城市轨道交通客流短时预测方法研究

Research on short-term prediction method of urban rail transit passenger flow

对轨道交通进行客流预测,不仅是进行轨道交通运营的基础,同时也决定了轨道交通的运营效率。为了提高轨道交通的短时预测效果,提出了LM算法和GA遗传算法的优化思路,对BP神经网络短时预测模型进行优化。此外,为了验证优化模型的有效性,研究S市两个站点的客流实测数据进行仿真实验,并以MAD、MAPE、MSE 3个指标评价模型的预测性能。研究结果显示,在优化后的GA-LM-BP算法模型中,它的预测值相对误差控制在±20%以内,且评价指标所反映的预测精度和稳定性较好。此次研究所提出的优化思路显著提高了预测模型精度和稳定性,优化后的预测模型具有良好的适用性。希望这次研究成果可以为提高客流短时预测效率提供一些参考,同时为提高轨道交通运营效率提供一些思路。

Passenger flow forecast of rail transit is not only the basis of rail transit operation,but also determines the operation efficiency of rail transit.In order to improve the short-term prediction effect of rail transit,the optimization ideas of LM algorithm and GA genetic algorithm are proposed to optimize the short-term prediction model of BP neural network.In addition,in order to verify the effectiveness of the optimization model,the simulation experiment is carried out with the measured passenger flow data of two stations in S City,and the prediction performance of the model is evaluated by the three indexes of MAD,MAPE and MSE.The results show that in the optimized GA-LM-BP algorithm model,the relative error of the prediction value is controlled within±20%,and the prediction accuracy and stability reflected by the evaluation index are better.The optimization ideas proposed in this study significantly improve the accuracy and stability of the prediction model,and the optimized prediction model has good applicability.It is hoped that the research results can provide some reference for improving the efficiency of short-term passenger flow forecasting,and provide some ideas for improving the operation efficiency of rail transit.