Identification of EMG activity with machine learning in patients with amputation of upper limbs for the development of mechanical prostheses

Expotecnología 2018 Research, Innovation and Development in Engineering

工业技术（总论）

Zuleta, J.N.^1 ; Ferro, M.^1 ; Murillo, C.^1 ; Franco-Luna, R.A.^1^2

Tecnoacademia Risaralda, Servicio Nacional de Aprendizaje SENA, Carrera 21 con 73 bis, Dosquebradas, Colombia^1

Facultad de Ingenieria, Universidad Tecnológica de Pereira, Carrera 27 # 10-02, Pereira, Colombia^2

关键词: Acceptable performance;    Accuracy percentages;    Classification system;    Descriptors;    Electromyographic signal;    Hand pose;    Machine learning techniques;    Upper limbs;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/519/1/012010/pdf
DOI  :  10.1088/1757-899X/519/1/012010

学科分类：工业工程学

来源: IOP

【 摘 要 】

A study of electromyographic signals (EMG) in subjects with partial hand amputation using machine learning techniques (ML) is presented in this document. The EMG were analyzed for five hand poses. We used the Fast Fourier Transform (FFT), and Wavelet transform as descriptors for the feature extraction, the identification and classification system was implemented based on Vector Support Machines (VSM). Percentages of accuracy greater than 90% were obtained in the cases of close hand, left hand, right hand and relax hand, while open hand obtained an acceptable performance with accuracy percentages lower than 90%.