In art and design education, students’ learning paths present diversified characteristics, and there are significant differences in the degree of knowledge mastery. In this paper, a DKVTMN-DTCN knowledge tracking model is constructed, which realizes accurate tracking of students’ knowledge status in art and design education courses by integrating temporal convolutional network and forgetting mechanism. The model adopts a dual-feature processing architecture, using TCN to process temporal feature data and CART decision tree to process nontemporal feature data, and introduces a forgetting time effect mechanism for learning ability differentiation on the basis of the DKVMN baseline model. The experimental results show that on the DPA_2023 dataset, the AUC of the DKVTMN-DTCN model reaches 0.8358 and the ACC reaches 0.9358, which is improved by 2.44% and 0.06%, respectively, compared with the best-performing SPKT method. On the PP_2023 dataset, the model’s recall reaches 0.9921 and F1 score reaches 0.9790, both of which outperform the existing baseline method. The knowledge state analysis shows that the model can effectively capture students’ forgetting behavior in discontinuous learning periods, which is in line with the law of Ebbinghaus’ forgetting curve. This study provides technical support for intelligent curriculum optimization in art and design education, which helps to realize personalized teaching and accurate learning assessment, and promotes the development of the education model in the direction of data-driven intelligence.
