Artificial hand control method based on deep learning

Abstract

The invention discloses an artificial hand control method based on deep learning. The method comprises the steps of selecting an action type of the hand; collecting a motion imagery electroencephalogram, and dividing the motion imagery electroencephalogram into a training sample and a sample to be tested; preprocessing the motion imagery electroencephalogram, wherein low-pass filtering and Laplacespatial filtering are involved; extracting features in the training sample by utilizing wavelet transform, and generating a time-frequency two-dimensional image of the training sample; constructing aconvolutional neural network model with the time-frequency two-dimensional image as input and the motion imagery action type as output, training adjustment parameters, and obtaining a trained model through multi-fold cross validation; extracting features in the sample to be tested by utilizing wavelet transform, generating a time-frequency two-dimensional image and an input model, and obtaining and outputting the corresponding motion imagery action type to serve as a control instruction to control an artificial hand to complete corresponding action. According to the method, common hand actionin life is selected as a classification target and is closer to a natural state, and the method has the advantages of more sufficient information utilization and higher stability and accuracy.

Description

technical field [0001] The invention relates to a method for controlling a prosthetic hand based on deep learning, and belongs to the technical field of brain-computer equipment control. Background technique [0002] Brain-computer interface (Brain-computer interface, BCI) technology takes the brain as the control center, uses the computer to receive brain signals, and after processing and analysis, controls the external equipment to complete the corresponding control instructions. Offered a new way of life. A basic BCI system includes the following four parts: signal acquisition, feature extraction, feature classification and execution of control instructions. BCI technology has great research and application value in the fields of life assistance for the disabled, rehabilitation training for patients with limb injuries, game entertainment and smart home. [0003] The realization of BCI application depends on the good accuracy and robustness of electroencephalogram (Elect...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art, solve the problem of difficulty in extracting motor imagery EEG signal features, and low classification accuracy. The present invention provides a prosthetic hand control method based on deep learning, using deep learning theory, A new means of signal feature extraction and classification in BCI system, based on deep learning to realize natural prosthetic hand control

Images