Structural Model Of Road Network And Research On Vehicle Category Mining

Urban transportation is the premise and foundation for human activities and social economic development.As the number of cars increases and the acceleration of urbanization,some new problems has arisen while improving people's travel convenience.For example,the expansion of road network,road traffic accidents and traffic congestion.Therefore,it is necessary to conduct theoretical analysis and applied research on a large number of vehicle trajectory data generated by intelligent transportation systems to alleviate the above problems.These research results provide valuable information for traffic management departments,public security investigations,etc.The specific work is as follows:1.Establish a spatio-temporal general propagating deterministic zero-sided L system.The super-large road network generated by the vehicle trajectory data has self-similarity,asynchronous concurrency,spatio-temporal and randomness.The current mathematical model used to characterize the road network is basically a directed graph with simultaneous analysis by using complex network theory,which is not sufficient to formalize the vehicle trajectory with spatio-temporal.Therefore,the traditional L system was promoted by referring to the growth pattern of plant roots and its morphological structure model method.The concept of spatio-temporal general propagating deterministic zero-sided L system is proposed,which adds time and space symbols,direction symbols,driving time and probability symbols.It makes the system well-characterized vehicle trajectory and the generated road network with spatio-temporal,asynchronous concurrency,randomness and self-similarity.The system is used to model the morphological structure of the road network,formally describe the vehicle trajectory with its generating language,and combine the dynamic and flexible data structure to design the road network generation algorithm.The related case analysis shows that the model can overcome the shortcomings with synchronization and static model of road network structure.2.Calculation of the probability of the vehicle's driving trajectory.The prediction of the vehicle's driving path is realized by mining analysis these massive trajectory sequence data.Based on the spatio-temporal general propagating deterministic zero-sided L system,a mathematical model is established by combining the discrete Markov chain to the state of the vehicle,and a joint system is constructed to transform the vehicle trajectory into a state sequence of discrete Markov chains.The transition probability of the Markov chain gives a probability formula for calculating the vehicle trajectory.The corresponding vehicle trajectory probability calculation method is illustrated by a case analysis combining vehicle trajectory.The corresponding vehicle trajectory probability calculation method is illustrated by a case analysis combining vehicle trajectory.This work will serve as a good early warning for road traffic accidents,It can also be combined with the public security big data in the criminal investigation to play the role of analysis and analysis from person to car to case.3.Vehicle category mining algorithm.Based on the historical trajectory data of the vehicle,according to the type of vehicle,license plate color,and whether the trajectory has a periodic pattern,the private car,taxi and bus types defined separately which commonly used in three modes of travel,so as to propose corresponding mining algorithms.Base on a large amount of vehicle trajectory data based on bayonet video monitoring,verification of the proposed vehicle definition and the effectiveness of the mining method by vehicle driving data for a certain week in a city,the method can be applied to research on applications of people's travel modes.Combining vehicle category mining algorithm with the point of interest(POI),the residential area is taken as an example to analyze the potential reasons for people to choose the mode of transportation comprehensively.