The Research Of Wearable Fall Detection Method Based On Multi-Sensor Data Fusion

With rapid development of living standards and growth of life expectancy,the aggravating trend of aging population has become a serious problem faced by world.The aging trend in China has presented characteristics of fast speed and high level.With the serious empty nest phenomenon,the situation is more severe.Meanwhile,falling is one of the major threat that cause injuries even death to the elderly.If they cannot get timely salvation,it will not only increase the injury degree,but also lead to huge medical investment and economic burden for family and society.Therefore,the research and application of fall detection technology has positive effect on improving medical response level and improving the life quality of the elderly.However,there are still many deficiencies in the current fall detection technology.Fall detection system based on context awareness often has defects such as high cost,complex computation and privacy violation,while detection system deployed in wearable devices has defects such as insufficient accuracy,simple judgment criteria and poor anti-interference.Therefore,the deployment of multi-sensors and the application of data fusion have become the mainstream scheme to improve fall detection effectiveness and has brought significant improvement to wearable system.This paper presents the research and application of real-time fall detection system based on wearable data fusion.Firstly,it builds the activity model of elderly people based on three-dimensional attitude angle,designs and develops the sensor board integrated with three-axis accelerometer,gyroscope and Bluetooth,to collect the activity data of the elderly in real time and send them to the smart mobile phone through Bluetooth.Secondly,it extracts three-dimensional attitude angle and acceleration signal vector magnitude as the features of fall detection,to accomplish the denoising and fusion of acceleration and angular velocity data by Kalman filtering.It uses the sliding window and k-NN algorithm based on Manhattan distance to implement the system that can sense the fall of the elderly and give an alarm.Thirdly,the influence of various distance formulas and Kalman filter on k-NN algorithm was tested by simulation experiment.Finally,the performance of k-NN algorithm is strictly compared with various classification algorithms and experiment shows that the sensitivity and specificity of the k-NN algorithm based on binary classification are up to 98.9% and 98.5% respectively,which proves that the system has the feature of good real-time performance and high accuracy.The project provides strong technical support to improving medical response level.Meanwhile,it also provides new ideas and guidance for the further research and application of multi-sensors data fusion technology in daily life.