Research And Application Of Fall Detection System

With rapidly expanding of elderly in China, aging has become increasingly serious. Social health of the elderly is also doubly concerned. The elderly great easily fall, because of aging of their body and control ability of the central and peripheral nervous system decreasing. Fall is one of the important causes of illness and death the elderly suffering from, and it's a serious threat to the physical and mental health of the elderly. Checking and timely assistance for the elderly falls not only greatly reduce the damage caused by falls for the elderly but also help to improve the quality of life of older people. According to demand of national natural science foundation project, the main content of this paper is to study a set of fall detection system which monitor real-time the elderly daily activities and raise the alarm in the event of a fall. The main purpose of system is collection activities data of human behavior by wearable sensor module at the waist and identification behavior activity of the wearable by detection algorithm stored in the processor. When fall incident happens, in order to provide timely assistance and reduce the damage caused by fall, the GPRS/GPS module sends the location geographic to corresponding relatives or medical institution.The main work and results are as follows:1. The disadvantage for the fall detection algorithm based on a threshold of anisotropic and cost sensitivity with considering shortage, fall detection algorithm which different ages set different the three-dimensional acceleration amplitude is proposed by collecting behavior activity data of different age sand comparing with analyzed fall data.2. Both inadequate Pattern Recognition training set sample and the particularity classification problem about normal or fall behavior in the fall detection, as a result, this paper proposes the improved AdaBoost with Support Vector Machine algorithm to train the classifier.3. Combined with high accuracy based fall detection algorithm threshold value and simplicity based on pattern recognition fall detection algorithm, this paper proposes fall detection algorithm based on cascade structure.4. In the fall detection system overall design, this design was analyzed from two aspects of hardware and software solutions to select modules.Finally, experimental results show that the proposed fall detection system for different ages the average sensitivity, accuracy, non-response rates respectively is 94%, 92.1%, 5%. And with the increase of age, the detection performance is superior.