基于遗传算法优化支持向量回归机的区域物流需求预测  

Logistics Volume Forecast Model of Support Vector Regression Optimized by Genetic Algorithm

作　　者：曹志强[1] 杨筝[2] 刘放 CAO Zhi-qiang;YANG Zheng;LIU Fang(Guangxi Normal University for Nationalities,Chongzuo 532200,China;Economics and Management School,Wuhan University,Wuhan 430072,China;State-owned Assets Supervision and Administration Office of the Wuchang District,Wuhan 430062,China)

机构地区：[1]广西民族师范学院,广西崇左532200 [2]武汉大学经济与管理学院,湖北武汉430072 [3]武汉市武昌区国有资产监督管理办公室,湖北武汉430062

出　　处：《系统科学学报》2018年第4期79-82,90,共5页Chinese Journal of Systems Science

基　　金：广西民族师范学院服务旅游与商贸产业专业群科研项目(LYYSMKY2016003);广西民族师范学院校级科研项目(2014YB007);广西职业教育教学改革工程项目(GXGZJG2016B061)

摘　　要：本文采用支持向量回归机对物流需求进行预测,研究SVR在物流需求预测中的可行性。为了建立有效的SVR模型,SVR参数需要精心设置,针对物流需求非线性的特点,本文首次提出用遗传算法(GA)优化支持向量回归(SVR)机进行预测,采用最优参数构造SVR模型。以1990-2015年广西货运量为时间序列数据,实验结果表明,以平均绝对误差(MAPE)和均方差比值(MSE)为精度检验标准,GA-SVR比ARIMA和灰色预测具有更优的预测效果。This study applies a novel network technique and support vector regression(SVR) to forecast logistics volume. The aim of this study is to examine the feasibility of SVR in logistics volume forecasting. In order to establish an efficient SVR model,the parameters should be set elaborately. The nonlinear characteristics of logistics volume,support vector regression( SVR) machine optimized by genetic algorithm( GA) are firstly proposed for prediction,and then the optimal parameters are adopted to construct the SVR model. Taking Guangxi freight volume in 1990-2015 years as time series data,the results indicate that GA-SVR is more outstanding than the ARIMA model and grey forecasting method based on the mean absolute percentage error( MAPE) and mean square error( MSE).

关 键 词：遗传算法 支持向量回归机 物流需求 预测 

分 类 号：N949[自然科学总论—系统科学]