摘要
为改进传统出行生成预测以单个出行为分析单元、难以反映个体出行行为的不足,建立了基于活动模式的工作者出行生成预测模型.采用相关性分析和多项logit方法研究了1617个有效样本的居民个体属性、家庭属性、区位属性、出行属性、活动属性所引起的活动和出行生成模式的差异.通过建立包含显著变量的6种典型活动模式效用函数,获得个体选择概率.按概率集计法分析各交通小区各类活动模式总量,结合活动模式中包含的出行链平均长度,进行工作者出行生成量预测.基于活动模式的工作者出行生成预测模型有助于以出行链整体为单元考察多次出行之间的关联性,增强出行生成模型与行为模型的融合.
To overcome the deficiency of traditional trip generation forecasting method, which is based on separate trip and short of individual travel behavior, commuters＇ trip generation model based on activity patterns is investigated. Correlation analysis and the multinomial logit method are applied to 1 617 samples to examine how individuals＇ activity and travel patterns are influenced by individual/household socio-demographics, locational factors, and travel and activity characteristics. By calibrating utility functions of six typical activity patterns including the significant variables, the individual choice probabilities can be analyzed. After aggregating the total number of each activity pattern in each zone, combined with the average length of each trip chain, commuters＇ trip generation is forecasted. The activity pattern-based commuters＇ trip generation model can not only improve the connection of single trips based on trip chain unit, but also help to contain individual＇s behavioral aspects.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2008年第3期525-530,共6页
Journal of Southeast University：Natural Science Edition
基金
国家重点基础研究发展计划(973计划)资助项目(2006CB705501)
国家自然科学基金重点资助项目(50738001)
关键词
活动模式
出行生成
MNL模型
出行链长度
集计
activity pattern
trip generation
multinomial logit model
trip chain length
aggregate
