This study examines the explanatory structural model’s inter-level division technique and how it relates to the directed graphs representing the system’s constituent parts. The directed graphs without loops are successfully identified by analyzing these directed graphs, and the critical elements in the set of highest-level elements are found accordingly. The reduced reachability matrix is used in the study to directly locate the corresponding sinks, and this approach not only optimizes the hierarchical effect of the elements of the complex system, but also significantly improves the computational efficiency of the explanatory structural model. We also used this refined model to examine the main variables influencing how well colleges and universities teach physical training. To maximize the effectiveness of the physical exercise teaching method, a strategy of moving from the deep to the surface was chosen, considering the hierarchical link between these components. The study’s findings demonstrated that applying this optimization method raised the subjects’ overall test scores before and after training by an average of 20.14 points. In addition, the subjects’ lung capacity increased from 4.21 liters to 4.86 liters, and the left ventricular end-diastolic volume and output per beat in terms of cardiac function increased by 11.7% and 18.9%, respectively. The study’s findings offer a significant theoretical and practical foundation for advancing physical education techniques and developing college students’ athletic potential.