Research On Short-term Prediction Of Entry And Exit Passenger Flow Of Urban Rail Transit Under Large-scale Activities

In recent years,urban rail transit industry has developed rapidly and has become an important way to people.At the same time,with the development of the economy and society,Large-scale activities are becoming more frequent.Large-scale activities will cause a large number of people gathering in a shorter period of time and a small space,resulting in a sudden large passenger flow in urban rail transit,which will have an impact on passenger safety and station operation.Accurate prediction of sudden passenger flow caused by large-scale activities is a necessary support for reasonable passenger flow organization and safe operation.Therefore,it is of great significance for short-term prediction of urban rail transit passenger flow under large-scale activities.Based on the research on the passenger flow law of urban rail transit under large-scale activities,this paper proposes a method for constructing and extracting the features of urban rail transit passenger flow under large-scale activities,and establishes a combined prediction model to realize the accurate prediction of entry and exit of urban rail transit passenger flow under large-scale activities.The main work of this paper is as follows:(1)Research and analysis on the changing law of urban rail transit passenger flow under large-scale activities.Firstly,the concept,characteristics and classification of large-scale activities are explained,and the concept and description indicators of urban rail transit passenger flow under large-scale activities are further explained.Secondly,the influencing factors of urban rail transit passenger flow under large-scale activities are summarized.Finally,the change process,scope and change characteristics of change are summarized.,and select the concentrated explosive activities passenger flow for further prediction research.(2)The feature construction and extraction process under the machine learning framework of large-scale activities passenger flow prediction are given.Firstly,the framework of urban rail transit passenger flow prediction under large-scale activities based on machine learning is established.The features are constructed from two aspects:conventional features and activity features.Secondly,the specific construction basis and method of the conventional features and activity features are expounded,and innovatively puts forward the construction method of activity features.Finally,the Lasso algorithm is adopted to select the features of the constructed full feature set,so as to obtain the input features of the constructed model..(3)The RF-GBRT passenger flow combined prediction model combined with gradient boosting regression tree(GBRT)and random forest(RF)is constructed.Firstly,it analyzes the difficulties of urban rail transit passenger flow prediction under large-scale activities,and proposes solutions based on error analysis.Secondly,the gradient boosting regression tree(GBRT)model and random forest(RF)model in ensemble learning are used as the prediction basis model.The improved particle swarm optimization(PSO)algorithm was used to optimize the parameters,and the prediction models were combined to obtain the RF-GBRT combined prediction model suitable for large-scale activities.And the cross-validation method and evaluation index for model verification and evaluation were determined.(4)A case study of the Beijing Subway DSST Station is carried out.The AFC data during DSST ball games period were selected and verified experimentally by Python tools.The prediction results under different feature sets are compared to prove the validity of the feature construction and extraction methods proposed in this paper.Compared with the common passenger flow prediction methods such as KNN,SVR,LR and so on,the proposed RF-GBRT model prediction method has better predictive effect.