Study On Methods Of Extracting Resident Trip Characteristics Based On Cellphone Location Data

To collect residents travel information by the use of mobile phone positioning technology has advantages of real-time, low cost, a large number of samples, and being easy to implement, and complemented with the traditional travel survey, it becomes another important data source of travel characteristics research. However, the mobile phone positioning data time distribution is not uniform, and the positioning accuracy is uncertain, the influence factors are complicated, the current trip feature extraction methods have some deficiencies. Based on the large scale real mobile phone positioning signaling data, from the point of view of the spatial and temporal characteristics of the data, the study tries to establish a more accurate and more feasible extraction method of travel characteristics, and it will have the important significance of theory and engineering practice.Firstly, in order to have a comprehensive understanding of the occurrence mechanism of positioning data, the cellular communication network and the basic principles of mobile phone positioning technology is introduced, and the in-depth of the connotation of mobile phone signaling data is interpreted deeply. On this basis, combined with the general travel definition and the characteristics of the mobile phone positioning data, the study defines a "mobile travel", and further illustrates the applicability of the extraction travel characteristics by the use of mobile phone positioning technology to pave the way for the follow-up study.Secondly, data processing and feature analysis. The study analyses deeply positioning data generation mechanism, and draws lessons from existing research results, builds a more perfect multi-level data processing methods, including a variety of invalid data filtering, a variety of noise data recognition processing, and provides high quality data sources for travel feature extraction. Then, from multiple dimensions of event type, time, distance, average speed, it analyses data characteristics to provide data for the establishment of travel feature extraction model.Thirdly, the establishment of travel feature extraction model. On the basis of the above research and in-depth analysis of time and space characteristics of mobile phone positioning trajectory point, the study finds that mobile phone users travel trajectory point is mainly composed of the circular stay area which represents stay and the long travel area which represents travel. Based on this, the research proposes a spatial-temporal clustering algorithm for identification of circular region, and the trajectory point is divided into staying and moving point, at the same time to obtain residence duration, position and other travel information. Based on the above information, it establishes a variety of calculation methods for travel characteristics index including the travel time, travel distance, travel speed, stay area. Finally, from the multiple perspectives the research analyses sampling expansion influence from the mobile phone travel to users travel and builds multi-layer sampling expansion methods.Fourthly, verifying model by experimental analysis. First of all, based on the definition of "mobile phone travel", each parameter in the model is analyzed and calibrated. Then, the study selects a large scale mobile phone positioning data of a working day in Shanghai city, and contrasts the model analysis results and the fourth Shanghai city travel survey data. It finds that the multiple travel feature indexes are highly related with the residents travel characteristics, and the r value of CORREL inspection is above 0.92. It suggests the travel feature extraction method that the study puts forward has good reliability and applicability.Finally, the study summarizes the shortcoming of the research method, and puts forward the prospects for future research.