Modeling Research on the Relationship between Physical Training and Vocational Ability Enhancement of College Students in Public Security Colleges and Universities Based on Deep Learning Models

At present, the physical training of public security police has not formed a unified training system in the country, and various places ignore the cultivation of other aspects of the ability to take skill training as the leading role, and solve the problem of how to train through the construction of the system, so as to ensure that the physical training of public security police is carried out effectively.This paper explores the impact of physical training on college students' professionalism in public security colleges, constructs the K nearest neighbor classification algorithm, and introduces the relevant activation function to deal with more complex students' physical training exercise trajectories.ATT-DAN multitarget tracking model is constructed to extract the feature information of college students' physical fitness training, obtain the target movement trajectory, and parameterize the representation of students' physical fitness training programs.The correlation ranges of frequency, average score, highest grade score of physical fitness training and occupational ability were between 0.415~0.632,0.452~0.769,0.412~0.715,respectively, and the credibility and stability of the occupational ability characteristics were good.Meanwhile, the linear regression of the two showed that the correlation Pvalue of age, 30-second deep squat, pull-up, 3200 meters, and 15-second repetitive straddle with occupational ability was less than 0.05, and there was a positive correlation between the two.