Abstract:  The pilot demonstration area of Well MH1 in the northwest margin of  Junggar basin (MH1 well  area) exhibits fractures of varying scales in the Permian and Triassic formations. The formation mechanism of these fractures is complex, the subsequent stimulation is frequent, and the existing forms are various, which seriously restrict the development process of horizontal wells. Based on improving the signal-to-noise ratio and resolution of seismic data, this paper establishes the small-scale hidden fault identification process and model, and further guides horizontal well drilling through a combination of optimal azimuth superposition scheme, continuous wavelet transform seismic data frequency boosting, matrix singular value decomposition for denoising, iterative processing of structure-guided filtering, fracture enhancement filtering, multi-attribute fusion, and artificial intelligence (AI)-based fault recognition. The results show that by combining various geophysical methods with AI technology, this paper establishes a small-scale concealed fault identification process and model suitable for MH1 well area, significantly improves the fault identification accuracy, optimizes the drilling trajectory of horizontal wells in MH1 well area, successfully avoids high-risk fault zones, reduces the incidence of leakage events by about 20%, and reduces the problem of pressure and channel interference by about 15%.

Key words:  , Junggar basin, fracture prediction, seismic attribute, multi-attribute fusion, AI prediction