基于灰色关联分析的LS-SVM铁路货运量预测

摘要

On the basis of analyzing the influencing factors of railway freight volumes, the LS-SVM railway freight volume forecast method with grey correlation analysis was proposed to improve the predicting accuracy and modeling speed of railway freight volumes. The influencing factors of railway freight volumes were divided into social demand factors and railway supply factors. Correlations between the two-category factors and railway freight volumes were analyzed respectively by grey correlation analysis. The input variables of LS-SVM were screened by the grey correlation degree value together with qualitative analysis to simplify the LS-SVM structure. Finally, the stochastic inertia weight PSO (SIWPSO) algorithm was used to optimize the parameters of the LS-SVM model. Statistics of the railway freight volumes from 1980 to 2009 indicate that the proposed forecast method provides a better convergence rate and higher predicting accuracy.%为提高对铁路货运量的预测精度及建模速度,在分析货运量影响因素基础上,提出基于灰色关联分析的LS-SVM铁路货运量预测方法.将货运量影响因素分为社会需求与铁路供给两方面因素,采用灰色关联分析法对两方面因素与货运量进行相关性分析,根据灰色关联度值,结合定性分析筛选LS-SVM输入变量,简化LS-SVM结构,再通过随机权重粒子群(SIWPSO)算法优化选择LS-SVM模型参数.通过对我国1980～2009年铁路货运量实例分析表明:该方法具有较快的收敛速度和较高的预测精度.