Abstract: Faults, as an important geological structure, are ubiquitous in nature.In 3d modeling of geological body, the difficulty of fault processing lies in obtaining data.Nowadays, under the condition of geological big data, different geological semantic constraints can be used to predict the occurrence of faults.However, as fault structures are often very complex, it is necessary to clarify the morphological classification of faults and their data models in order to use geological big data for constraint modeling of fault structures. This paper studies classification of faults and extracts factors that affect occurrence of faults. These factors are taken as part of fault data model and as mining target in the big data mining of geology. They can constitute table of geological semantic information of faults to provide data support for subsequent fault modeling.On this basis, a new fault modeling idea is proposed, that is, fault is regarded as a whole, and section fault geological semantic information table is classified by Aprirori algorithm to find standard model of each fault. Then, the fault model is constrained by geometric form by using the fault semantic information table to form a fault model with the specific geological semantics. Finally, through smooth transition of the fault traction structure, the fault is embedded in established stratigraphic framework model to complete three-dimensional modeling of the fault.