Physical fitness plays a crucial role in sports competitions, and enhancing physical fitness is one of the key factors in athletes’ training. This study proposes a dynamic capture and simulation model construction method for athletes’ human postures based on sports biomechanics and sports coordinate transformation methods, then collects electromyographic information from athletes’ muscle surfaces during physical training, designs a physical fitness assessment and monitoring system, and optimizes each athlete’s physical fitness training program through the output results of the system. It was verified that the errors between the movement posture angle, residual force and EMG data recognized by the system and the real data were found to be small, which supported the subsequent intervention experiments. The physical training intervention practice found that the athletes in the optimized training group assisted by the intelligent system significantly improved their physical qualities in terms of speed, sensitivity and strength (P<0.05). This paper proves that the use of the intelligent training system in athletes’ physical training is conducive to improving the athletes’ physical fitness level, so as to achieve the effect of overall fitness.