Sedimentary microfacies analysis, as an indispensable part of oil and gas reservoir research, has important theoretical and practical significance. This paper mainly studies the intelligent design system of petroleum sedimentology model based on artificial intelligence algorithm. This paper first analyzes the basic concept of sedimentary microfacies, which lays a foundation for the subsequent research. In view of the multi-scale nature of well logging, the following improvements are made on the basis of U-net network: removing the pooling layer to reduce the loss of spatial characteristic information; Multi-scale convolution block is introduced to realize multi-scale mining. By adding a one-dimensional convolution layer and realizing single-direction segmentation, a well logging sedimentary microfacies identification model IU-net with multi-scale characteristic constraints is constructed. Finally, an intelligent system for sedimentary microfacies analysis and management is constructed and tested.