Short Time Prediction And Traffic Organization Research Of Urban Rail Transit Based On Spatio-temporal Correlation Analysis

Passenger flow research has always been the focus of research and analysis in the whole process of urban rail transit from planning and construction to operation and then to dealing with emergencies.With the overall development of urban rail transit system in China,the research on passenger flow has also put forward higher requirements: more accurate,faster and more comprehensive.The short term passenger flow of urban rail transit affects all aspects of its operation plan closely,which causes the research demand for rail transit short-term passenger flow prediction.In the past,the research on short-term prediction mainly focused on the prediction of road traffic flow.In this respect,the research of researchers at home and abroad has formed a more mature achievement.Later,it used it for reference and applied to the forecast of urban rail transit passenger flow,so that the short-term passenger flow prediction of urban rail transit has stepped into a new stage.However,the current prediction method is far from enough,and the accuracy of prediction is far from enough.Therefore,this paper puts forward some thinking and exploration.The main research work is as follows:(1)The temporal and spatial distribution characteristics of short time passenger flow are analyzed from the aspects of research scope and research angle,in terms of year,month,week and day,so as to help better grasp its rules.(2)In order to make the prediction of the short-term passenger flow of urban rail transit more accurate,the influence of the passenger flow in the rail transit section and the passenger flow of the station is analyzed from the angle of space related and the time related,and how to put forward the prediction method for this effect is also discussed.All the sections are divided into related groups according to the spatial correlation,and the sections in each related group are predicted,which avoids the mutual influence between the unrelated sections,which is more reasonable and accurate.By means of autocorrelation analysis and partial autocorrelation analysis,the time interval of passenger flow prediction is determined.(3)The support vector machine regression model is applied to the short-time prediction of the urban rail transit passenger flow,and the selection of the kernel function,the selection of parameters and the evaluation index are analyzed in detail,and the K nearest neighbor algorithm is applied to the short time prediction of the passenger flow in the import and export of urban rail transit,and the sequence of the passenger flow sequence is calculated.The correlation coefficient is assigned to the K nearest neighbor value to get a more reasonable result.(4)How to make use of the obtained passenger flow forecast results to make a reasonable driving plan for the traffic and operation of urban rail transit,improve its passenger capacity and service level,and reduce the cost and waste at the same time.(5)The correctness and correctness of the theory and algorithm are verified by the support of line 2 of Zhengzhou urban rail transit.