A Study On Travel Chain Activity Type Recognition

With the popularity of mobile Internet,mobile Internet applications contains a huge opportunity,traffic patterns of automatic identification obviously has a broad application value both in field of traffic planning and LBS(location-based services).At this stage,smartphones have become more and more powerful functions,communication is no longer the only function of mobile device,smartphone not only has powerful lies in its communication function,but strong perception,computing and storage capacity.In the current study of traffic pattern recognition based on smart phone,single source of data is difficult to make the traffic pattern recognition algorithm with good adaptability,in addition,the pattern recognition algorithm itself also has different ability to adapt to different scenarios,which is great obstacles for improving traffic pattern recognition accuracy.In this paper,the traffic pattern recognition method is based on random forest,research contents mainly includes the following several aspects:1,Introduce a variety of data sources,solve the problem of the single source of data in the traffic pattern recognition,and study the influence on identification accuracy when new data source is introduced.2,Research the influence of different traffic pattern recognition algorithm on recognition accuracy,especially random forest algorithm and other algorithm with their similarities and differences in performance.3,Traffic pattern recognition system based on smart phone demand analysis and design,including system demand analysis,system design,detailed design,etc,in terms of design,detailed study of the system data acquisition,feature extraction and construct model based on random forest,in terms of the detailed design,mainly completed the design of the system interaction module,memory module design,data acquisition module design,the design of the feature extraction module,pattern discriminant module design,the design of the main control module.4,Traffic pattern recognition based on smart phone system implementation and testing,mainly including system function modules,traffic mode discrimination test results,etc.This paper implemented the transportation recognition system based on smart phone,after the test,the static,walking,running,bike,bus,drive,subways,seven kinds of traffic patterns discriminant accuracy reach 90.5%,has the very good use value.