Study On Abnormal Detection Of Elder Travel Behavior Based On Hidden Markov Model

China has already been into aging society, and the aging of the population brought aboutmany social problems. The aged people may be lost or injured in daily life since their physicalquality have been weak. With the development of technology, the precision of GPS has beenimproved significantly, and intelligent equipment with the GPS module has been appliedwidely in our daily life. Hence, it is convenient to monitor the elderly travel by the intelligentequipment. Making these devices from recording time and location to recognizing behaviorfor intelligent elderly care, it is of great social significance.In the process of monitoring the elderly, intelligent devices can obtain amounts of GPS data,but these trajectory data cannot be directly observed. This thesis focuses on how to effectivelyextract the information of people’s activity behavior from trajectory data of trip, and use thisinformation to figure out whether the elderly travel is abnormal.The main study contents of this research include:(1) Research on the problem of motion analysis based on trajectory data. According to themotion characteristics, divide the trajectory into movement and stop sequence, and find thestop points. Then analyze the correlation of the elderly travel purpose and travel path, andutilize clustering analysis to extract points of interest (POI) where the elder people gofrequently. Finally, the experimental result indicates that POI recognition algorithm hasexcellent performance.(2) Research on the problem of detection of abnormal travel behavior for elderly. We utilizedirection angle of trajectory to train Hidden Markov Model (HMM). Meanwhile, it improvesthe way to identify hidden states, and builds the model of travel behavior. In addition, use themodel to determine whether the behavior is abnormal and to develop classified alarm system.The experimental result indicates that this approach has outstanding performance on detectionof abnormal travel behavior.(3) Develop software to assist the research. The system can manage the data of travelling,extract the elder people’s POI of travelling, build HMM of travel behavior，and detectabnormal travel behavior.