Research On Passenger Travel Path Selection In Rail Transit Network Based On Learning Effect

Under the strategic policy of giving priority to the development of rail transit in my country,the optional routes of passengers in the road network are becoming more and more diverse,but at the same time,the randomness and diversity of passenger route selection methods also increase the complexity and inconsistency of the rail transit network.certainty.Mastering the rules of passengers’ route selection is of great significance for the organization and operation management of urban rail transi.However,the traditional route selection research methods rarely consider the adaptive learning behavior of passengers during travel,that is,individuals face complex Self-adjusting decision-making processes that arise when the environment changes.According to the shortcomings of the existing research,this paper takes the traveling passengers in the rail transit network as the research object,uses experience weighted attraction and reinforcement learning respectively to explain the learning effect of passengers in the routing behavior,and constructs a daily-variable routing model based on the corresponding theory.,analyzes the passenger’s route selection law and the evolution of route flow,and finally uses the multi-agent-based simulation method to verify the model results,and summarizes the relationship between the microscopic behavior of passengers and the change characteristics of macroscopic road network flow.The main work of this paper is as follows:(1)Summarize the properties of the experience-weighted attraction model and the reinforcement learning model,and analyze the applicability of the two models by comparison,so as to provide theoretical support for the subsequent update of the daily-variable path selection model.(2)Inductively analyze the influencing factors of passengers’ route selection behavior in the rail transit network,and design a questionnaire on this basis to obtain research data and establish a basis for parameter setting in the model.(3)Construct the generalized travel cost function of the route and introduce the delay time cost,and at the same time improve the congestion cost coefficient in various situations.In the experience weighted attraction model,the revenue value of the passenger route selection result is determined by the route generalized travel cost.In the reinforcement learning theoretical model,the passenger perceived travel cost is derived according to the route generalized travel cost and used as a satisfaction evaluation index to update the passenger’s travel cost.Path selection probability.(4)Using the experience weighted attraction to redefine the route selection update rules of passengers,establish a daily-variable route selection model with the passenger’s travel experience as the main parameter.Rational choice probability and empirical choice probability are added to describe the influence of individual rational cognition and travel habits on passengers in the process of route selection.(5)Using reinforcement learning and establishing a daily-variable route selection model according to the passenger’s satisfaction evaluation index for the route selection results.The evolution properties of path traffic under the influence of different strengths of learning behavior ability are verified and analyzed by simulation.