基于UBGPM-Markov的铁路货运量预测方法

针对铁路货运量发展变化的非线性特性,采用非线性灰色模型中的无偏GM（1,1）幂模型进行预测,并用状态划分更为精细后的马尔可夫链修正预测值,从而建立优化后的UBGPM-Markov模型.通过对我国2000~2012年铁路货运量实例分析,与改进GM（1,1）模型、无偏GM（1,1）模型2种预测方法进行了比较,比较显示无偏GM（1,1）幂模型具有较高的预测精度.优化后的UBGPM-Markov模型更是显著提高了预测精度,将平均绝对百分误差（MAPE）由2.11%减少到0.55%.

Method Study of Railway Freight Volume Prediction Based on UBGPM-Markov Model

Aiming at non-liner characteristics of development and changes for railway freight volumes,the future development of railway freight volumes is forecasted by the unbiased GM（ 1,1） power model,and the forecast accuracy is modified to a great extent by the optimized Markov Chain method,which divides states more accurately for establishing the optimized UBGPM-Markov model. Through analyzing and predicting railway freight volumes in China between 2000- 2012,the prediction results are compared with the improved GM（ 1,1） and the unbiased GM（ 1,1） models. The conclusion shows that the prediction effects of the unbiased GM（ 1,1） power model are superior to the other two models,and the optimized UBGPM-Markov model significantly improves the prediction accuracy,reducing the average relative prediction error from 2. 11% to 0. 55%.