Study on the Prediction Model of Short-term Bus Passenger Flow Based on Big Data

Abstract

Prediction of short-term bus passenger flow can help bus managers timely and accurately get the changes of the passenger flow and make scientific and reasonable vehicle scheduling to meet passengers' needs. In this paper, a SLMBP model is constructed to predict the bus passenger flow. The SRCC(Spearman rank correlation coefficient) method is used to determine the factors that have significant influence on passenger flow changes. The Levenberg-Marquardt algorithm is used to optimize the BP neural network to avoid getting stuck in local optimal solutions and prompt the convergence speed. A SLMBP neural network parallel algorithm is constructed to perform multiple stations prediction. The experimental results show that the SLMBP neural network parallel algorithm can not only guarantee the accuracy of short-term passenger flow prediction, but reduce the time spent on model learning and prompt the prediction speed.