A Deep Research on Stress Classification of Table Tennis Athletes Based on Perceived Stress Scale and EEG

Stress for athletes may influence competition performance, and electroencephalography (EEG) is usually used as one of the approaches to stress classification. This study involves three phases: collection of EEG signals; pre-processing of EEG signals; and analysis of EEG signals. The collection of EEG signals phase uses the perceived stress scale and EEG to collect the brainwave data of table tennis athletes. The pre-processing of EEG signals phase uses the Butterworth filter and fast Fourier transform to process brainwave data. The analysis of EEG signals phase constructs the machine learning and deep learning models to analyze brainwave data. This study constructs the capsule networks (CapsNet) model of the accuracy of 2-level stress and 3-level stress at 96.91% and 96.91%. Finally, the results of model prediction provide information to the coaches and athletes that can prevent and reduce stress in advance.