Research And Design Based On Wearable Human Fall Detection System

Human posture plays an important role in many fields.Such as some patients'rehabilitation training,human-computer interaction games,some special effects in movies,etc.How to detect and identify some human postures quickly and accurately has become a popular research topic.At present,the programs of human posture detection can be divided into two categories:One is by detecting human body's surrounding information,such as human video and images,sound,electromagnetic waves,infrared and other information changes as a basis for judging the determination of the posture;The other is by wearing some devices on human body,such as triaxial acceleration sensor,pressure sensor and inertial sensors to collect data which related to human posture,by which we can infer and identify the human posture.Combining the current empty nest of the elderly with the frequent occurrence of falls,sometimes the elderly do not get help timely after falling down.Considering to easy to carry,low power consumption,real-time three aspects,this paper designs a detection system which only based on a single three-axis acceleration sensor.The device is suitable for wearing on the waist of human body.In addition.the system designs automatic help and active help function.Then the fallen people can get help in time after falling down.The detection device consists of signal acquisition module,processor module and communication module.The sensor module is responsible for collecting acceleration data,then filtering.Processing module extracts acceleration rate SMVA,combined acceleration differential absolute mean MADS,acceleration signal strength SMA,human body's tilt angle BTA and other related eigenvalues.The communication module is responsible for positioning the user and sending help message.According to the detected data,we use the threshold method,the improved threshold method(multi-level fall detection algorithm)which is proposed in this paper and the singular decomposition algorithm of support vector machine respectively,to detect whether the human body fell.By comparing the accuracy of the three algorithms,the improved multi-level attitude detection algorithm is more accurate than the general threshold method,but the accuracy of support vector machine algorithm has small differences with the improved multi-level attitude detection algorithm.Considering the real-time and accuracy,we decided to use the multi-level fall detection algorithm as the final scheme.The hardware and software design of the fall detection system is completed base on this.By testing the various functions of the system,it can get a good realization of the human body fall detection function and automatically send SMS function.The accuracy and real-time of the examination effect both achieve the desired purpose.