The relationship between the perceptual elements of environmental design and aesthetic emotions was explored, providing a basic framework for the application of generative artificial intelligence in architectural and environmental design. Through questionnaire surveys, data on eight perceptual elements including spatial sense, brightness sense, scale sense, rhythm sense, order sense, variability, colorfulness and naturalness, as well as six emotional indicators including security sense, pleasure sense, attractiveness, interest sense, impression sense and dependence sense were collected. The research using multiple linear regression analysis and multi-layer perceptron regression model found that the sense of order significantly enhanced the sense of security, variability and naturalness improved the sense of pleasure, and rhythm and variability stimulated the sense of interest. These results provide a reference for the optimization of generative artificial intelligence algorithms, and are helpful for designing design schemes that are more emotionally resonant and user-centered.