基于误差融合模型的地铁车站客流监测方法研究

Research on Passenger Flow Monitoring Method for Metro Stations Based on Error Fusion Model

针对城市轨道交通拥堵问题和智慧地铁发展要求,提出了一种基于误差融合的城市轨道交通短时客流预测模型。采用灰色关联分析方法对客流关键影响因素进行识别,基于线性回归模型对车站客流进行预测,并进一步利用神经网络模型对预测误差进行修正,以广州地铁十八号线某站的进、出站客流数据为例进行案例研究。研究结果表明:与传统统计模型及单一的神经网络模型对比,基于误差融合的短时客流预测模型均方根误差分别降低了38.0%与29.6%,平均绝对误差分别降低了46.4%与35.1%,证明了该模型在短时客流预测方面的准确性、可靠性,为地铁车站客流监控提供了技术支撑。

In view of the traffic congestion problem of urban rail transit and the development requirements of the intelligent metro, a short-term passenger flow prediction model based on error fusion was proposed for urban rail transit.Firstly, the key influencing factors of passenger flow were identified by the grey correlation analysis method. Secondly,the station passenger flow was predicted on the basis of the linear regression model, and the neural network model was used to correct the prediction error. Finally, the inbound and outbound passenger flow data of a station on Line 18 of Guangzhou was taken as an example for analysis. The results show that compared with the traditional statistical model and the single neural network model, the short-term passenger flow prediction model based on error fusion reduces the root mean square error by 38.0% and 29.6%, respectively, and the average absolute error by 46.4% and 35.1%,respectively. This proves the accuracy and reliability of the proposed model in short-term passenger flow prediction,which provides technical support for passenger flow monitoring of metro stations.