Study On Fall Detection System Of Old People

Elder monitor has received much attention in modern society. It is a very urgent issue to be solved for the family as well as society. The old man’s is in the degradation, so it is necessary to have the problem monitored to avoid accidents. The fall is the most frequently encountered case which also brings serious hurt even causes life risks for empty-nest elderly, The prevention of the fall becomes very difficult due to the difficulty of predication and its abruptness, we must find out the situation as soon as possible and timely rescue in order to reduce the harm caused by falls. It is necessary to prepare a comprehensive fall detection device therefore. In this paper, we have designed a real-time wearable monitor for daily activities of the elderly with the purpose of to notify the guardian in case of fall timely to release his concerns at the same time to secure the empty nest elder.The human body will produce all kinds of motion signals in the activity then based the status of body is monitored, but in the human activity in nonlinear non-stationary acceleration and angular velocity signal will lead to fall detection not accurate. In order to solve these problems, the paper proposes a Hilbert Huang transform data preprocessing. Firstly, the signal is decomposed by EMD, then the IMF function is analyzed by Hilbert transform. The method accurately restores the original signal characteristics and solves the problem. The algorithm solves the problem of nonlinear and non-stationary signals. The threshold value of the proposed algorithm is accurate. In addition, the threshold algorithm is used to detect and judge the human body state, which could be used to determine the accuracy of the human activity data, and to use the support vector machine algorithm to extract the feature vectors from the human activity. Finally, the hardware and software architecture of the whole fall monitoring device is realized based on the improved threshold method and the support vector machine algorithm. At last, the test is used to judge and detect the fall detection algorithm, and the accuracy of the whole system is verified.Through simulation experiments, the monitoring model machine system can timely monitor and accurately determine the human body’s fall in this paper. When the human body falls, the model machine system quickly determines the status, the average accuracy rate is 94.3%; and after the determination of human falls, the monitoring system timely sends fall information to the guardian. The research of this paper has a considerable significance to solve the problems of the elderly and to the management of the elderly.