Using Deep Learning Techniques to Study the Effects of Psychological Stress on College Students’ Performance in Physical Education Classes

Abstract With the increasing prominence of the problems of accelerated pace of life and aggravated academic pressure, psychological stress appears in many college students as a group. The article summarises the basic characteristics of PPG signals, implements the extraction of HRV features of psychological pressure in PPG signals, investigates the Transformer model in deep learning, and adapts the encoder in the model and optimises the parameters to be suitable for the analysis of psychological pressure signals. It is confirmed through relevant experiments that the psychological stress feature recognition and extraction method based on the one proposed in this paper has high consistency in the degree of psychological stress and can be used to assess the psychological stress of college students. The regression analysis of college students’ psychological stress and performance in physical education class shows that college students’ psychological stress significantly negatively predicts physical education class exercise β =−0.358 (t=−14.632,p<0.001). In summary, the deep learning technique proposed in the article is important for studying the psychological effects of college students’ stress on their performance in physical education classes.