Research On Medthod Of Fall Detection Baseed On Three Axis Acceleration Sensor

With the increasing of aging population and empty nest families, the health care of older persons has been widespread concern in China. Fall is an important factor that influence the health of the elderly, Fall not only hurt the old people’s physical and mental health, but also bring economic losses to the society and family. If we can detect the fall action accurately and quickly when the elderly fall, then the elderly can get timely treatment, so that not only the damage caused by falls in the elderly but also unnecessary economic loss will be reduced.In this paper, a fall detection algorithm based on body acceleration is presented, and gives a method to product a wearable fall detection device. The main work of this paper can be classified into three parts: ① the research on the algorithm of fall; ② the implementation of fall detection device;③make experiment to verify the performance of the fall algorithm,what more, I put forward a improvement method to the fall algorithm.We found that the body acceleration and great changes a lot when human fall down. So, wo use combined acceleration,acceleration mean and body great as judgment of fall detection. Through the analysis and comparison of the merits between method based on the threshold judgment and method based on pattern recongnition, we designed a fall detection algorithm based on threshold judgment.In this paper, we propose hardware and software implementation of the fall detection device. Hardware selection is mainly based on low power requirement of wearable devices. We set a button function in order to prevent the occurrence of under-reporting and over-reporting. When the user does not fall but the fall detection device alarms, the user can cancel the alarm by the button. When the users falls but the fall detection device does not alarms, the user can push the button to get help.In order to test the performance of the fall detection system,we invited multiple experimenters with different height and different weight to perform a large number of experiment, the experiments show that the fall detection has good detection effect. At the same time, we found that the acceleration in Y-axis has a very significant change while the body falls. Thus,we proposed a improved fall detection algorithm based on Y-axis acceleration. And the improved fall detection algorithm has a good performance. The correct detection rate increated by 0.8%.