Analysis And Application Of Bus Passenger Trip Characteristics Based On Activity Stability

With the development of intelligent transportation systems and the improvement of bus IC card and mobile device swiping technology,a huge amount of travel data has been formed in the background of bus operation,which is widely used in the field of urban public transportation planning and management.Past studies have mostly analyzed the passenger flow of the entire public transport network from a macro perspective,or divided the bus market according to travel time,occupation and age,ignoring the differences in passenger travel patterns.As a result,the existing public transport service improvement strategies are not sufficiently targeted.Based on the long-term travel time and space law of individual passengers,this paper uses an improved clustering algorithm to segment the public transport market,and tries to obtain the demand for public transportation services for various trips.This paper takes conventional bus travelers as the research object.First,the IC card data,vehicle operation data,and static data of routes and stations are matched based on passenger travel characteristics to complete the inference of passenger boarding stations,alighting stations and transfer behavior.Furthermore,the individual passenger’s travel trajectory information for multiple consecutive days is constructed with a single-trip behavior recording unit.Then,based on the long-term travel time and space law of individual passengers,the concept of traveler activity stability is proposed.Furthermore,according to time stability and space stability,passenger travel is divided into time and space stable travel,space stable travel,time stable travel,and irregular travel.Then,the applicability of the DBSCAN algorithm for mining individual traveler’s spatiotemporal patterns is introduced.Furthermore,combining the limitations of the algorithm’s low efficiency when processing large samples and the characteristics that travelers will repeatedly visit the same stations during long-term trips,the WP-DBSCAN algorithm is proposed.Then,the WP-DBSCAN algorithm was used to cluster the passengers’ boarding time,boarding station and alighting station.In the clustering of boarding time,the time window was used to pre-classify the time points.Furthermore,the result of bus market segmentation based on the stability of individual passenger activities is obtained.According to the clustering results,the characteristics of each type of travel are analyzed from the basic travel indicators,time distribution and spatial distribution.Then the demands of various types of travel for public transport services are summarized.Finally,the application suggestions of this research in public transportation operation management and network optimization are given.The results show that: 1)The efficiency of the WP-DBSCAN algorithm is significantly higher than that of the DBSCAN algorithm when there are a large number of overlapping points in the sample.And as the sample size increases,the WP-DBSCAN algorithm has more obvious advantages.2)Trips divided by time stability have obvious differences in the distribution of time and trips divided by spatial stability have obvious differences in the distribution of space.It indicates that it is reasonable to divide the bus trips based on the time stability and space stability of the individual travel of passengers.3)Travellers corresponding to stable trips in time and space care more about improvement road conditions and services in the rush hours.Travelers corresponding to irregular trips are more concerned about improving the accessibility of bus services.Trips with time stability are mainly concentrated in the rush hours,and the corresponding travelers have higher requirements for punctuality.Travels without time stability are more flexible in time.Travels with spatial stability are mainly concentrated in the core area.Travels without spatial stability are more scattered,and the corresponding travellers pay more attention to the scope and convenience of bus services.In this paper,a more efficient clustering algorithm is designed to mine individual passenger travel time and space patterns.The classification results will help the public transportation management department understand the characteristics of passenger travel,and adjust the operation strategy and the network structure in a targeted manner to improve public transportation services and operations.The research provides a certain theoretical support and application value for the refined management and service of public transportation.