Mining Ride-sharing Model Of Private Vehicles Using Big Data Of Electronic Registration Identification Of The Motor

The number of private vehicles in China has reached 189 million,accounting for 57.8% of the total number of motor vehicles.China has entered the vehicle society.The traffic congestion,traffic chaos,exhaust pollution,rise in oil prices and other problems have become the main problems of the current society.The ride-sharing is an important way to alleviate the above problems.Ride-sharing is a mode of transportation for residents to travel together by sharing vehicles.By sharing vehicles,the occupancy rate of vehicles can be increased,and the number of vehicles on the road can be reduced.Thus,the total traffic flow can be reduced,the traffic pressure can be alleviated,the emission of exhaust gas can be reduced,and the green traffic can be promoted.This thesis mined the ride-sharing model based on the data of Chongqing's electronic registration identification(ERI).The goal is to analyze the data of electronic registration identification to provide data-driven services.By mining the travel behavior of private vehicles,we develop public transport ride-sharing model and establish long-term and stable personalized ride-sharing model.The main work is as follows:(1)Vehicle trip chain extraction method based on electronic registration identification data.In order to understand the travel behavior of vehicles,we need to extract vehicles' trajectory chain and trip chain according to the characteristics of electronic registration identification data.Because of the cross-reading and missing-reading of RFID reader,we can get the complete trajectory chain by cleaning redundant data and completing trajectory based on Markov decision process,and extract the trip chain by setting the time threshold by quartile screening method.(2)Mining hot routes of private vehicle based on frequent sequence patterns.The hot route of private vehicles refers to the route that a large number of private vehicles pass in a certain period of time,which reveals the demand for urban ride-sharing.To illustrate the need of ride-sharing with data,we propose the PSSS algorithm based on the idea of frequent sequential patterns,which is used to mine and analyze the hot routes of private vehicles.(3)Mining regular travel behavior of private vehicles based on data cube.In order to develop personalized ride-sharing model and establish a long-term stable carpooling,we mine the regular travel behavior of private vehicles based on data cube,and provide important information for the recommendation of carpooling.(4)Similarity-based ride-sharing recommendation.After mining the individual's regular travel behavior,we recommend ride-sharing according to the similarity of travel habits.We measure the degree of recommendation from three aspects: spatio-temporal similarity of travel behavior,similarity of vehicles and similarity of departure place.Two people with high degree of recommendation are recommended for carpooling,so as to establish a long-term and stable personalized ride-sharing.Finally,the experiment is carried out with the real-world electronic registration identification data of the main urban area of Chongqing.Experiments show that carpooling can reduce 6.09% of private vehicles,3.35% of total mileage,and save more than 5 tons of gasoline during morning commuting period.The reduction of the number of private vehicles going out will reduce the total traffic volume and alleviate the urban traffic pressure.At the same time,the decrease of total mileage and the saving of gasoline are conducive to building a green and civilized society and reducing the cost of travel of residents.Therefore,the ride-sharing model can bring certain value to society,which is worth studying.