Study On Passenger Flow Effects Analysis And Prediction Model Of Rail Transit Stations Under The Influence Of Built Environment

China is in the stage of rapid urbanization and motorization development,urban population and small car ownership are rapidly increasing,leading to urban sprawl,environmental pollution and traffic congestion problems are increasingly prominent,urban land development mode(TOD)oriented by public transportation such as urban rail transit is an effective strategy to enhance station-city integration and alleviate urban disease problems.The built environment is closely related to urban residents’ traffic travel behavior,and the lack of effective articulation and matching guidance between rail transit planning and construction and urban functional layout leads to high peak full load rate of rail transit,uneven distribution of line network operation load,and significant spatial variability in station flow restriction and line network density homogeneity,etc.How to capture the influence of complex spatial patterns of passenger flow of different types of stations from TOD built environment on how to capture the influence mechanism of complex spatial patterns of passenger flow on rail stations from the impact of TOD built environment,and consider the spatial dependence and interaction effect of built environment,it is important to carry out research on station traffic prediction.Existing studies have not focused on theoretical exploration and empirical research on the spatial influence effect of TOD built environment on different station passenger flow,interaction mechanism and passenger flow prediction in rail station influence area.Therefore,this paper takes the interaction between TOD built environment and urban rail transportation as the theoretical basis,and captures the spatial characteristics of rail station passenger flow and TOD built environment based on the fused and processed multi-source heterogeneous data.Finally,the research focuses on the supply and demand relationship between TOD built environment and rail station passenger flow,and predicts the passenger flow of classified stations under the spatial dependence and interaction effect of built environment.The study covers the following five aspects.(1)Normalize the multi-source heterogeneous data,extract the temporal distribution pattern of passenger flow at rail stations by kernel density and probability density distribution function,and visualize the spatial distribution characteristics of passenger flow at rail stations and the TOD built environment in the 800 m direct influence area and 1500 m radiation influence area with the help of spatial analysis technology,so as to provide a data basis for later research on the relationship between passenger flow at rail stations and the spatial characteristics of TOD built environment and the mechanism of interaction influence.(2)The "node-place" model was improved by adding accessibility indexes,and the global Moran index was introduced to build the evaluation method of TOD built environment development level in the rail station influence area;for the situation that passengers’ spatial and temporal distance perception of subway lines changes with the station platform in the actual travel interchange process,the spatial syntax theory and the method of measuring accessibility and walkability is based on spatial syntax theory and expanding convex spatial partitioning.The study shows that the improved "node-place" model has better identification of "balanced stations";the higher the synergy between the level of TOD built environment development and passenger flow in the influence area of rail stations,the greater the influence of the level of built environment development on station passenger flow.(3)A regression model based on global constant parameters and local variable parameters was developed to analyze the interdependence and spatial effects between passenger flow and TOD built environment at rail stations in Shenzhen,and the performance of five multiple linear regression models was compared and evaluated in terms of accuracy and fitting effects;the spatial variability of the development level and factors of TOD built environment on passenger flow at classified stations was explored.The study shows that the geographically weighted regression model(GWR)has more reliable estimation results in portraying the spatial heterogeneity of the dependence between passenger flow and TOD built environment variables;the effects of TOD built environment factors on passenger flow show significant regional differences.The differences between the factors of grade II and grade IV stations are the smallest,while the differences between the factors of grade V and grade VI stations are the largest.(4)Discuss the influence of the modifiable areal unit problem(MAUP)on data aggregation and modeling results,verify that 800 m space scale is the optimal influence area of TOD built environment on passenger flow,and use geographical detector from three aspects of direct individual effect,indirect interaction effect and interaction effect variation rate(IEVR).The built environment driving factors influencing the spatial differentiation of passenger flow at six railway stations are explored,and the influencing mechanism of passenger flow at different railway stations was analyzed.It is found that density,walkability,service ability and mixing degree are sensitive to MAUP,and their interaction effects play an important role in the passenger flow rebalancing strategy of rail stations.IEVR not only helps to quantify the difference between passenger flow and TOD built environment,but also provides a new quantitative method for exploring the influence mechanism of passenger flow.(5)The nonlinear regression function between the nature of land use and the passenger flow demand of rail stations was constructed with the constraints of road network distance and land use location information,and the attenuation law of passenger flow generation of various land uses with the distance to the rail stations was obtained.On this basis,considering the spatial dependence effect and interaction effect of TOD built environment of classified stations,a passenger flow prediction method based on GWR model and random forest(RF)algorithm was proposed,and the differences of the two prediction models in the accuracy,fitting effect and applicability of six types of rail stations were compared and evaluated.It is found that the decay law of passenger flow generation rate with accessibility is more in line with the exponential decay model,and the overall passenger flow generation rate decays fastest when the distance between different land use and rail station is 800 m.The passenger flow prediction method based on classified stations improves the prediction accuracy.RF algorithm is better than GWR model in terms of overall prediction accuracy and fitting effect,and the average absolute percentage error is reduced by 3.33%.This model is more suitable for grade III stations to grade V stations,and its prediction results is not effective when dealing with grade VI stations.The research results enrich the theories and methods of the relationship between the built environment and rail transit passenger flow,and can provide basic support and empirical reference for dealing with urban diseases caused by the imbalance and disconnection between rail transit development demand and urban planning.