The seat structure of passenger flow is the specific performance of passenger travel demand characteristics under certain transport supply conditions. In this paper, the influence of income level, transit time, and the relationship between supply and demand on passenger flow structure is considered as independent variables, and the proportion of seats at different classes is a dependent variable. The long and short-term memory neural network model (LSTM) based on deep learning is constructed, and the simulated annealing algorithm (SA) is used to optimize the hyperparameters of the model, forming an SA-LSTM model for passenger flow structure prediction of high-speed railway. Using passenger flow data of the Yangtze River Delta, the prediction accuracy of this model reaches 97.36%. Compared with other models, the proposed model has better prediction accuracy and generalization ability.

High-speed rail;passenger flow seat structure;LSTM;simulated annealing algorithm