深度学习算法在城市轨道交通客流短时预测中的运用分析

Application Analysis of Deep Learning Algorithm in Short-Term Prediction of Urban Rail Transit Passenger Flow

何姜姜 ¹张鹏 ¹王芳玲¹

扫码查看

作者信息

1. 广州铁路职业技术学院广东广州 511300
折叠

摘要

为准确把握轨道交通短时客流量的变化情况,在深度学习算法基础上,基于长短期记忆网络(Long Short-Term Memory Network,LSTM)和北方巷鹰优化算法-变态模态分解-长短期记忆网络(Northern Goshawk Optimization-Variational Mode Decomposition-Long Short-Term Memory Network,NGO-VMD-LSTM)联合预测模型,提出了短时客流量预测的方法.以某轨道交通换乘站和邻近小区的居住站为对象,以30 min为时间段,对客流时序数据进行训练,预测一段时间内的客流量变化趋势.根据研究结果可知,NGO-VMD-LSTM模型能够充分提取客流波动特征,可提高高峰时段的车站管理效率,为轨道交通运营部门的车辆调度、乘客管理等提供参考依据.

Abstract

In order to accurately grasp the change of short-term rail transit passenger flow,on the basis of Deep Learning Algorithm,this paper puts forward the short-time passenger flow prediction method based on the combined prediction model of Long Short-Term Memory Network(LSTM)and Northern Goshawk Optimization-Variational Mode Decomposition-Long Short-Term Memory Network(NGO-VMD-LSTM)respectively.With a rail transit transfer station and the neighboring residential stations as the object,it trains the passenger flow time series data for a period of 30 minutes,predicts the flow trend over a period of time.According to the research results,the NGO-VMD-LSTM model can fully extract the characteristics of passenger flow fluctuation,improve the efficiency of station management during peak hours,and provide a reference for the vehicle dispatching and passenger management of the rail transit operation department.

关键词

深度学习算法/轨道交通/客流短时预测/LSTM模型

Key words

Deep Learning Algorithm/Rail transit/Short-term passenger flow prediction/LSTM model

引用本文复制引用

出版年

2024

科技资讯

北京国际科技服务中心北京合作创新国际科技服务中心

科技资讯

影响因子：0.51

ISSN：1672-3791

段落导航