Por favor, use este identificador para citar o enlazar este ítem:
http://inaoe.repositorioinstitucional.mx/jspui/handle/1009/324| EEG signal processing based on stochastic resonance | |
| LUCIO FIDEL REBOLLEDO HERRERA | |
| GUILLERMO ESPINOSA FLORES VERDAD | |
| Acceso Abierto | |
| Atribución-NoComercial-SinDerivadas | |
| Stochastic resonance Weak signal detection Duffing system Nonlinear Stochastic dynamics | |
| In this dissertation the phenomenon of stochastic resonance was investigated, i.e. the optimal noise-induced increase of a dynamical system's sensitivity and its ability to amplify small periodic signals. We have considered bistable stochastic systems fed with a small periodic signal (two-state Markov chains) with discrete and continuous time and diffusions in double-well potentials. In the Markov chain case, one-step or infinitesimal probabilities are periodically modulated. In the diffusion case, depths of the potential wells are periodically changed in time. We introduce several measures of goodness for tuning, the most important one of which is the coeficient signal-to-noise ratio and harmonic distortion (SINAD), which describes the spectral content carried by the averaged random output corresponding to the frequency of a small deterministic periodic perturbation. Quartic double well (QDW) modulation was performed by moving the wells separation, depth and friction, supposing a particle forced by stochastic Brownian movement described by the modified Duffing system forced by white Gaussian noise. This approach leads to Langevine equations definition, analyzed by its corresponding Fokker-Planck equation and sustained by two state theory, to acquire the transition rate between wells for Stochastic Resonance (SR) induction. Also, QDW modulation will be shown to recover the input signals by pushing the Duffing system into pseudo-separatrix states and, thus, forcing a limit cycle by the stochastic force. The proposed methodology was algorithmically implemented based on Runge-Kutta integration and tested with input signals under -10dB. Discrete wavelet transform DWT and autocorrelation were compared with the modulated QDW methodology with interesting results. Moreover, Electroencephalogram (EEG) signals where processed under our methodology, showing enhanced results compared with other methods. | |
| Instituto Nacional de Astrofísica, Óptica y Electrónica | |
| 2017-01 | |
| Tesis de doctorado | |
| Inglés | |
| Estudiantes Investigadores Público en general | |
| Rebolledo-Herrera L.F. | |
| ELECTRÓNICA | |
| Versión aceptada | |
| acceptedVersion - Versión aceptada | |
| Aparece en las colecciones: | Doctorado en Electrónica |
Cargar archivos:
| Fichero | Descripción | Tamaño | Formato | |
|---|---|---|---|---|
| RebolledoHeLF.pdf | 12.28 MB | Adobe PDF | Visualizar/Abrir |