Research On Travel Time Prediction And Reliability Estimation Of Expressway Based On Floating Car Data

With the rapid development of social economy,the requirements of urban residents for travel experience are also increasing.Travelers not only need to know the accurate travel time,but also want to master the delay and reliability of travel time,so as to reasonably plan the trip and arrange various affairs.Urban expressway is an important channel for residents to travel.Due to the imbalance of traffic supply and demand,serious traffic congestion appears,which brings great difficulty to travel time prediction.Therefore,it is necessary to carry out in-depth research on travel time prediction and travel time reliability estimation of expressway based on more advanced methods.In this paper,the grids are used to represent the road section.The vehicle trajectories of expressways are matched with the floating car data,and the grid travel time is extracted.Then an iterative path travel time prediction method is proposed,which effectively reduces the error of travel time prediction results.Finally,the travel time reliability estimation indexes are improved,and the travel time reliability is estimated by combining the travel time prediction value and the historical path travel time distribution.The specific works of this paper are as follows.1)Matching vehicle trajectories on expressway and establishing grid travel time database based on grid method.This paper uses GPS data of online car-hailing and GIS digital map of Beijing Expressway to divide the expressway area into grids and represent the road sections with grids.Considering the driving habits of online car-hailing drivers under the condition of carrying passengers,this paper proposes an effective method to identify the track points of the main road and the auxiliary road of the expressway.Taking the Fourth Ring Expressway in Beijing as the research object,the effective continuous track points of vehicles in the same direction are extracted.Then the grid travel time is obtained by linear interpolation and the grid travel time database is established.2)An iterative prediction method of path travel time is proposed.The grid travel time prediction is completed by selecting suitable short-term prediction model and determining different model accuracy evaluation indexes.Considering the comprehensive and diverse travel time demands of travelers,this paper proposes an iterative prediction method for path travel time,and conducts some case studies on the Fourth Ring Expressway in Beijing.The path travel time prediction effect is good,and the prediction error ranges from 3% to 11%.Compared with the direct prediction method,the iterative prediction method can also update the more accurate arrival time in the travel,and reduce the prediction error of the remaining travel time to 50% of the original error at most.3)Combined with the travel time prediction results,the travel time reliability estimation method is proposed.Travel time reliability is defined as the probability that the traveler can complete the travel within the travel time prediction value.Then,the distribution model is used to fit the historical path travel time data.In order to facilitate travelers to understand of the travel time reliability,this paper selects and improves the travel time reliability estimation indexes.The case results show that distribution models can well fit the historical path travel time distribution of the expressway with the right skewness or multimodal,and it has passed the Kolmogorov-Smirnov Test.The travel time reliability estimation method can indicate the potential delay of travelers through historical arrival rate,planning time index,and average travel time,and provide travelers with decision-making information including buffer time before travel.The travel time prediction and travel time reliability estimation method are proposed in this study can provide more accurate and valuable travel time information for travelers before and during travel,and help travelers plan their future travel reasonably.