A Classification Method of Surface EMG Signal Based on Capsule Network

Abstract

The invention discloses a surface electromyography signal classification method based on a capsule network, which uses a window analysis method to preprocess the original surface electromyography signal; extracts the features in the signal to form a feature sequence, and uses a two-dimensional method to convert it into a feature matrix; the original signal is simply cut and stacked to generate a two-dimensional signal matrix; the two matrices are convolved through convolution kernels of different sizes, and the channel superposition is performed after obtaining the feature map of the same size; the obtained abstract feature map Send it to the capsule network for training and save the network weights. The invention improves the method for generating abstract features, has high accuracy and strong robustness, especially for actions with similar muscle movements.

Description

technical field [0001] The invention relates to a surface electromyographic signal classification problem. Background technique [0002] In human's daily life and work, hands play an extremely important role. In recent years, with the increase of population aging and the occurrence of various diseases, accidents and natural disasters, the number of elderly people with inconvenient hands and feet and patients with physical disabilities has increased year by year. Intelligent prosthetics can imitate the way of human hands and provide basic assistance in life for the elderly and the physically disabled. Human hand movements are dominated by consciousness, which is transmitted to the whole body in the form of bioelectrical signals, and the corresponding muscle groups execute corresponding actions after receiving the signals. For the control of prosthetics, a signal source is also required, and the most common one is surface electromyography (sEMG). [0003] Surface electromyo...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the current problems, the present invention provides a classification method based on capsule network that fuses original signals and extracted features, adopts a new deep learning model capsule network, and integrates multi-level features in the network, aiming to solve existing Problems in technology, and finally solve the problem of surface electromyography signal classification with higher robustness and accuracy

Images