Multi-source Data Analysis And Application Research On Urban Traffic Demand Analysis

Traffic demand analysis is the core of transportation planning,aiming to provide scientific and objective decision support by quantitative analysis.Whether traffic development rules can be accurately grasped depends on the understanding of the status of the relevant data and the rationality of the analysis method.With the in-depth development of information and accelerated iterative of data acquisition technology,the new data with large sample size,small scale,multi-dimension and rapid update greatly expand the traditional data set which is mainly composed of planning texts and traffic surveys.These new data provide more complete support for fully understanding how urban spatial structure affects traffic flows and reasonably grasping the traffic development trend.In this paper,mobile location data and open Internet geographic information data are emphatically analyzed.Based on data applicability and model universality,the application value of the above two kinds of data in the basic data mining of traffic demand analysis,such as employment and resident distribution,land use,traffic generation and distribution,will be discussed.In terms of mobile location data,mobile communication and mobile map location data cover similar data application scenarios but differ in data acquisition mechanism and principles.Therefore,space applicability of these two types of data is compared and analyzed.Firstly,the population structure was extracted from the demographic data to provide a benchmark for the expansion of employment and resident population and OD matrix.Secondly,the mining results of employment and resident population and OD matrix from mobile communication and mobile map location data were expanded to the population.Finally,expansion results were compared on district,subdistrict,and traffic analysis zone respectively.The results indicate that the mobile communication data using base station location technology are suitable for the analysis of macroscopic and mesoscopic scale,while mobile map location data using integrated location technology have more advantages than mobile communication data at the micro scale.Due to integrating the mechanism of data acquisition triggered by passive events,such as position switching and fixed time interval,mobile communication data has a stronger ability to continuously observe individuals and are more stable than the other one.In terms of open Internet geographic information data,the applications of points of interest and area of interest in land use identification,employment and resident population distribution and traffic demand forecasting were explored.For land use identification,the points of interest with area and weight attributes obtained via inverse geocoding interface and complementary area of interest from Amap and Open Street Map can support the analysis on traffic analysis zone scale.The former has better identification effect in construction land,while the latter can supplement the result of non-construction land.For employment and resident population distribution and traffic demand forecasting,firstly,a Back-Pro Pagation Network is built with number of points of interest by category as input variables.The prediction results of the model for employment and resident population and traffic generation on subdistrict scale are basically consistent with the mobile map location data,and the model also performs well in spatial portability.Based on results of employment and resident population distribution and traffic generation,population-weighted opportunity model and radiation model are compared with dual-constrained gravity model.Traffic distribution results show that population-weighted opportunity model and radiation model with no argument can also get the consistent results as the dual-constrained gravity model with the optimal parameters based on the natural attenuation mechanism of population distribution.As a result,employment and resident population distribution and traffic demand prediction are feasible on macroscopic and mesoscopic scale based on open data.Compared with mobile location data,it has the advantage of no data cost,which provides a new way for analysis without the support of sufficient data.The research results provide theoretical support for data selection under specific traffic demand analysis scenarios.Especially the successful application examples of open data based on points of interest and area of interest will expand the data acquisition channels for transportation planning.