| With the development of aging society, the proportion of the elderly population is growing year by year. Physiological characteristics and other factors cause the special behavioral characteristic-easy to fall, which is a serious problem for elderly people. In China, fall as a phenomenon in the elderly is more worthy of attention. Therefore, it is significant for the elderly that falls are accurately detected and the alert measure is taken to obtain timely assistance.Firstly, the paper has analyzed the research status of fall detection system at home and abroad. Then a fall detection alert system based on wearable sensor is proposed. The system can accurately detect falls, realize locating, and send alert message to the rescuer who can give timely assistance.Secondly, according to the system solution, this paper has completed hardware and software design of the wearable sensor node, conducts data acquisition experiments, and extracts features from the data. Experiments are designed to acquire the three-axis acceleration and three-axis angular velocity of daily activities and falls. Then the sensor data are extracted through the sliding window technique to be feature set for classification.Then, this paper has optimized the classification algorithm and trains classifiers. DT and RF classifiers are used to classify feature set for training and testing. In order to optimize the classification algorithm, the improved-Relief feature extraction algorithm is proposed. It firstly reduces the dimension of the feature set by selecting and then train and test through the classifiers. The test results show that the optimized classification algorithm improves the sensitivity of detection, achieves high accuracy.Finally, according to the system solution, an alert software on intelligent mobile phone is designed which implements Bluetooth communication, falls detection, location, sending alert messages and other functions. Then experiments are conducted to demonstrate the effectiveness of the fall detection alert system.Fall detection based on wearable sensor is studied in the paper, and the fall detection alert system is designed. The system can effectively detect falls, realize locating, and send alert message to family or rescue center. |