融合等维新息的灰色马尔科夫模型的地铁中长期客流量预测

Medium and Long-Term Subway Passenger Flow Prediction Based on a Grey Markov Model with Integrated Equi-Dimensional Information

摘要: 地铁中长期客流量预测在单一模型中很难同时满足数据的稳定性、周期性等特征导致其预测结果较差。针对中长期客流预测精度较低的问题，本文使用融合等维新息的灰色模型与马尔科夫模型组合的地铁客流预测模型。首先对原始数据序列进行预处理，其次建立灰色GM(1, 1)模型并融合等维新息的思想来提高中长期预测精度。然后将马尔科夫链纳入融合等维新息的灰色模型来修正残差。最后，选用2013~2019年上海地铁日均客流量进行预测，结果表明融合等维新息的灰色马尔科夫模型在地铁中长期客流量预测精度为I级(优)，高于单一模型。

Abstract: Medium and long-term subway passenger flow prediction in a single model is difficult to simultaneously satisfy the stability of data, periodicity and other characteristics leading to its poor prediction results. In response to the issue of low accuracy in medium and long-term subway passenger flow prediction, a grey Markov model with integrated equi-dimensional information is proposed for prediction. Firstly, the original data sequence is preprocessed; then, followed by the establishment of a grey GM(1,1) model which is improved using the equi-dimensional information concept; next, the Markov chain is incorporated into the grey model with integrated equidimensional information to correct the residual. The proposed method is experimented using daily passenger flow data from Shanghai subway between 2013 and 2019. The results show that the implementability and advantages of the grey Markov model incorporate equi-dimensional new information in the application of metro passenger flow prediction in the medium and long term.