Predicting Urban Rail Traffic Passenger Flow Based on LSTM

Abstract

Accuracy is important in Urban Mass Transit System (UMTS), especially for traffic control. Traffic passenger flow prediction has a long history and is still a difficult problem due to highly nonlinear and stochastic characteristics of transit systems. The existing models such as Auto Regressive Integrated Moving Average (ARIMA) is mainly linear models and cannot describe the stochastic and nonlinear nature of traffic passenger flow. In recent years, deep-learning-based methods have been applied as novel alternatives for traffic passenger flow prediction. However, which kind of deep neural networks is the most appropriate model for traffic passenger flow prediction remains unsolved. In this paper, we use Long Short Term Memory (LSTM) and hierarchical cluster method predict short-term traffic passenger flow, and experiments demonstrate that Recurrent Neural Network (RNN) based on LSTM perform better than ARIMA, Support Vector Machines (SVM) and Extreme gradient boosting (Xgb).