Gait Phase Recognition in Multi-Task Scenarios Based on sEMG Signals DOI Creative Commons
Xin Shi,

X. Zhang,

Pengjie Qin

и другие.

Biosensors, Год журнала: 2025, Номер 15(5), С. 305 - 305

Опубликована: Май 10, 2025

In the human-exoskeleton interaction process, accurately recognizing gait phases is crucial for effectively assessing assistance provided by exoskeleton. However, due to similarity in muscle activation patterns between adjacent phases, recognition accuracy often low, which can easily lead confusion surface electromyography (sEMG) feature extraction. This paper proposes a real-time method based on multi-scale fuzzy approximate root mean entropy (MFAREn) and an Efficient Multi-Scale Attention Convolutional Neural Network (EMACNN), building upon concept of entropy. MFAREn used extract dynamic complexity energy intensity features sEMG signals, serving as input matrix EMACNN achieve fast accurate phase recognition. study collected signals from 10 subjects performing continuous lower limb movements five common motion scenarios experimental validation. The results show that proposed achieves average 95.72%, outperforming other comparison methods. this significantly different compared methods (p < 0.001). Notably, walking level walking, stairs ascending, ramp ascending exceeds 95.5%. demonstrates high accuracy, enabling sEMG-based meeting requirements effective interaction.

Язык: Английский

Gait Phase Recognition in Multi-Task Scenarios Based on sEMG Signals DOI Creative Commons
Xin Shi,

X. Zhang,

Pengjie Qin

и другие.

Biosensors, Год журнала: 2025, Номер 15(5), С. 305 - 305

Опубликована: Май 10, 2025

In the human-exoskeleton interaction process, accurately recognizing gait phases is crucial for effectively assessing assistance provided by exoskeleton. However, due to similarity in muscle activation patterns between adjacent phases, recognition accuracy often low, which can easily lead confusion surface electromyography (sEMG) feature extraction. This paper proposes a real-time method based on multi-scale fuzzy approximate root mean entropy (MFAREn) and an Efficient Multi-Scale Attention Convolutional Neural Network (EMACNN), building upon concept of entropy. MFAREn used extract dynamic complexity energy intensity features sEMG signals, serving as input matrix EMACNN achieve fast accurate phase recognition. study collected signals from 10 subjects performing continuous lower limb movements five common motion scenarios experimental validation. The results show that proposed achieves average 95.72%, outperforming other comparison methods. this significantly different compared methods (p < 0.001). Notably, walking level walking, stairs ascending, ramp ascending exceeds 95.5%. demonstrates high accuracy, enabling sEMG-based meeting requirements effective interaction.

Язык: Английский

Процитировано

0