Ride Sharing Group Discovery For Commuting Private Car Based On Spatio-temporal Semantic Similarity

According to the statistics of the traffic control bureau of the Ministry of public security,by the end of 2020,the number of private cars in China has reached 244 million.Private car has become one of the main modes of transportation for urban residents.It is found that commuting private car sharing group experience is an effective means to solve urban morning and evening peak traffic congestion.In the past,the research on carpooling mainly comes from taxi data and public transport data,while private car data is rarely involved at home and abroad because of its high difficulty in obtaining privacy.Electronic vehicle identification technology provides a driving force for the research of commuting private car sharing.Uberpool,Di Di carpool and other large-scale carpool service platforms provide users with carpool services based on passenger requests.However,studies show that only a small number of commuters choose to use such carpool software in the morning and evening rush hours.The fundamental reason is that commuters expect to get a long-term and stable ride sharing service.On the premise of ensuring the similarity of travel time and space,if the characteristics of each carpool user can be considered,the success probability of commuting private car carpooling can be improved.Therefore,this thesis proposes a commuting private car sharing group discovery model based on spatio-temporal semantic similarity.The model takes into account the similarity of commuting time and space,and adds the commuting destination semantic as the similarity measure.The semantic part is composed of the recognition results of urban functional areas,and takes the out of the commuting private car sharing group as an example Based on the recognition results of urban functional areas,this thesis explores and analyzes the traffic pattern.The main work of this thesis is summarized as follows(1)Aiming at the problem of urban functional area identification,considering the heterogeneous data collection and management in the form of grid,this thesis proposes an urban functional area identification model based on multi classification support vector machine.Firstly,the study area is defined and the traffic grid is established.The POI data and road network data are mapped into the traffic grid to complete the traffic district division.Then the urban functional area identification model is constructed to identify the functional area of the traffic district.Finally,the experimental results are analyzed.(2)Aiming at the problem of commuting private car sharing group discovery,this thesis proposes a commuting private car sharing group discovery model based on spatio-temporal semantic similarity,considering the distance measurement from three aspects of time,space and semantics.Firstly,the representative travel of commuter is extracted.Then,with the representative travel of commuter as the input,the space-time semantic distance measurement function is defined,and the commuting private car sharing group discovery model is realized.Finally,the experiment is completed with the private car data of Chongqing.Combined with the recognition results of urban functional areas,the travel rules of the sharing group in the experiment are explored and analyzed.