基于SVM与人工神经网络组合模型的物流规划车辆行程时间预测  

Vehicle travel time prediction in logistics planning based on SVM and NN model

作　　者：叶创鑫[1] 谭满春[1] 

机构地区：[1]暨南大学数学系,广东广州510632

出　　处：《暨南大学学报（自然科学与医学版）》2010年第5期451-456,共6页Journal of Jinan University(Natural Science & Medicine Edition)

基　　金：广东省自然科学基金项目(9151001003000005)

摘　　要：行程时间预测在交通和物流规划中具有重要的作用.传统方法大部分是针对单一路段的行程时间进行短期预测.考虑了一个与传统行程时间预测不同的背景,研究物流车队行程时间预测问题.结合人工神经网络的学习能力和支持向量基对复杂非线性预测的处理能力,提出了一种基于支持向量基和人工神经网络相结合的方法,对物流规划中车辆行程时间进行有效的预测.把得到的结果和真实值比较,说明所提出的预测方法是可行和有效的.Travel time prediction plays an important role in transportation and logistics planning. Most of the work done so far is aiming at the development of the short-term prediction methods for the single road of the transportation systems. In this paper the travel time prediction of vehicle fleets in logistic com- panies is considered, which is a different scenario from the conventional transportation sector~ one. Com- bining the good learning ability of neural network （NN） and the strong nonlinear forecasting ability of support vector machine （ SVM）, a new approach is proposed for the time prediction of the long-range trips in a logistics fleets, based on the NN and SVM. The comparisons between the prediction values and the real values prove that the proposed approach is feasible and effective.

关 键 词：支持向量基 神经网络 行程时间预测 物流规划 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] U491.14[自动化与计算机技术—控制科学与工程]