Study On Abnormal Identification And Short-term Prediction Of Subway Passenger Flow Under Station Closure Of Large-scale Activities

With the continuous expansion of people’s demand for urban rail travel,problems such as urban rail traffic congestion have become the norm.Especially under the station closures of short-term large-scale activity,a large number of gathering and distributing passenger flow and the transfer of passenger flow at closed stations will eventually cause congestion on the network around the event site.Based on the prediction of inbound and outbound passenger flow,it can provide data support for station staff to carry out prospective passenger flow control.Therefore,it is of great significance to study the prediction for inbound and outbound passenger flow under the station closures of short-term large-scale activity.However,there are still problems in the existing research on the site that have not been solved:On the one hand,due to the complex transmission mechanism of passenger flow,the identification of the affected temporal-spatial scope of the station in the case of station closure is inaccurate.On the other hand,due to the dual effects of the short-term large-scale activity and station closure measures lead to inaccurate passenger flow prediction results.In response to the above problems,this paper has done the following work:(1)Analysis of passenger flow characteristics under the station closures of short-term large-scale activityFirstly,the problem of station closures for short-term large-scale events is described and the research boundary is determined;secondly,the temporal-spatial characteristics and passenger flow components under the station closure of short-term large-scale activity are summarized;finally,a station classification index system based on multi-source data is constructed.It is found that the passenger flow is phased and abrupt in temporal characteristics and spatially has the characteristics of decreasing with the increase of distance and transfer times.The affected passenger flow has fluctuation and large amount;The characteristics of the network are related,which lays the foundation for the following research.(2)Research on the temporal-spatial scope of the impact of station closures of a short-term large-scale activityIn order to distinguish the affected and unaffected passenger flows under the station closure of short-term large-scale activity,a novel combination method of symbolic approximate clustering and derivative dynamic warping algorithm(SAX＿CP-DDTW)is proposed to identify the affected temporal-spatial scope of stations.On the one hand,the time series is segmented based on the abrupt points of the passenger flow time series,and on the other hand,the spatial and temporal impact degree of the station closures of short-term large-scale activity is quantitatively analyzed by synthesizing the abnormal changes in the number and trend of the inbound and outbound passenger flow.The model is compared with the SAX＿CP and SAX-DTW algorithms through a numerical example,which shows the accuracy of the method in the identification of quantity anomalies and trend anomalies and the stronger interpretability of the identification results.(3)Research on passenger flow prediction based on residual error correction of spatiotemporal influence rangeIn order to achieve accurate prediction of overall passenger flow and affected passenger flow,an ensemble Dynamic Factor Model and Extreme Learning Machine optimized by Sparrow Search Algorithm an extreme learning machine combined error correction prediction(EDFM-SSA-ELM)model is proposed to predict the passenger flow in and out of the station in the event of a short-term large-scale event closure.First,a DFM model based on temporal and spatial features is constructed to predict the overall passenger flow;secondly,according to the above-identified prediction results in the affected temporal and spatial scope,the SSA-ELM model is used to predict the residual,and the The preliminary forecast results are revised to obtain the final inbound and outbound forecast results.(4)Case Study of Beijing SubwayTaking the station closure of short-term large-scale activity of Beijing Subway Olympic Sports Center Station as an example,the case analysis verifies the validity of the identification method and prediction model proposed in this paper.The results show that: 1)the SAX＿CP-DDTW model proposed in this paper is more interpretable and accurate than the existing methods in the identification results of the space-time influence range under the event of a short-term large-scale event closure;2)the EDFM-SSA-ELM passenger flow proposed in this paper The predictive model has a certain improvement in the predictive performance of the existing baseline model.It solves the problems that it is difficult to accurately identify the impact range of passenger flow and the accuracy of short-term passenger flow prediction is low under the station closures of short-term large-scale activity.The method proposed in this paper provides theoretical and technical support for station passenger transport managers to formulate forward-looking passenger flow control measures in the case of short-term large-scale event closures.Figures:44,Tables:41,References:59.