Klasifikasi Sinyal EEG Dengan Power Spectra Density Berbasis Metode Welch Dan MLP Backpropagation

Abstract

Electroencephalogram (EEG) signal is a signal that could become an information for study about disorders of brain function such as Epilepsi. EEG that detected in epileptic seizures produce patterns that allow doctors to distinguish it from normal conditions. However, a visual analysis can not be done continuously. This study proposed a new hybrid method of EEG signal classification using Power Spectral Density (PSD) based on Welch method, Principle Component Analysis (PCA), and Multi Layer Perceptron Backpropagation.There are 3 main stages in this study, firstly preprocessing the dataset of EEG signals by Power Spectral Density (PSD) based on Welch method, then Principle Component Analysis (PCA) as a method of dimensionallity reduction of the EEG signal data and the Multi Layer Perceptron Backpropagation for classifying a signal. Based on experimental results, the proposed method is successfully obtain high accuracy for the 80-20% training-testing partition (99.68%).

Keywords

• classification
• electroencephalogram
• power spectra density
• principle component analysis
• multi layer perceptron backpropagation