Research On Electromyographic Sensing And Motion Identification Of The Lower Limb

Surface electromyogram of human lower limb Signal)is a kind of slight bioelectricity signal which is produced by muscle contraction and relaxation when people walk,jump and squat normally,and can be collected by lower limb electromyography sensor.The electric signal in the process is very convenient and harmless to human body.Now it is widely used in medical care,military industry,AI manufacturing industry,etc.The EMG signal of human lower limb muscle is still in the primary stage of development in the field of compliance control system of flexible booster suit,and the corresponding technologies are still not very mature,there is a certain gap from the actual production and application.The purpose of this paper is to hope that the flexible power suit based on sEMG can be popularized and used in sports assistance,to divide the gait of human lower limbs,and to study the sEMG sensing technology of lower limbs.Then,the non local wavelet transform domain algorithm is used to reduce the sEMG signal,and to transport the sEMG signal of eight muscles of human lower limbs The optimized BP neural network is used for motion identification.This paper is divided into the following parts:1)Firstly,the research of the lower limb electromyographic sensing technology is carried out.Through the study and analysis of the human hip,thigh,calf and foot muscle groups in physiological medicine,the movement mechanism of each joint is analyzed when the lower limb is walking,and then the gait of the lower limb is divided.The open Sim Software is used to model and simulate the lower limb muscles of human body,and to compare the two layouts of array and distribution.The layout of the lower limb EMG sensor patch is determined,and then the EMG signals of the lower limb muscles of human body are collected and processed.2)This paper analyzes several common feature extraction methods of sEMG,including iEMG,ZC,VaR and SD.These methods are used to extract the features of SEMG.The non local wavelet transform domain filtering method is used to preprocess the sEMG signal.The experimental results show that the denoising method is effective.3)In this paper,k-nearest neighbor algorithm and support vector machine are used to compare the accuracy of gait phase identification of human lower limbs.Both of them have high identification rate.Then the optimized BP The neural network is used to identify the seven motion modes of the lower limb movement.Through the test,different prediction results are achieved.Finally,the optimized BP neural network has better prediction results in the classification and recognition problem.The recognition rate of the optimized BP neural network is 5.1% higher than that of the traditional BP neural network,and the average recognition rate is 93.7%.