Caracterización de señales electromiográficas para la clasificación de cuatro movimientos de la mano empleando técnicas temporales y frecuenciales
Manipulating bionic prosthesis through electromyographic signals captured in humans or animals, it requires the detection of the existing correlation between the temporal characteristics and/or frequency of the electromyographic (EMG) signals captured with the type of movement executed by the muscle...
| Autor Principal: | Cruz Salazar, Mónica Andrea |
|---|---|
| Otros Autores: | Aguilar Zapata, Olga María, Acosta Calle, Steven |
| Formato: | info:eu-repo/semantics/bachelorThesis |
| Idioma: | spa |
| Publicado: |
Ingenierias
2017
|
| Materias: | |
| Acceso en línea: |
http://hdl.handle.net/10819/4129 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: |
Manipulating bionic prosthesis through electromyographic signals captured in humans or animals, it requires the detection of the existing correlation between the temporal characteristics and/or frequency of the electromyographic (EMG) signals captured with the type of movement executed by the muscle or muscles under study. This features or signal parameters are measured in different time intervals, generating a feature vector V = [V1, V2,...,VN]ERN, , where each parameter Vi ER is measured in the i- tenth time signal. Literature has proposed various techniques to characterized EMG signals that have shown to be very effective in detecting the type or types of movements performed, such as the coefficient technique of the autoregressive model (AR), the Zero Crossing, the temporal behavior of the mean and variance, the Short Time Fourier transform, the wavelet transform signal and the energy behavior in time among others. In this paper we will present the development of some temporary - Frequency techniques for the characteristics extraction of the EMG signals for 4 types of movements performed by a hand (arm supination - pronation, fingers Extension - Flexion) and application of principal component analysis (PCA) technique for the dimensions reduction of the characteristics vectors, in order to use the obtained features for each type of movement in a neural network classifier designed by the work team, which will autonomously discriminate the types of movement from the measurement of the EMG signals of the muscles under consideration.
Finally, the experimental measurement of the matrix for the probability of success of the different techniques for characteristics extraction will be performed + Neural Network, to select the technique with greater success rate to be used at a later stage of the project for the online recognition of hand movements. |
|---|