Method for detecting defects on machined surface 3D point cloud based on diffusion model, comprising: obtaining 3D point clouds of machined surfaces of defect-free industrial products and obtain defect 3D point clouds, thereby constructing a training set; establishing improved diffusion model based on displacement iterative reconstruction and training the model using training set until loss function converges; obtaining 3D point cloud of machined surface of industrial product to be detected, performing point cloud preprocessing, reconstructing it via reconstruction model; detecting and segmenting 3D point cloud to be detected and its reconstructed point cloud via detection function to obtain defect detection classification and localization results. Present invention is applicable to defect detection based on 3D representations, solve slow inference speed and high video memory consumption in existing 3D defect detection methods, contributes to realization of efficient, rapid, and accurate 3D defect detection and quality inspection of industrial assembly line products.