Fall Detection Based On Doppler Radar

With the increasing trend of population aging,the total number of empty nesters is increasing.The health problems of the elderly have attracted more and more attention and attention.Accidental fall is a major killer that endangers the physical and mental health of the elderly.How to detect accidental falls in elderly people and avoid the two injuries and complications caused by falls is an urgent problem in health care for the elderly.The detection method based on radar has strong penetration ability,non intrusive sensing and good privacy protection.It is not easy to be affected by environmental factors,such as weather,temperature,light and so on.It has the ability to compete with traditional video surveillance and wearable sensor.In recent years,it has been widely used by scholars at home and abroad.Note.But at present,the method is still in its infancy,and the radar echo signal processing method is not yet mature,especially the recognition ability of different human gestures,such as station,sitting,kneeling and other movements,remains to be strengthened.Therefore,this paper uses Doppler radar to study the fall detection,and focuses on the echo signal processing methods,including signal preprocessing,feature extraction and classification recognition.The main research work of this paper includes:1.The main methods and research status of fall detection are introduced,and the advantages and disadvantages of various methods are analyzed;expounds the principle of Doppler radar fall detection,and introduces the method of radar echo signal processing.2.The Doppler radar system based on single frequency continuous wave isbuilt and debugged,and the corresponding algorithm for detection and recognition of radar fall detection is designed.First,the original radar echo signal is filtered by digital filter,and then the time-frequency analysis of the filtered signal is carried out by short time Fourier transform.Finally,the support of the optimized parameters is used.A classifier is used to classify the fall events.The experimental results show that this method can realize accurate recognition of fall action.3.A Doppler radar system based on vector network analyzer is built,and the corresponding algorithm of radar drop detection and recognition is designed.The experimental results show that the system can determine the direction of human fall by the positive and negative frequency signal after the time-frequency transformation,and has a higher recognition accuracy for different posture of the human body.