Research On Visual Analytic Method For Anomaly Detection And Diffusion Of Mrtro Passenger Flow

The urban rail transit system has become an important part of China’s urban public transportation due to its own advantages such as high speed,large capacity,and high safety.However,some emergencies or large-scale activities can cause sudden changes in daily passenger flow,which in turn leads to the formation of major passenger flows on the rail network and seriously affects the normal operation of rail transit and the safety of pedestrians.Therefore,the detection of abnormal passenger flow in rail transit and the excavation of abnormal passenger flow propagation rules can help formulate effective emergency measures and improve rail service levels.In this context,this paper fully considers the characteristics of the periodic change of metro passenger flow based on the metro smart card data,studies the abnormal fluctuation of rail passenger flow in Beijing,and proposes an abnormal passenger flow detection model.In addition,based on the results of anomaly detection and data analysis,the diffusion laws of abnormal passenger flow in the metro network is explored,and the detection,verification,and diffusion of abnormal passenger flow are displayed in a visual manner.Finally,a set of emergency decision-making support platform for subway passenger flow intelligent monitoring and control system in the face of heavy passenger flow is designed and implemented.Specifically,the research content of this article is divided into the following three parts:(1)Combining dynamic mode decomposition method,an abnormal passenger flow detection model based on low-rank dynamic mode decomposition is proposed.The model fully considers the influence of noise existing in the original passenger flow,decomposes it into a noise matrix,a sparse anomaly matrix,and a low-rank passenger flow matrix,and introduces dynamic mode decomposition into the low-rank matrix to obtain the internal characteristics of passenger flow.Compared with traditional anomaly detection methods,the model has higher detection accuracy and robustness.Combined with the spatiotemporal diffusion algorithm,the diffusion law of passenger flow in the metro is studied.(2)Based on abnormal passenger flow detection results and diffusion analysis,a visual analysis method is designed and implemented,which visualizes abnormal passenger flow detection,verification and diffusion analysis respectively,and supports multi-view interactive exploration of metro anomalies.(3)The intelligent monitoring and control system of metro passenger flow is designed and developed.The system uses prediction technology and monitoring technology to realize the monitoring module of line network,station detection module and statistical analysis module for passenger flow monitoring and early warning.The system provides technical support for coping with the impact of large passenger flow under the network operation of rail transit.This thesis is based on smart card data,combined with Weibo data,through a large number of experimental results proved that the proposed method is usefulness and validity.The obtained diffusion laws of abnormal flow can provide preliminary theoretical reference for subway managers and related researchers,the proposed visualization methods also offer an effective way to display and analyze the anomalies.