Research On The Distribution Prediction-model Of City Vehicle Velocity

In the past few years,with the constantly increasing of vehicle quantity in China,the pollution of vehicle has been focused by many areas in society.Our government has also legislated for controlling the problem of exhaust emission.Vehicle driving cycle is one of the most important parts of fuel consumption,measuring and analyzing vehicle exhaust.For different types of city traffic environment,it is improper to use only one driving cycle to describe complex city driving situation and test the exhaust emission precisely.Therefore,our country needs to construct different typical driving cycles according to our actual city traffic environment.Using what kind of index or distribution to define the category of one city is one of the priorities for nowadays.The existing literatures have boundedness in this research area.For choosing typical cities of driving cycle,the existing papers do not define a quantitative evaluation index.For building prediction models of vehicle velocity,existing papers need a large number of velocity data as the input of the model to predict the velocity for this moment.And mostly these kinds of prediction models are used only in part of the city.City traffic environment is composed of vehicles,roads,intersections and many other traffic environment characteristics which effecting the vehicle velocity in different ways.But no matter how complex the traffic environment is,vehicles always drive in one road section at one time and decide the next move based on traffic lights.That is to say,the intersection-road model is one of the basic models in city traffic environment.Therefore,we proposed to use the probability distribution of vehicle velocity as a quantitative evaluation index and aimed to build a prediction model by the entry point of intersection-road model.The prediction model used city traffic environment characteristics as inputs,including the road data,traffic density,vehicle quantity,the average travel time of city residents and so on,and obtained the probability distribution of vehicle velocity by related calculating.In this way,the prediction model could realize to define the category of one city without collecting a large number of velocity data.The main work and results were summarized as follows.First,this paper finished the data processing of original velocity data and the data extraction of the digital map of Chang Chun main city area.The original velocity data was offered by the transport department of Chang Chun.We choose part of the valid data which satisfied our purpose of research.The digital map was built by the Vehicle Operation Simulation Group.We extracted the road data,intersection data and some other traffic environment data from the digital map.Also we got the data of average time of urban travel,the average frequency of urban travel of city residents from literatures and calculated the city traffic density.By using the latitude and longitude data of vehicle,this paper accumulated appearing times of vehicles in different regions and divided Chang Chun main city area into four areas.Secondly this paper built an intersection-road model.For a single intersection,we used the road length,the acceleration of the vehicle,the percentage of idle vehicle,the total number of vehicle and some other characteristics to deduce the equation of vehicle velocity and these characteristics.Then we applied the single intersection model to the city road net and got an intersection-road model by conversing local characteristics to macroscopic characteristics.Then we coped with the road length data and vehicle velocity data to get the fitting model respectively by using the method of statistical inference.The accuracy of fitting models is verified by a_ab as an evaluation index.Next this paper built a distribution prediction-model of vehicle velocity.We analyzed implications of different characteristics in the fitting model and found the inner relationship between characteristics and road length distribution,intersection-road model by connecting with city traffic environment.Then we could got the equation of different characteristics in the fitting model and the city traffic environment characteristics.That is to say,for taking the area as an object,we built a prediction model by using area traffic environment characteristics as inputs and getting the distribution prediction-model as an output.Then based on it,we got a prediction model by taking the city as an object which using city traffic environment characteristics as inputs.At last,this paper verified the prediction distribution-model of city vehicle velocity.The verification data included four groups,which were composed of twenty sub-databases randomly selected from the database.The paper took area and city as objects respectively to verify the prediction model and used a_ab as the evaluation index.