Research On Urban Traffic Operation State Identification And Travel Time Prediction Method Based On Grid Model

As the booming growth of the auto ownership,the traffic problems of the megacities have become more and more serious.In addition,the road networks of the megacities possess the characteristics like the high probability of incidents,various traffic conditions,etc.Therefore,it is significant to monitor and analyze the real-time traffic conditions.Meanwhile,with the development of big data technologies,data collection,transmission,and analysis have also growing mature,which improves the efficiency of the road network operations in order to help the travelers to complete their trips more safely and efficiently.The traditional approach of using floating vehicles to extract the transportation parameters is usually based onmatching the trajectory data and the GIS vector data.In brief,it evaluates the traffic conditions according to the positions and velocities of the vehicles.The drawback is that this approach highly depends on the accuracy of the GIS map data.However,in this research,we drop the basis of GIS vector data but utilize the trajectory data of floating vehicles to evaluate the traffic congestion level and travel time.In short,this method manages the humongous trajectory data to divide the research area in togrids and extract the corresponding traffic parameters on grid basis so as to reflect the traffic condition of the area.This paper will include the following sections.First,preprocessing of the trajectory data will be illustrated.Then,a sample grid model will be developed to simulate a large-scale area.Furthermore,a detailedmodel will be constructed with traffic flow parameters,such as travel path and travel time.Based on these parameters,congestion prediction,and travel time estimation.Eventually,this paper will contribute from the following five aspects:(1)By analyzing the floating vehicle data,summarization of the quality of the data will be done.Then this research will develop a cleansing standard and framework for relevant trajectory datasets according to the objective of the following chapters.Programmingdeveloped to preprocess the data based on the framework and traveltime prediction.Finally,the program will validate and clean the humongous Beijing floating vehicle trajectory data of Oct 2012 to qualify the following research objectives.(2)Based on matching the trajectory data and gridding map,this research will construct a corresponding sample grid model to extract traffic operation parameters and study the traffic condition of the gridding megacity networks.We will finally propose a classified grid congestion map according to different times,spaces,and road network structures.(3)The grid model will be constructed in four categories including grid generation,trajectory data extraction,static characteristic extraction,and characteristics of traffic dynamics.Eventually,the time-dimension will be incorporated to form a three-dimensional model.Then,we will propose a methodology to match the trajectory data and grid models to simulate the traffic conditions effectively.Meanwhile,we will also extract the static traffic characteristics by referring to the historical trajectory data and the improved clustering algorithm to further achieve the network features and traffic condition attributes in the grids.By analyzing the travel time within the grids,we can differentiate the frequent and accidental congestion spots and time periods.The validation process will also be done to evaluate the effectiveness of the approach.(4)We will propose a travel time prediction model by using historical travel time,real-time travel time and applying Multiple Linear Regression and KNN nonparametric regression.Based on the trajectory data and the detail travel paths,the research will comprise grid sets to represent a new way of travel paths.According to the demand of traffic operators,new grid sets of travel paths will be developed to analyze the travel time distribution and reliability for different travel distance.By researching the travel time within the continuous grids,we will propose an effective travel path extraction approach using the trajectory and OD data.Finally,we will combine the travel path extraction approach and travel time estimation by nodes to propose the overall travel time estimation(5)A travel time prediction system based on historical data will be programmed by combing the technologies of.NET Framework and ArcGIS Engine.The system provides users with frequent/accidental congestion spots and time periods both historically and in real-timewhen importing the historical trajectory data of float vehicles.It will also construct the travel time estimation model based on the trajectory datasets and the OD parameters from the users.Finally,it will provide an estimated travel time and visually display the functionality by applying the dynamic link library of ArcGIS Engine.