Dynamic OD Matrix Estimation Based On Cell Phone Location And Travel Survey

ABSTRACT:The OD (Origin-Destination) matrix can describe the traffic demand in a certain time period, which is the essential part in modern urban transportation research. With the development of information technology, it becomes possible to obtain traffic information by using cell phone location technology. However, the research of cell phone location information has been limited to the stage of theoretical modeling so far, and trip recognition technology is not fully mature. In this context, the research will analyze the regional dynamic OD data estimation by using cell phone location information and travel survey data.First, the research conducts the literature review about existing OD matrix estimation methods, analyzes the data collection method for OD estimation and summarizes the advantages and disadvantages of the existing methods. Then, the existing studies of cell phone location technology are summarized. After that, the research contents are determined.Secondly, for the cell phone location data and travel survey data, the research analyzes the characteristics of raw data and extracts the traffic OD information. For the cell phone location data, the research designs the methodology of obtaining the trip chaining by using cell phone location data, and identifies the OD point by using thresholds of speed, time and distance.Then, the triggering mechanism of cell phone location data is analyzed, and based on this, the location data is divided into two categories:active mode and passive mode. After that, the research analyzes the characteristics of time distribution for each mode and summarizes the advantages and disadvantages of both modes. The research designs different modification methods for each mode based on the travel survey data. For the active mode, the research modifies the data based on the time distribution of the OD’s amount, and then, adjusts the single OD pair respectively. For the passive mode, based on the data characteristics, the research designs the modification method by using the time distribution of travel survey data.Finally, a typical area is used to verify the regional dynamic OD data collection technology in the research. The raw data is obtained and modified by the designed modification method and travel survey data. For the all-mode data and passive data, the research verifies in different dimensions. For the travel amount, the error is reduced from30.08%to7.11%. For the relationships of production-attraction between zones, the correlation coefficient improves significantly. For the correlation coefficient between matrixes, the rise is not obvious, because of the nature difference between different data sources and the assumptions of the research.