Study on the Efficiency of Urban Railway Transportation Based on the Hybrid Model of CNN and BiLSTM

This study aims to construct a hybrid model that incorporates a convolutional neural network (CNN) and a bidirectionallong-and-short-term memory network (BiLSTM), and cleverly incorporates a self-attention mechanism to strengthen the model's ability to capture features of the spatiotemporal data, to improve the accuracy of short-term passenger flow prediction in urban rail transit systems to meet the operational challenges of the peak passenger flow. Firstly, CNNs are introduced into the hybrid model in this paper to extract spatial features from passenger flow data, while the self-attention mechanism strengthens the model's ability to identify key features and ensures that the model is able to understand changes in passenger flow at different times and locations. Secondly, the inclusion of BiLSTM enables the model to better understand passenger flow dynamics in the time dimension, and the bidirectional structure ensures that the model captures the impact of both past and future information on the current prediction. Finally, in this study, the dynamic time warping (DTW) method is used to cluster the time series of stations to identify clusters of stations with similar passenger flow patterns, which improves the model's ability to generalise and predict the flow trends of various types of stations. At the same time, considering the impact of external factors such as rainfall on passenger flow, the study shows that the inclusion of these data on a larger time scale significantly improves the accuracy of the predictions.