Research On Gait Feature Evaluation Method Based On Internet Of Things Platform

With the continuous development of sensor technology and Internet of things technology,the interaction between human and nature is gradually affected.Relevant scholars at home and abroad apply the Internet of things technology to the field of intelligent medical,and strive to explore a new scheme of remote real-time monitoring of human health information.In recent years,with the deepening of population aging and the increasing rate of disability caused by traffic accidents,the number of patients who need lower limb gait rehabilitation training has increased.Therefore,the combination of Internet of things technology and gait rehabilitation evaluation scheme is of great significance to alleviate the pressure of medical and health resources in China.This thesis proposes a gait feature evaluation method based on the Internet of things platform,which solves the problem that the traditional lower limb walking rehabilitation scheme needs the assistance of medical care personnel or children,improves the utilization rate of medical and health resources,and reduces the cost of rehabilitation training.For the patients in the late stage of gait rehabilitation,the walking aid robot,the surface electromyography signal acquisition system of Internet of things and the gait feature evaluation scheme of Internet of things are developed and designed.Under the condition of ensuring effective walking training,real-time monitoring of health information data,and classification of gait features to achieve gait evaluation.Specifically,this paper has carried out the following research work.Firstly,the surface EMG signal acquisition system of Internet of things is designed.STM32 is used as the device control processor to control the acquisition,preprocessing and transmission of s EMG signal data in the process of walking rehabilitation.GPRS communication protocol is used to transmit the s EMG signals of lower limb muscles collected in the experiment through the 4G communication module on the acquisition device,and the relevant data is stored and processed on the Internet of things platform to realize the real-time monitoring of walking rehabilitation.Secondly,according to the mechanism and characteristics of s EMG signal,two data mining processes,machine learning and deep learning,are proposed.This paper analyzes and compares the algorithms from the dimensions of surface EMG data preprocessing,feature extraction and pattern recognition.Aiming at the machine learning of artificial feature extraction,the time-domain,frequency-domain and time-frequency-domain eigenvalues of surface EMG signals are analyzed and compared.Support vector machine and BP neural network are selected as machine learning classification algorithm,and convolution neural network is compared in gait feature recognition rate.Finally,based on human anatomy and gait cycle theory,the gait rehabilitation evaluation scheme of Internet of things is designed.A large number of s EMG signals of lower limb muscles in walking rehabilitation were obtained through the design and acquisition experiments,which provided data samples for this research topic.Through the construction of classical machine learning and convolution neural network model,the algorithm model was trained by using the collected gait feature data.The experimental results show that the convolutional neural network model trained by a large number of data samples is superior to other algorithm models in classification accuracy,robustness and parameter adjustment complexity,and can achieve 91.5% recognition accuracy.