Hand Motion Recognition And Interaction Based On Surface EMG Signal

Hands play a vital role in human daily life and work.The number of patients with hand disabilities is increasing due to illness or accidents,and the quality of life of patients is severely affected.With the development of society and medical care,people's requirements for the quality of life are increasing,and then the effect of rehabilitation is more precise and the pursuit of perfection.To improve the self-care ability of patients with hand disabilities,medical auxiliary products such as smart artificial hands have gradually become the focus of research,and to some extent,the daily lives of disabled patients have been improved.Accurate recognition of hand movement intentions is one of the key technologies for smart artificial hands,so improving the accuracy and reliability of hand movement recognition has important research value.This article takes hand motion as the research object,conducts in-depth exploration based on sEMG signals for hand motion recognition,and uses a simulated artificial hand as a control object to conduct real-time human-computer interaction control experiments.The research results are expected to improve the intelligence level of rehabilitation equipment,help disabled people improve their life autonomy,and promote the harmonious development of society.The main research contents and contributions of this article are as follows:(1)To deal with the complex low-frequency noise of sEMG signals,this article uses a third-order Butterworth high-pass filter to process sEMG signals,remove low-frequency noise below 20 Hz in sEMG signals,and retain the effective information in sEMG signals.As a result,the signal-noise ratio of the sEMG signal is increased by 14.2% to 12.71 dB,and the denoising effect is obvious.(2)Aiming at the feature extraction of sEMG signals,an improved spectrum map(SPM)feature based on principal component analysis is proposed.Under three classes of support vector machine,K-nearest neighbor and linear discriminant analysis,the recognition accuracy of the improved SPM feature is 81.88%,80.08%,and 81.72%.Compared with the original SPM feature,the running time is reduced by 23.5%,15.4%,and 16.7%.The comprehensive performance of the feature is the best.(3)Aiming at the problem of hand motion recognition based on sEMG signals,a motion recognition method based on dual parallel channel(DPC)convolutional neural network(CNN)is proposed.DPC-CNN re-extracts the effective information of the improved SPM feature from different focusing angles and overcomes the limitation that a single channel cannot fully reflect hand movement intentions.Using the data set to train CNNs with different structures,DPCCNN performs best,and the average recognition accuracy of the test set reaches 93.79%.In the control experiment of a simulated artificial hand,the average recognition accuracy of DPCCNN is 92.75%,the average running time is 263.15 ms,and the real-time control effect of the algorithm is the best.(4)Based on the above research results,this article carries out research on the application of DPC-CNN hand motion recognition method in artificial hand simulation control.The specific implementation process of the simulation experiment is expounded,and the DPC-CNN hand motion recognition method is applied to the simulation of the artificial hand control experimental study.The experimental results further verify the feasibility and effectiveness of the research results in this article.