Process Control Operator EEG Feature Extraction Based on Empirical Mode Decomposition and Spectral Analysis

The aim of this study is to extract salient features from EEG signals that reflect the process control operator functional state. The EEG feature extraction process contains two stages. Firstly, the segmented EEG signals are decomposed into IMFs via empirical mode decomposition. And then Welch´s method for power spectrum estimation is applied to four lower-order IMFs, of which the frequency ranges from 0.5 to 30 Hz. After that the features, including peak frequency, peak power, gravity frequency, absolute power and relative power of the IMFs are calculated. The correlations between features and operator task load, subjective mental workload measurements are analyzed and the features significantly relating to operator functional state are selected.