High-Speed Railway Station Safety Passenger Flow Prediction by GMM Clustering and CNN-BILSTM

Aiming at the existing high-speed railway station security check passenger flow prediction research mostly for the prediction under normal conditions, without considering the change trend of security check passenger flow under abnormal and unexpected circumstances, a convolutional neural network-BiLSTM high-speed railway station security check passenger flow prediction model based on Gaussian mixture model clustering with fusion attention mechanism is proposed. First, the original data set is clustered using date features and delay features using GMM clustering algorithm, and different CNN-BiLSTM passenger flow prediction models are constructed based on different daily security check passenger flow scenarios obtained from the clustering. The experimental results show that compared with the existing prediction methods, this method can accurately predict the security check passenger flow in each time period under different scenarios.