To gain a more comprehensive understanding of students' affective experiences in designing and teaching artistry and dance in higher education, and to understand the role of metaphorical expression in dance learning. In this paper, we utilize affective computational analysis to capture students' affective experiences from dance instruction, using body movements and facial expressions to identify and categorize students' affective states in dance instruction. A better understanding of students' emotional experiences is achieved by collecting affective data, including affective intensity levels, affective dimensions, and basic features of basic emotional expressions. In terms of emotion feature extraction, Mel frequency cepstrum coefficients, convolutional neural networks and long and short-term memory neural networks are applied to extract the emotional features exhibited by students in dance teaching. Finally, unsupervised classifiers are integrated to construct an emotion computation model to help teachers and students in the field of design and dance teaching in college art studies to better understand, express and apply metaphors. The analysis found that the efficiency of emotion transfer reached 100%, and the difference between the scores of the control group and the experimental group was slightly close to the significance level P<0.10. The construction of emotion is expected to provide useful insights for educational practice and instructional design, and to enhance the effect and experience of teaching Artistic Design and Dance in colleges and universities.