The social environment faced by athletes is constantly changing, and there are various ways for them to contact. Compared with before, today's athletes are more precocious and sensitive to many things in society. The social environment they are faced with is constantly changing, and their ways of contact are various. In order to improve the accuracy of athletes' positive psychological fitness assessment, this paper analyzes and studies athletes' positive psychological fitness based on system engineering method and DM method, and constructs a positive psychological fitness assessment model based on Fuzzy C-clustering (FCM) and Secure Neural Network as examples. The findings indicate that, following numerous iterations, this method outperforms the comparison algorithm in psychological crisis analysis. The error rate has been substantially reduced by 38.55%, while the recall rate has reached an impressive 96.87%, surpassing the comparison algorithm by 16.04%. These results demonstrate the model's capacity for self-learning, enabling online self-diagnosis of athletes' positive psychological disorders. Moreover, it provides valuable support to psychological counseling and the psychological fitness teams in universities. Given the widespread application of big data in the field of psychological fitness, utilizing big data for studying new approaches to pre-alarm psychological fitness crises has emerged as an important research direction.