A hybrid kernel principal component analysis and long short-term memory approach for EEG signal-based fatigue evaluation

Brain fatigue will have a negative effect on shooting accuracy, mental state, long-term training effect and sports career. Electroencephalogram (EEG) signals, which reflect the electrical activity of the brain, are promising for fatigue detection. In this study, we combined kernel principal component analysis (KPCA) with long short-term memory (LSTM) for fatigue detection based on EEG data. KPCA extracts nonlinear features from high-dimensional EEG data, maps them to a high-dimensional space, finds hidden patterns and reduces dimensionality while retaining key information. These features then enter an LSTM network, which is good at processing sequential data and capturing temporal dependencies. Over time, it can learn the changes in EEG features associated with fatigue, enabling accurate predictions. Experiments on a dataset validate the effectiveness of our approach and produce high accuracy. This study provides a new approach for fatigue detection based on EEG, which achieved 97.1 % accuracy in fatigue detection. For shooters and other high-attention sports, EEG fatigue detection is of great significance. It can detect fatigue in time by monitoring brain wave changes, help athletes make reasonable adjustments in daily training, make accurate decisions in competition, and promote long-term training optimization to ensure athletes' performance and growth in an all-round way.