Research On The Elder Fall Detection Algorithms

According to the incomplete statistics of related research institutes,by the end of 2014 in China,population of 60 years age or older people reached 221 million,which accounted for 15.5% of the total population.In recent years,as the growth of the economic and cultural level,Chinese aging problem has become increasingly serious.The protection,rescue and health problem of old people has become the main problem of the society.If it isn't handled properly,not only the government institutions,social groups and individual residents will face great trouble,but also it will affect the political stability of the society.This thesis starts with this point,concerning about the problem of elder falling,combining the universal recognition of high-precision,low-cost,easycarry miniature sensors,and putting forward a real-time monitoring and automatic alarm system to detect the elder fall.This system is based on the frequent elder fall phenomenon,and focuses on how to more accurate and convenient to identify the activity status of the elderly.By analyzing the current old people fall detection methods,it points out the disadvantages of various methods,and puts forward an optimization method based on acceleration and the variation of the angular velocity.This algorithm is different from the commonly used fall detection algorithms based on acceleration and angular velocity differential extremum.It provides an easier way to distinguish fall action from daily actions through threshold,so as to improve the accuracy and reliability of the algorithm.The subsequent experiment results proved the feasibility and effectiveness of the algorithm.This thesis finally implements a complete fall detection system.The hardware part of the system adopts the Arduino as access point with external three-axis accelerometer.Human motion information can be gathered by the angular velocity sensor,and be transferred to cell phone via bluetooth transmission module for subsequent processing.The software part completed an alarm software under the Android system.When the software receives the action data by the bluetooth,it extracts the body acceleration rate and angular velocity gradient,judging by the proposed fall detection algorithm.If discovering the falling action,alarm will be activated.The alarm types include sending a text message for help,emergency calls,and reminded bell.To validate our system,the thesis collected the normal movements including walking,sitting,crouching and bending.There were 10 participants,including five men and five women,aged 22 to 28.Each of the experimenters walked,sat,squatted,stooped and fell and waited four times.The experimental results show that the fall detection algorithm based on acceleration rate can effectively identify fall with an acceptable range's accuracy,and the algorithm is simple,efficient and can be easily embedded into all kinds of mobile devices.