Research On Travel Model Split Based On Activity Patten Classification

Activity theory has certain advantages in explaining urban traffic demand, because of traffic demand rooting from the production and living activities. Recently, it has been become the focus that the traffic demand analysis method based activity theory in the field of urban traffic planning and management in China. For the traditional mode split methods, they always pay attention to resident single trip, which takes no account of personal activity patterns and the links between the different trips in one day. This dissertation will do some works on activity patterns and mode split, seek after the combination between activity theory and traditional modes split methods.Activity patterns split principles and methods were put forward. In this dissertation, activity patterns was divided into eight categories based on the mainly activity type, whether it's suspended or not, whether there was work sub-trip or not, and whether there was subordinate trip or not. According to the existing residents'trips survey data, a pattern data extraction procedure was designed. Work activity patterns were studied in depth from analysis of the contingency table. Then, the relationships between activity patterns and personal attribute, family attribute, and etc were analyzed, the main influencing factors were selected, and the MNL model of the activity pattern probability choice was set up.The trip data were statistically analyzed with the various patterns, it's clear that trip mode choices and personal attributes were different. Through contingency table analysis, it indicated that trip mode choices and personal attributes had strong correlation. According selected characteristic variables, established discrete probability models of model split choice respectively and used the mixed logit model attemptly, then recurred for the simulation algorithm to solve. We got the integration of the activity theory and mode split, which based on the joint model between activity patterns and mode split using probability theory.The models for traffic demand forecasting setting up in this dissertation can be used to forecast traffic demand, improve forecasting accuracy. Also for traffic management planning with a more accurate data and for the useful improvement of traditional demand forecasting methods.