Research On Short-term Passenger Flow Prediction Of Urban Rail Transit Based On TCN-LSTM Combined Model

Fine-grained short-term passenger flow prediction can provide key data support for the formulation of service plan and precise control of passenger flow of urban rail transit.However,current domestic research only considers the subway passenger flow as a single variable,without considering the impact of external factors such as weather and holidays on passenger flow,resulting in low prediction accuracy.Moreover,current research mainly focuses on the prediction of station entrance flow,and there is an urgent need for more fine-grained passenger flow indicators,such as entrance flow,Original-Destination flow transfer flow,interval and train flow.Temporal Convolutional Network(TCN)has demonstrated strong predictive performance in time-consuming and accuracy when applied to multiple sets of time series datasets.Long Short-Term Memory(LSTM)can extract the long and short-term dependencies between passenger flow and external factors such as holidays and weather.The combination of TCN-LSTM model can effectively improve the accuracy of short-term passenger flow prediction.Therefore,a study on the urban rail transit short-term passenger flow prediction method based on the TCN-LSTM combination model has important engineering application value.Based on the analysis of urban rail transit passenger flow prediction algorithms and their research status,the TCN-LSTM short-term entrance passenger flow prediction model based on multi-dimensional predictable features is proposed in this paper.,which improves the prediction accuracy.To achieve fine-grained state prediction of predicted entrance passenger flow,such as passenger travel path and train frequency,a passenger flow spatiotemporal distribution prediction model based on the PCA-Kmeans method and dynamic trains is proposed.Based on the PCA-Kmeans method,the starting and ending point OD passenger flow prediction of urban rail transit is realized,the K-shortest path set calculation is improved based on the A* algorithm,and the passenger flow allocation model based on dynamic train crowding is used to predict passenger flow routing path and train frequency.On the basis of research on short-term passenger flow prediction algorithms,the overall design of the urban rail transit short-term passenger flow prediction algorithm based on the TCN-LSTM combination model is presented,which includes six parts: algorithm data interface,data acquisition module,entrance passenger flow prediction module,OD passenger flow prediction module,passenger flow dynamic allocation module,and prediction data statistics module.Finally,the short-term passenger flow prediction method for urban rail transit was experimentally verified using a city’s subway system in Southwest China as an example.Experimental results show that the urban rail transit short-term passenger flow prediction method based on the TCN-LSTM combination model proposed in this paper can achieve short-term prediction of multidimensional passenger flow indicators with high accuracy.It can provide fine-grained predictive data support for urban rail transit passenger flow control and transportation organization planning,and provides a new approach to realizing short-term passenger flow prediction and operation auxiliary decision-making data support for urban rail transit.