged,the simulation process finished.(4)The validation of empirical study.Beijing metro network was taken as the empirical study and the route-based Beijing topology network was proposed.After practical survey in the network,the simulation parameters such as accessing and transfer time are obtained.Comparing the simulated travel time and actual travel time,which validates the feasibility and availability of the optimal stagey based passenger simulation.Analyzing the passenger behavior and section loading rate from simulation results,the suggestions for better operation plan are proposedThe innovation points of this thesis are stated as below:(1)The optimal strategy based simulation and behavior inference.This study proposed the route-based topology network.After analyzing the passenger behavior under the multi-operation plans,the optimal strategy based passenger simulation method and system were proposed.The empirical study demonstrated the feasibility and availability of the optimal strategy simulation method.(2)The depth-first optimal strategy path generation algorithm with the stagey nodes.After analyzing the optimal strategy,the stagey node concept and the depth-first optimal strategy path generation algorithm were proposed.(3)The capacity-constrained optimal strategy model and algorithm.Based on the basic optimal strategy model,this study considered the capacity constraints and proposed the effective frequency to simplify the model.The simulation-mixed iteration algorithm was proposed to solve this problem.The under-over index based cost updating algorithm.Comparing the simulated travel time and actual travel time,the under-over index was proposed to determine the updating direction.And the waiting time filtering process guarantees the feasibility of the updating process.