基于支持向量机的通勤日活动-出行持续时间预测  

Prediction of commuter's daily activity-travel duration time with support vector regression

作　　者：宗芳[1] 王占中[1] 贾洪飞[1] 焦玉玲[1] 吴杨[1] 

机构地区：[1]吉林大学交通学院,长春130022

出　　处：《吉林大学学报（工学版）》2016年第2期406-411,共6页Journal of Jilin University:Engineering and Technology Edition

基　　金：国家自然科学基金项目(50908099;51278221);教育部人文社科青年基金项目(14YJC630225);吉林大学国家优秀青年科学基金后备人选培育计划项目

摘　　要：构建了通勤者的日活动-出行链,并给出其时间构成要素,对比分析了Hazard和回归支持向量机模型的预测精度,应用回归支持向量机模型建立了出行耗时预测模型和活动持续时间预测模型,预测了通勤日活动-出行链中各次出行的出行耗时和各次活动的持续时间,并应用已建模型考察公交优先策略对出行耗时的影响。研究表明:本文所建模型能够以较高精度预测通勤者的日出行耗时和活动持续时间,在出行耗时预测中回归支持向量机的预测精度高于Hazard模型,相关影响因素分析也有助于掌握居民的通勤出行时间决策规律。研究成果可以为构建通勤日活动-出行行为预测模型系统提供连续时间预测模块,也可以为交通需求管理政策的制定和实施提供决策依据。The commuter＇s daily activity-travel schedule is proposed and its key time allocation is investigated.According to the comparison of hazard model and Support Vector Regression（SVR）model,travel time prediction models and activity duration prediction models are developed by employing SVR.Then,the continuous time allocation,i.e.all the travel times and activity durations in daily activity-travel schedule are derived and the transit priority policy is evaluated using these models.The results indicate that the model system has a high level of prediction accuracy,and the goodness-of-fit of SVR models is higher than that of Hazard models.This study provides useful insights into commuter＇s activity-travel time allocation decision.It also serves a foundation that future models of full-scale daily activity-travel pattern can be built on.Moreover,it provides potential for transportation demand management policy analysis.

关 键 词：交通运输工程 通勤出行时间 回归支持向量机 

分 类 号：U491.1[交通运输工程—交通运输规划与管理]