Research On The Control Algorithm Of Fall Protection Device And Its Internet Of Things System

Due to the deepening aging of the population and the serious health problems that falls can cause in the elderly,the development of intelligent fall detection systems has received increasing attention.The application of fall protection devices to implement accurate and rapid fall detection and fall protection can greatly reduce fall injuries,which is of great significance to the safety of people who are prone to falls.This paper studies the wearable fall protection device.Based on the analysis of the research status of fall detection technology at home and abroad,a fall detection method based on machine learning is proposed,and a fall detection controller and its Internet of Things system are developed.The main work of this research is as follows:(1)First,the data preprocessing and training data extraction methods in this paper are proposed for the fall detection problem,and then different fall detection models are designed and optimized using the artificial feature method and the convolutional network feature extraction method respectively.Train models using public datasets and evaluate different models from metrics such as accuracy,slack time,algorithm latency,and more.(2)A fall detection device is designed based on STM32 chip,which realizes the collection,data processing and transmission of human motion information.The model trained on the computer side was transplanted to the STM32 chip,and in view of the limited hardware resources of the embedded side,a two-level framework for the machine learning model to run in real time on the embedded side was proposed,which improved the efficiency of the algorithm and saved computing costs.(3)In order to let the guardian know the fall time and location of the faller at the first time,a fall detection Io T system was developed.Based on the three-layer architecture of the Internet of Things,the perception control layer,the network communication layer and the application service layer are designed respectively.Use 4G communication module and MQTT protocol to transmit edge information,develop WEB services based on front-end and back-end separation mode and Springboot framework,and develop mobile APP based on Android.The platform realizes the release of fall alarm information,the persistent storage of fall information and the query of fall status.(4)The fall detection algorithm designed in this study is experimentally tested,and the functional test of the Io T system is carried out,which proves the feasibility and effectiveness of the fall detection system in this paper.