Study On The Travel Distribution Characteristics Of Urban Traffic Network Based On Taxi GPS Data And Its Impact On Congestion

The urban transportation system is a complex giant system,and its operation law is extremely complicated.The complex network method that emerged at the end of the twentieth century provides a new perspective for studying urban transportation system.Empirical studies shows that the actual traffic network topology has a small-world or scale-free characteristic.Some scholars study the network congestion characteristics of different topologies in combination with traffic flow theory,and further explore the connection between the traffic network topology and traffic congestion.However,most of the existing researches pay less attention to the impact of traffic distribution characteristics and spatial distribution characteristics on the network congestion.The development of multiple technologies such as mobile communications,the Internet,spatial positioning,and big data has greatly expanded the access to traffic travel data.These travel data range from the initial census,questionnaire surveys,and US dollar circulation records to today’s various high-precision vehicle GPS trajectories data,mobile phone signaling data,and public transport IC card data.Human understanding of the traffic travel has made a qualitative leap in scale and dimension.The research results on the spatial and temporal characteristics of traffic travel demand keep emerging.A large number of empirical studies show that human traffic behaviors are highly bounded,periodic and regular.By mining largescale spatial interactive data,existing studies have proposed a variety of macro traffic distribution prediction models to discover general patterns of user travel.These models show a high prediction accuracy on the actual data set,which provides a very valuable reference for the prediction of traffic travel distribution.It is of great practical significance if the spatial characteristics of traffic demand excavated from actual data are applied to the study of network topology and traffic dynamics.Starting from the data,this paper focuses on the study of the travel distribution of different cities and its impact on the road network congestion.The OD distance distribution of passengers in different cities is first extracted by deeply mining the GPS track data of taxis in multiple cities.An algorithm is designed to identify the travel "thermal area" based on the GPS data,so as to find the relationship between the shape of the thermal area and the OD distance distribution of different cities.Then a demand construction method that satisfies the needs of specific travel distance distribution is designed to generate the travel demand that meet specific spatial characteristics on different networks.Finally,combining the travel demand of different spatial distribution and the traffic network dynamics theory,the traffic congestion characteristics of road networks with different topologies are evaluated.The main contributions of this paper are four aspects as follows:1.Based on the historical taxi GPS data of nine cities in China,the OD linear distance is extracted according to the passenger carrying status.The OD distance distribution of taxi trips in different cities is analyzed.The results show that the probability density of urban taxi OD distance follows the logarithmic normal distribution,and the complementary cumulative probability density follows the exponential distribution.In addition,the OD distance distributions in different cities are significantly different.2.In order to further explore the causes of the differences in the travel distance distributions in different cities,a thermal region recognition algorithm based on OD points is designed to identify the high-density region boundary of plane scattered points in this paper.The algorithm is then applied to the taxi GPS data in nine cities in China to identify the boundaries of the travel thermal area in different cities.Finally,the relationship between the geometry shape of different urban thermal areas and OD distance distribution is analyzed,and the possible causes are discussed.3.A method for generating traffic demand distribution is designed,which can generate demand OD that meets specific requirements on different networks according to the specified distance distribution function and the total demand.By comparing the traditional uniform distribution generation method and the PWO model prediction method,it is verified that the method proposed in this paper can ensure that the total Vehicle Kilometers Traveled(VKT)of the travel demand is controllable,so as to ensure the comparability of travel demand on different topological networks.4.In order to explore the optimal traffic network topology under different spatial characteristics of travel distribution,the method based on the distance distribution function is used to generate travel distributions with different spatial characteristics.Then congestion characteristics of the different topological traffic networks are re-evaluated according to the static traffic balance assignment.