基于大数据的轨道交通短时客流预测研究

Research on short-term passenger fl ow prediction of rail transit based on big data

轨道交通以其高效、快速、准时的优势,得到了城市内旅客出行的青睐。轨道交通站内及线上的客流变化情况直接影响了轨道交通运营的安全性和高效性,尤其是针对特殊情况客流产生巨大增长的情况,因此,开展短时客流预测研究具有重要意义。大数据技术为短时客流预测研究带来了便利和挑战。本文立足于大数据时代,对近年来城市轨道交通短时客流预测采用的模型、方法以及对时间粒度的选择进行了综述,研究主要面向普通机器学习算法和深度学习网络开展。结果表明,具有时空特性的长短时记忆模型(LSTM)是客流预测领域一种应用广泛且效果优异的模型。此外,时间粒度的选择与预测的对象、模型和方法有关,15和30min的时间粒度适合于大多数情况。

Rail transit has been favored by passengers in the city for its advantages of high efficiency,speed and punctuality.The change of passenger flow in and on the rail transit station directly affects the safety and efficiency of rail transit operation,especially for the huge growth of passenger flow in special cases.Therefore,it is of great significance to carry out short-term passenger flow prediction research.Big data technology brings convenience and challenges to the research of short-term passenger flow forecasting.Based on the era of big data,this paper summarizes the models,methods and the choice of time granularity used in short-term passenger flow prediction of urban rail transit in recent years.The research is mainly aimed at ordinary machine learning algorithms and deep learning networks.The results show that the long short-term memory model(LSTM)with spatial and temporal characteristics is a widely used and excellent model in the field of passenger flow prediction.In addition,the choice of time granularity is related to the predicted object,model and method,and the time granularity of 15 and 30 min is suitable for most cases.