A new approach to the prediction of passenger flow in a transit system

A non‐linear model is proposed for predicting the rate of passenger flow in a transit system, and its chaotic characteristic is observed. Using wavelets analysis, the passenger flow data for a whole day are decomposed in a multi‐scale way to obtain decomposition sequences. Subsequently, a neural network approach is used to predict the sequences. Finally the passenger flow value can be predicted when the predicted sequences are reconstructed. Results show that the present approach is a feasible method for passenger flow prediction.