Research On Multi-semantic Feature Analysis And Space Optimization Of Rail Transit Station Area

In the context of urbanization and motorization,China is facing problems such as traffic congestion,environmental pollution,and disorderly urban expansion.Urban rail transit system,with its large carrying capacity,punctuality,and environmental friendliness,has gradually become the core backbone of the modern integrated public transportation network,easing many urban problems.However,with the rapid development of urban rail transit,there has been an imbalance and mismatch between rail transit and urban space in some areas.The station area is the place where the rail transit system is most closely connected to the outside world,and it is also the area where the contradiction between traffic supply and travel demand is most prominent.The imbalanced development of rail transit and station area space makes it difficult to fully utilize the transportation function and land use efficiency,causing congestion and resource waste.Therefore,it is urgently needed to comprehensively grasp and deeply understand the spatiotemporal characteristics of rail transit and station area space,as well as the interaction between the two from multiple perspectives,in order to provide scientific decision-making suggestions for rail transit construction and station area space planning.This paper,using multi-source geographic big data and artificial intelligence technology,combined with cross-disciplinary theoretical methods including urban planning and geographic information,researched three aspects of refined identification of rail transit station area characteristics,nonlinear effects of station ridership,and land use optimization modeling,along the research thread of the interaction between rail transit and station area space.This research provided a theoretical and practical basis for promoting coordinated development of urban space and transportation systems.Shanghai Metro is regarded as case study in this paper.The research content and main conclusions are as follows:(1)The identification of multi-perspective spatiotemporal characteristics of station areas is fundamental for finely characterizing rail transit and station area spaces,and exploring their current development and interaction.This study introduced the street landscape dimension to expand the characterization of station area vertical features,and extracted structured mobility-location-visual semantics using semantic models.Station area identification was achieved through a clustering method based on these semantic features.The results showed that mobility semantic clusters revealed passenger flow distribution patterns of rail transit such as early in-early out,early out-late in,double peaks in the morning and evening,and interaction features,which can better reflect the mobility of residents among stations compared to the traditional static ridership temporal characteristics.The location semantic clustering results demonstrate commercial and industry,residential,public,mixed-use,transportation hub,and undeveloped land use functional types of station areas.The visual semantic is a quantitative expression of the streetscapes features,and the clustering results showed a significant ring distribution.In order to further explore the interaction between rail transit and station area space,this study analyzed the co-occurrence and heterogeneity of multiple semantics from the aspects of attribute and location,validating that there is a close coupling relationship between transit ridership,land use,and streetscapes.The study found that stations with similar multiple semantics are clustered in the central commercial area and the residential area located in the junctions between the downtown and suburbs,while the differences in morphological function and development level still result in local heterogeneity of the characteristics of "partially similar" station areas.(2)Ridership is a core indicator for evaluating the operating status of rail transit and reflecting residents’ travel demand.Analyzing the impact mechanism of built environment elements on station ridership is a substantial part in understanding the interaction between rail transit and station area space,and providing theoretical support for optimizing station area space.To address the limitations of existing studies on the impact of passenger flow,which have failed to consider the nonlinear characteristics and spatial effects simultaneously,this study proposed a GW-XGBoost model to reveal the local characteristics of the impact of built environment factors on station-level ridership.Based on the fact that land use and street landscape are closely related to ridership,this study extracted station characteristics,multimodal connections,and socio-economic factors and analyzed the impact mechanism of ridership based on global and local nonlinear models.The results showed that compared with linear models and global models,the GW-XGBoost model had a higher fitting degree,indicating that the impact of the built environment on station ridership is more fitted with nonlinear and spatially heterogeneous research hypotheses.The ridership showed nonlinear changes in different directions as the built environment factors increased,and there were generally change thresholds.Among them,betweenness centrality,entrance and exit quantity,and commercial land use had the most significant impact on ridership.The impact mechanisms of factors had local differences.Commercial land use and bus stations had a greater impact on passenger flow in peripheral areas than in central urban areas,while the impact of entrance and exit quantity and elderly proportion was the opposite.(3)Transit-oriented development(TOD)is a critical concept in the integration of public transportation and land use,emphasizing the orderly and compact development of urban space under the guidance of public transportation.Employing TOD to guide station area land optimization is a significant measure to promote the beneficial interaction between rail transit and station area space and promote their coordinated development.Based on the TOD principles and strategies,this study proposed six planning objectives from different perspectives of people,transportation,land use,and environment.A TOD planning model was constructed by combining multi-sources geographic big data and linear and nonlinear fuctions,alleviating the oversimplification issue of existing TOD models.Then,the land use layout optimization schemes were obtained by using multiobjective genetic algorithm.The effectiveness of the optimization model was verified using the Fanghua Road subway station in Shanghai as an example.The results showed that the TOD planning model obtained through the developed optimization algorithm provided good candidate schemes in various optimization directions.This study found that there were consistencies and conflicts between planning objectives,and planners need to pay attention to the contradictions between rail transit ridership benefits and environmental burdens,and improve regional functional mix while maintaining local land use compactness.The optimized land use in the station area displayed high density and diversified development centered on the station,effectively improving passenger flow and land compactness.Compared to the pre-optimization state,the station area now resembles the TOD mode more closely.This study enriches and expands the research on the interaction between rail transit and station area space and has important theoretical and practical significance for optimizing station area space under the guidance of urban rail transit.