Study On Urban Public Transportation Mode Based On Human Behavior Characteristics

High-speed urbanization causes social problems such as traffic congestion and environmental pollution,making it a consensus for all levels of government to prioritize the development of public transport system.The study of urban public transport,especially the regularity and pattern of passenger transit trip,will contribute to optimizing the allocation of public transport system resources and improving the operation efficiency,so it is the focus of studying public transport.For a long time,the study of urban traffic trip is subject to limited questionnaire surveys about traffic trip,most of which are limited to the macroscopic level.With the popularization and application of information technology such as traffic IC cards and vehicle AVL,people have obtained more comprehensive and accurate spatio-temporal data about passenger flow than those obtained through conventional questionnaires,which provides favorable conditions for in-depth study of the internal mechanism of traffic trip.Under the background of big data,this study proposes an improved scheme for passenger bus travel trajectory mining based on passenger credit card data and vehicle station data.Based on this,the dynamic empirical analysis of bus travel characteristics and the study of bus travel mechanism are conducted to explore the travel time and space map.Information,explored the problem of bus travel path identification.The feedback of travel research results is applied to the trajectory mining of public transportation,which helps to further improve the accuracy and efficiency of data mining and form a virtuous circle.The main work carried out in this study includes:Firstly,full sample data individual travel trajectory mining.This study proposes a method for estimating the missing data of the station system based on the time stamp of the passenger IC card and the travel time of the interval.The probability model based on the travel time is established,and the information of the passenger boarding station is matched under the condition that the station data is missing.By establishing a cross-correlation function between the passenger swiping time series and the vehicle reporting time series,the optimal value of the time error is obtained,and the vehicle reporting data is corrected;the probability inference and the pre-estimation of the residence are introduced,and the conditions for missing data are used.The maximum time to restore the passenger's travel time and space track.Secondly,behavioral dynamics travel characteristics and travel patterns.This study demonstrates that the passenger travel time interval deviates from the power law distribution while demonstrating the power law distribution “universality” of passenger bus travel characteristics,and thus innovatively proposes that the deviation of the travel feature from the power law distribution represents hidden.Order and deep law,that is,public transportation mode.Secondly,the research on the bus travel mode driven by demand,such as survival and study,is carried out.The regression analysis clustering model and time spectrum function model based on behavioral dynamics method are proposed and applied to passenger travel pattern recognition.Thirdly,analysis of bus travel maps.By constructing the travel time and space map of the passengers,the behavior habits of passengers travel can be explored,and the passenger travel mode can be clustered.According to the passenger travel trajectory,the passengers can learn the existing bus line network and use it to plan and design the existing bus network.Evaluation;by constructing the space-time map of the bus station,for the function identification of the area where the site is located.Finally,research on multi-path identification of traffic travel.The Gaussian cloud transformation model of multi-path recognition of traffic network is constructed.The travel time spectrum function is obtained by using the travel time history data of the station,and the travel time cloud model of the station is established.The concept division(ie,the identification of the travel path)is realized by the adaptive Gaussian cloud transform method.The concept of concept division is controlled by setting the ambiguity index in the transformation process.The validity of the cloud model is verified by comparison with the Bayesian information criterion method.The accuracy of the model evaluation is further improved by the improved method of segmentation superposition.