This paper focuses on the dynamic characteristics of teachers’ professional development in the context of physical education reform, and constructs a career trajectory portrait system based on time series prediction model. A multi-seasonal frequency domain augmented PD-FEformer model is designed to analyze the non-linear evolution law and key turning points of teachers’ professional competence. Combined with cognitive network analysis to reveal the hierarchical features of professional competence evolution, the PD-FEformer model is used to capture the barriers to teachers’ professional competence development. Physical education teachers had the lowest covariance coefficients for A6-A7 (0.04) and A5-A7 (0.05), and the highest covariance coefficient between A3-A1 (0.46), followed by A2-A3 (0.42) and A3-A4 (0.41). The model identified shortcomings in the development of professional competence of physical education teachers, including 18% of physical education teachers who had not participated in open classes, 26% of teachers who had not won any prizes for their papers in the last three years, and as many as 36% of teachers who had no idea at all about professional development.
