Evaluation of Teaching Quality of Sports Applied Courses based on PSO-BP Algorithm

Evaluation system for assessing teaching quality in sports applied courses using a hybrid Particle Swarm Optimization-Backpropagation (PSO-BP) algorithm. Traditional evaluation methods often suffer from subjectivity, limited dimensionality, and poor adaptability. To address these challenges, we developed a data-driven framework that integrates PSO's global search capability with BP neural networks' local refinement to optimize teaching quality assessment. The model processes multidimensional inputs, including student feedback (engagement, skill progression), instructor performance (teaching methods, demonstration accuracy), and course structure metrics (curriculum design, resource allocation). Experimental results demonstrate that the PSO-BP model achieves 96.4% accuracy, 95.8% precision, and 96.0% specificity, outperforming conventional methods (Random Forest, SVM, standalone BP) by 15–25% in key metrics. The system's interpretability features reveal actionable insights, such as the correlation between timely feedback and skill mastery rates (+28% improvement). Additionally, the optimized architecture reduces training time by 40% compared to traditional BP networks while enabling real-time evaluation (1,100 FPS). Validated on datasets from three universities, this approach offers a scalable, objective, and adaptive solution for sports pedagogy enhancement, bridging the gap between theoretical evaluation frameworks and practical teaching improvement. The results highlight the potential of hybrid AI models in revolutionizing quality assessment in physical education.