基于长短期记忆网络的轨道交通短期OD客流量预测

Forecast of short-term origin-destination passenger flow of rail transit based on long short-term memory network

为了更准确地掌握轨道交通客流在线网中的时空分布,更高效地匹配客流需求与运输能力,实现提高轨道交通运输效率、改善运营服务质量的目的,提出了一种基于长短期记忆网络的短期OD(交通起止点)客流量预测方法。以历史客流数据为基础,定性分析车站间OD客流量的时空相关性,利用回归分析法定量分析客流影响因素,筛选出运营时刻、运营日特征、最低气温3个时间特征。为提高预测精度,以长短期记忆网络为基础,结合时间特征,为每对起讫点单独构建预测模型,形成了基于长短期记忆网络的轨道交通短期OD客流量预测模型。以苏州市为例进行验证,结果表明,加入了时间特征的短期OD客流量预测模型较移动平均模型、仅利用历史客流数据训练的基于LSTM网络的短期OD客流量预测模型,预测结果与真实值之间的误差降低了6.27%~8.58%,所提出的方法和模型可为轨道交通运营部门制定列车运行计划、组织客运工作提供更准确的数据资料。

In order to grasp the temporal and spatial distribution of passenger flow in the rail transit more accurately,match the demand of passenger flow and the capacity of rail transportation more efficiently,and achieve the purpose of improving rail transit transportation efficiency and operation service quality,a forecast method of short-term OD passenger flow(origin-destination)based on long short-term memory network was proposed.The temporal and spatial correlation of OD passenger flow between stations was qualitatively analyzed based on the historical passenger flow data.The influencing factors of passenger flow were analyzed by regression analysis method,and three time characteristics of operation time,operation day characteristics and minimum temperature were selected.Based on the long short-term memory network,the model was built for each OD combining with the time features to improve the forecast accuracy.Therefore,the short-term OD passenger flow forecast of rail transit based on long short-term memory network was formed.Suzhou railway transit was taken as an example to make an imitation test.The results show that compared with the moving average model and the short-term OD prediction model based on LSTM network trained only by historical data,the error between the real value and the predicted result of the short-term OD prediction model with the time characteristics is reduced by 6.27%~8.58%.In short,the proposed model and the methods can provide more accurate data for rail transit operation department to make the train operation plan and organize passenger transport.