| Urban traffic congestion has become a major problem to be solved in urban development,with the acceleration of urbanization process and the rapid increase of vehicles.Vigorously developing urban public transport and advocating green travel are the common initiatives to ease the traffic pressure in most big cities.As an important component of urban public transport,taxi has the characteristics of convenience and flexibility.However,subject to the information asymmetry between supply and demand,difficult to take a taxi,high taxi no-load rate,high taxi operating costs and so on,these issues seriously hampered the healthy development of the taxi market.In view of this,some scholars put forward the taxi carpooling concept in order to solve these problems,which quickly aroused resonance.The empirical data show that taxi carpooling has policy effectiveness in terms of alleviating urban traffic congestion,solving the plight of taxi operations and so on.This paper reviews and summarizes the research status and development trend of domestic and foreign taxi carpooling,analyzes and concludes the problems and shortcomings of the study of urban taxi carpooling.Through the analysis of the travel characteristics from the aspect of urban taxi and passenger,it constructs the urban taxi carpooling probability and taxi carpooling waiting time models,puts forward the urban taxi three stages algorithm,and establishes the urban taxi simulation system based on the data of Nanjing city.The main contents of this paper are as follows:(1)The paper discusses the origin,development,concept,type and organization model of urban taxi-carpooling,analyzes the present situation and existing problems of urban taxi-carpooling,and constructs and elaborates the technical framework of urban taxi –carpooling system.(2)Taking Nanjing city as the research object,The paper analyzes the travel characteristics of Nanjing taxi from the taxi mileage,taxi no-load rate and taxi passenger characteristics.It analyzes the changing characteristics of taxi passengers in Nanjing from a series of indexes,such as passenger travel distance,travel time,travel point equilibrium,trip point density and travel point distribution.Therefore,the influencing factors of urban taxi carpooling are put forward.(3)By using the Poisson distribution theory,the probabilistic characteristics of passenger vehicles in the taxi ride are studied,the model of taxi carpooling waiting time is constructed,which is based on the taxi carpooling Poisson distribution model and the N Bernoulli experiment principle,the taxis number,the no-load rate,the average travel speed,the destination distribution and other factors affected on the taxi carpooling probability and taxi carpooling waiting time have been analysed and discussed in detail.The model was validated by taking the taxi GPS data of Yuhua Road in Nanjing as the data source.The experimental results show that the taxi passenger vehicle data pass the K-S test,obey Poisson distribution.With the large number of taxis,the low no-load rate and the concentration of the destination,it will increase the probability of taxis and reduce the waiting time.The stability of the average driving speed has small effect on the probability of the taxi carpooling and carpooling waiting time.In the rush hour,the multiple carpooling method is used to improve the carpooling efficiency.(4)The paper proposes a taxi cluster algorithm based on data field energy and distance.By using the function of the data field energy,the field energy of each data point in the data is solved,by comparing the size of each data point in the energy value,each data point spacing is calculated,then according to each data point and interval value product field which determined by the threshold to determine the cluster centers point,sub data points including outliers and the clustering center.A case study of Nanjing passenger taxi track data,compared with the classical clustering algorithm,this algorithm has better clustering effect for the spatial distribution of a certain law,the irregular distribution of the taxi trajectory data.Therefore,the clustering algorithm is more suitable for urban traffic road vehicle clustering;this innovative research ideas and methods provide a new perspective for the study of urban traffic clustering(5)Based on the judgment of the distance threshold of the carpooling taxi destination and the passenger destination,the distribution scheme of the passenger and taxi can be obtained.According to the matter-element theory,the fuzzy matter-element matrix is constructed,which calculates the weight of 4 indicators as carpooling distance,distance of reaching the pick-up point,distance of reaching the destination point and the time of reaching carpooling pick-up point by the entropy method,and which uses the closeness degree to calculate the priority of the taxi distribution scheme.The experimental results show that the algorithm is feasible and effective,and the evaluation results are more objective,because the parameters of the algorithm are less constrained by human factors.(6)Based on the vectorization of the road,the topology structure of the network node is established,the road weight of way is constructed by use of the road length,average vehicle speed,crossing traffic waiting time value index.The weighted matrix between nodes is obtained by using the direct weighted matrix algorithm,the node identification based on region constraints and improvement of A* algorithm are proposed.By adjusting and judging the weight matrix of the nodes,the path planning and the carpooling cost estimation of the passenger and the taxi are achieved,and thus the reference proposals for the carpooling passengers are put forwardIn this paper,a lot of basic theories of taxi carpooling matching are studied,the research results can provide theoretical support for the urban taxi carpooling matching.A simulation system of urban taxi carpooling is constructed based on the above theoretical research in this paper,the content includes two parts: the overall design of the system and the functional structure of the system.The simulation results show that the proposed algorithm is feasible in practice,and this contribution provides an important support for the follow-up study to eliminate all kinds of carpooling assumptions and improve the carpooling technology plan. |
PDF Full Text Request