Urban Rail Transit Passenger Flow Forecast Based On Refined Land Use Data

For the new demand of refined city management for passenger traffic forecasting and the wide application of Internet data,this paper uses the Direct Demand Method(DDM)based on Geographical Weighted Regression(GWR)to research the rail transit passenger flow forecasting problem.By analyzing the surrounding environment factors such as land use,social economic factors and transportation facilities within the attraction level of the station,the relationship between the passenger flow and each element is established.This method has the advantages of low cost,quick response,and easy interpretation of results.This paper aims to research the characteristics of passenger flow,data acquisition and processing of passenger flow prediction,and model construction.The main contents include the following aspects:(1)Analyze the rail travel process of rail transit and influencing factors of urban residents from the perspective of land use.Summarize the characteristics of existing rail transit passenger flow forecasting methods,then combine spatial statistical techniques to propose a rail transit passenger flow forecasting method based on refined land use data.(2)Based on the influencing factors,we defined the target data sets and data sources required for the passenger traffic forecasting modeling of the rail transit stations.Due to the characteristics of the data and the difficulty of obtaining,a technical solution using various means is proposed for multi-source data collection.A data capture solution is designed for Internet data,and an applieation is developed to implement batch acquisition and regular update of Internet data.(3)Analyze the characteristics and existing problems of the acquired electronie map POI data,and use the semi-structured extraction method of text information to extract information from residential communities and office buildings to improve data availability.For the incompleteness,redundancy and complementarity of multi-source data attributes,the spatial attribute fusicrn algorithm and the non-spatial attribute fusion algorithm are used for data fusion respectively to obtain refined spatial geographic data.(4)Analyze the relationship between the passenger flow and the collected land use data.The multi-cornection method attracting range is calculated,and then we propose method of multi-connecting method to superimpose the area to achieve spatial division,which reflects the relationship of multi-circle layer discrete attenuation in the attraction range.By using exploratory regression to complete the modeling of variables and data preparation,we establish a geographically weighted regression model combined with spatial interpolation techniques to achieve direct forecasting of passenger flow at the station level.