English：

This paper focuses on the application research and practice of generative artificial intelligenceand its key technology（diffusion model） in the field of architectural engineering design. The principle ofimage generation is introduced, including convolutional neural network, generation countermeasurenetwork and diffusion model, and the advantages of diffusion model are mainly described. Through theapplication of three practical engineering cases, the drawing effect of Stable Diffusion and Midjourneysoftware design scheme based on diffusion model is displayed. The research results show that generativeartificial intelligence technology has been able to effectively assist architectural design tasks of differentscales from indoor rooms to large parks. For scenes with relatively simple requirements, the method ofgenerating images based on text can meet the rapid concept expression. For more complex projects, thecombination of fine⁃tuning model training and precise control image technology can achieve high⁃qualitycustomized output.