摘要
为提高城际铁路车站短时客流量预测的准确性,本文设计了一种基于双向长短期记忆网络(Bi-LSTM)的预测模型.首先分析了广珠城际铁路车站日客流量的时间分布规律,发现各车站日客流量存在着相似的分布特征,但工作日、周末和节假日客流量呈现出较大差异.以广州南站、珠海站以及小榄站3个车站的进站客流为例,选择时间窗步长为2和4分别进行预测分析,通过调整模型参数来提高预测精度.当时间窗步长为4时,预测精度明显提高.与长短期记忆网络(LSTM)的预测结果对比,Bi-LSTM网络的预测精度更高,在广珠城际车站日客流预测中具有更好的适用性.
In order to improve the accuracy of short-term passenger flow prediction of intercity railway stations,a prediction model based on bidirectional long-term and short-term memory network(Bi-LSTM)is designed in this paper.First,the time distribution law of daily passenger flow of Guangzhou-Zhuhai intercity railway station is analyzed.It is found that the daily passenger flows of stations have similar distribution characteristics.However,there are great differences in the passenger flows during weekdays,weekends and holidays.Taking the inbound passenger flow of Guangzhou South Station,Zhuhai and Xiaolan stations as examples,the time window steps of 2 and 4 are selected for prediction and analysis respectively.The model parameters are adjusted in order to improve prediction accuracy.When the time window step is 4,the prediction accuracy is improved significantly.The prediction accuracy of Bi-LSTM is higher than those by the long-term and short-term memory network(LSTM).It has better applicability in the daily passenger flow forecasts for Guangzhou-Zhuhai intercity stations.
作者
吕秋霞
钟晓情
任雅思
Lü Qiu-xia;ZHONG Xiao-qing;REN Ya-si(School of Rail Transportation,Wuyi University,Jiangmen 529020,China)
出处
《五邑大学学报(自然科学版)》
CAS
2022年第1期50-56,共7页
Journal of Wuyi University(Natural Science Edition)
关键词
城际铁路
短期客流
客流预测
双向长短期记忆网络
Intercity railways
Short-term passenger flows
Passenger flow forecast
Bi-LSTM