Research on Classification of Driving Fatigue Brain Regions Under Human Machine Co-Driving Based on Time-Frequency Domain EEG Features

Regarding the issue of driving fatigue in human-machine shared control environments, a brain region classification method based on multi-feature fusion and machine learning algorithms has been proposed. By extracting temporal features (mean, variance, skewness, kurtosis, root mean square, and Hiorth parameters) and frequency domain features (power spectral density) from the driver's electroencephalogram (EEG) signals, classification models for multiple brain regions, including the frontal pole region, frontal region, central region, parietal region, and occipital region, have been constructed. Commonly used machine learning algorithms such as Support Vector Machine (SVM), Naive Bayes, Backpropagation Neural Network (BPNN), K-Nearest Neighbors (KNN), and Logistic Regression have been utilized to conduct binary classification discussions on the states of driving fatigue and wakefulness for each brain region. The results indicate that the prefrontal lobe region exhibits the highest classification accuracy when using the BPNN algorithm, reaching up to 95%. This research not only provides a new method for real-time monitoring and early warning of driving fatigue, but also offers valuable insights into the brain activity patterns of drivers in human-machine shared control environments.