LSTM候机楼人流量预测的研究

Research on Airport Pedestrian Flow Prediction with a Long Short-Term Memory Network

人流量预测是目前智能交通的研究方向之一。精准预测人流量的趋势和峰值,能够合理调度警力和人力资源,并对由于人群众多引起的踩踏事件作出提前预警。随着各传感器的大量部署,交通系统已拥有大量可用数据,但是缺乏行之有效的分析方法。为此,本文通过深度学习的方式对人流量预测建立模型,提出一种基于LSTM神经网络的人流量预测模型,能够根据已知数据预测未来一段时间内人流量的趋势和峰值。与ARRIMA模型比较验证,LSTM网络模型具有更好的性能。

Pedestrian flow prediction is one of the research direction in intelligent transportation.Not only can we schedule the police and human,but also make early warning for stampede if we can accurately predict the trend and peak of pedestrian flow.With large numbers of sensors deployed,traffic system has already had a large number of available data,but hasn’t effective analysis method to process those data.This paper proposes a long short-term memory network(LSTM)-based approach to predict pedestrian flow.Based on known data,the proposed approach can accurately predict the trend and peak of pedestrian flow for the foreseeable future.It also has better prediction accuracy than the autoregressive integrated moving average model.