基于LSSVM和马尔科夫链模型的公路客运量预测  

Highway Passenger Volume Forecast Based on Least Squares Support Vector Machine and Markov Chain Model

作　　者：程丽娟 张仲荣[1] 

机构地区：[1]兰州交通大学数理学院,甘肃兰州730070

出　　处：《洛阳理工学院学报（自然科学版）》2017年第4期70-75,共6页Journal of Luoyang Institute of Science and Technology：Natural Science Edition

基　　金：甘肃省自然科学基金项目(1610RJZA057)

摘　　要：通过分析常用的客运量预测方法,提出了基于LSSVM和马尔科夫模型的组合预测模型。以2004年~2013年兰州市公路客运量的实际值为基础,通过LSSVM对客运量的预测,得到了2014和2015年的公路客运量的预测值,结合实际客运量计算出其预测结果的相对误差。对相对误差进行划分状态区间,运用马尔科夫模型对预测结果进行修正,进而得到高精度的客运量预测值。最后将所得结果与应用单一的LSSVM预测方法及时间序列方法预测所得结果进行对比。分析结果表明,基于LSSVM和马尔科夫链模型的组合预测模型预测精度较高,满足实际需求。In this paper, a combined forecasting model based on least squares support vector machine and Markov model is proposed by the analysis of the commonly used passenger volume forecasting methods. First of all, through the actual volume of highway passenger traffic in Lanzhou from 2004 to 2013, the values of highway passenger volume in 2014 and 2015 were predicted by least squares support vector machine and the relative error of the predicted results is calculated. Then, the relative error is divided into the state interval and revised by using the Markov model, so the high precision passenger volume forecast value is obtained. Finally, the results are com- pared with the results predicted by the method of a single LSSVM prediction and time series. The results show that the prediction accu- racy of the combined forecasting model based on LSSVM and Markov chain model is high, which can meet the actual demand.

关 键 词：LSSVM 马尔科夫模型 组合预测 公路客运量 

分 类 号：F570.82[经济管理—产业经济]