In order to study the comfort of tight-fitting sportswear in winter, this paper designed a series of motions, and explored the distribution of comfort perception under different sports conditions by evaluating the wearing perception of human body. Finally, through the acquired experimental data, intelligent prediction models were established, and the prediction results were visualized, which makes the comfort distribution more intuitive. The results show that there are great differences in the parts that affect the overall comfort perception under different sports conditions; different ages subjects have different perceptions of comfort; Particle Swarm Optimization-Cuckoo Search-Adaptive Network-based Fuzzy Inference System has better prediction accuracy, and could replace the wearing trials.