动态公交车辆运行时间预测模型详解.pdfVIP

25 3 V o.l 25 N o. 3 20 10 6 JOURNAL OF SYSTEM S ENG IN EER ING Jun. 2010 1 1 2 姚宝珍 , 杨成永 , 于 滨 ( 1. , 100044; 2. , 116026 : 准确预测公交车运行时间是先进的出行者信息系统 ( AT IS 的核心. 本文应用支持向量 ( SVM 进行公 交车辆的运行时间预测, 其目的是要验证 SVM 在运行时间预测领域的可行性. 为了调整不同阶段历史数据对预 测结果的影响引入了衰减因子, 并应用 一种自适应算法来动态调整预测误差. 然后以大连市 23路公交车对该模 型进行来了检验. 结果显示, 带有衰减因子和自适应算法的支持向量 算法具有很好的预测精度和动态性能. : ; ; ; ; : U 12 1 : A : 1000- 57 81( 2010 03- 0 365- 06 Dynamic bus traveltime predictionmodel 1 1 2 YAO Baozhen, YANG Chengyong, YU Bin ( 1. School of C iv il Eng ineering A rchitecture, Beijing J iaotong Un iversity, Beijing 100044, China; 2. Schoo l of T ransporta ition M anagem ent, D alian M arit mi e U niversity, Da lian 116026, China Abstract: E ffect ive pred ict ion o f bu s arriva l tmi e is a core for advanced trave ler inform ation system. Support vecto rm ach ines ( SVM are applied to pred icting bus travel tmi es. T he ob ject ive of th is pa per is to exam ine the feasib ility and applicability o f SVM in the veh icle travel tmi e forecasting area. A decay factor is introduced to ad ju st the w eights bewt een new and o ld data. A lso, an adaptive a lgo rithm is u sed to mi prove the predict ion results. The SVM w ith the decay factor is tested w ith the data o f the num ber 23 bu s route in Dalian c ity. R esults show that the SVM w ith the decay factor and the ad