<p>In the field of competitive sports, the scientific evaluation of the effect of sprinting training has long faced the core challenges of insufficient integration of multi-dimensional indicators and static weight distribution. Traditional methods rely on a single physiological indicator or linear model, making it difficult to quantify the nonlinear synergy of physical fitness, skills and mental ability, especially with significant limitations on the threshold effect of psychological factors and adaptability to dynamic training cycles. To this end, this study developed a fuzzy comprehensive evaluation system based on dynamic coupling of AHP-entropy weight method. By integrating the subjective experience weighting of Analytic Hierarchy Process (AHP) with the objective data drive of entropy weight method, a three-dimensional evaluation system covering nine secondary indicators including vertical jump height, step frequency coefficient of variation and anxiety control score was constructed. The system innovatively introduces the dynamic adjustment coefficient α to achieve adaptive optimization of subjective and objective weights at different stages of the training cycle, and uses the structural equation model to verify the path contribution rate of physical fitness, skill and psychological dimensions to the comprehensive training effect, where the marginal effect of skill parameters increases non linearly with the training intensity. The independent contribution rate of psychological factors in the pre competition period increased significantly to 31%. Empirical research shows that by using membership function quantification indicators to blur boundaries and combining multi-source data fusion techniques, the model reduced the overall prediction error by 47.7%, increased the model fit index (CFI) to 0.93, and improved the sensitivity of the psychological assessment module by 38.2% compared with traditional methods. The dynamic weighting mechanism successfully captured the cross-system association between the blood lactate threshold and psychological resilience. During the weighting transition from the enhancement period to the pre competition period, the skill weighting was dynamically adjusted from 0.46 to 0.53, and the psychological weighting jumped from 0.44 to 0.58, effectively matching the demand for neuromuscular coordination and stress control in the competition environment. The study confirmed that the millisecond-level optimized response of the fuzzy rule base to the coefficient of variation of step size and the time to touch the ground solved the problem of nonlinear modeling of technical parameters. The application of this system breaks through the empirical evaluation paradigm and provides a full-chain solution for sprinting training from data collection, dynamic decision-making to personalized intervention. Its core innovation lies in achieving three breakthroughs: dynamic weight distribution, psychophysiological collaborative quantification, and cross-cycle adaptive optimization.</p>