CEEMDAN fuzzy entropy based fatigue driving detection using single-channel EEG

Yunhe Liu, Zirui Xiang, Zhixin Yan, Jianxiu Jin, Lin Shu, Lulu Zhang, Xiangmin Xu

Abstract: Highlights•Establishing a Single-Channel EEG Dataset for Driving Fatigue: The research involves collecting single-channel EEG data related to driving fatigue, and the model proposed in this paper has demonstrated favorable classification results using our custom dataset.•Incorporation of a Novel Semi-Supervised Approach: A novel semi-supervised approach enhances the collected EEG data. This method involves the generation of pseudo-labels, combining self-training with high-confidence pseudo-labeling, resulting in a significant improvement in model accuracy.•Enhanced Recognition Model: The research incorporates a multi-scale fuzzy entropy algorithm based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN). This approach significantly enhances recognition accuracy by approximately 8 % compared to existing methods. The model is trained on a combination of labeled and pseudo-labeled data, making it more effective for intelligent vehicle systems.