Urban Residents' Travel Mode Research Based On Smart Phone

With the rapid development of urbanization and the continuous improvement of people's living standards,the number of motor vehicles is gradually increasing.The increase in motor vehicles' number has caused some traffic problems at the same time of bringing convenience to people's life.As an important part of traffic investigation,residents' travel survey has a great effect on the understanding of traffic problems.Most of the traditional residents' travel survey adopts the method of manual investigation,which depends on the respondents' subjective perception of the problem to a large extent.There are many problems in this way of investigation,such as the respondents' heavy burden,high rate of refusal to visit,and the need of a lot of time and manpower.In order to overcome the shortcoming of this method,the collection of travel information based on smart phone has come into being,which provides a new technical means for the study of residents' travel mode.The main contents of the thesis include the following parts:(1)Conducting travel survey based on smart phone and network survey technology.(2)The selection of eigenvector.The thesis chooses travel distance,travel time,average speed,acceleration,95 percentile speed,95 percentile acceleration these six kinds of feature vectors to identify the way of travel.(3)The method of travel pattern recognition based on GPS data is proposed.The thesis uses the two-stage method to identify the way of travel.Firstly,the way of subway and walk are identified by pattern recognition,and then the remaining four kinds of travel modes are classified as motor vehicles and non-motor vehicles respectively.The recognition accuracy89.5%.On the basis of pattern recognition,SVM classification algorithm is used to identify bicycles and electric vehicles,cars and buses.Both of these two pairs of travel modes are easily confused with one another.And the recognition accuracy is 91.9%.Among them,the identification rate of electric vehicles and buses is the lowest.The innovation of the thesis is mainly embodied in the following aspects:(1)A travel investigation method based on smart phone and network survey technology is proposed.In order to reduce the burden of volunteers and improve the accuracy of travel characteristics,the thesis constructs an online platform which includes individual socioeconomic attribute data and GPS data storage and trajectory data processing.Besides,this thesis discusses the methods of the investigators' participation in the verification of volunteer travel characteristics.(2)A two-stage method for the identification of travel patterns using GPS data is proposed.In the first stage,it is proposed that the six ways of travel namely walk,bike,E-bike,car,bus and subway are identified as subway,walk,motor vehicle and non-motorized travel modes.Then,in the second stage,the support vector machine is used to focus on non-motor vehicles and motor vehicle travel.The remarkable characteristics of the subway,walk and other travel modes are fully considered by using the two-stage classification method,and a new idea is put forward for improving the precision of the travel pattern recognition.(3)In this paper,a rule-based method,namely pattern recognition method and machine learning method support vector machine algorithm,are used in combination with R language to identify bike and E-bike,car and bus these two groups of travel modes,both of which are with lower precision.