Method for Recognizing and Classifying Electricians'Hand Movements Based on Improved Transformer Model
WEI Xinran
HE Xinyi
YANG Xiuxiu
TAN Hanzhong
ZHU Li
TAN Jianjun
Abstract：In order to fill the gaps in the application of motion recognition technology of wearable devices in the field of electrician operations,a method for recognizing and classifying electricians' hand movements based on improved Transformer model was proposed.Firstly,this method recognized and classified electricians' hand movements through multi-sensor data fusion,utilizing flex sensors and surface electromyography(sEMG)electrode patches to collect finger bending data and electromyography.Secondly,low-pass filters and time warping techniques were employed for preprocessing to address data interference and instability.Finally,multi-head attention(MHA)mechanism capturing global features and single-head attention(SHA)mechanism enhancing local features were combined to address the issue of insufficient feature depth in MHA and improve the accuracy of motion recognition.The results demonstrated that the recognition accuracy of the improved Transformer model increased by 5.21 percentage points compared to the classic Transformer model,effectively capturing and recognizing common electrician hand movements.This research could provide technical support for virtual reality-based electrician training,and enhance training efficiency and operational safety.
Keywords：motion classificationfinger curvaturesurface electromyographyTransformerelectrician operationsvirtual realitydata fusion
Publication Date：2025-06-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 253-258 )