Abstract: Low-grade faults serve as primary pathways for hydrocarbon migration and redistribution, thereby controlling reservoir distribution. However, their identification is challenging due to subtle displacements, small sizes, and weak discontinuous seismic responses that are often obscured by noises. To address this issue, we adopt an identification technique using single-frequency phase to identify low-grade fractures, which comprises four key steps: (1) calculating single-frequency phase volumes via time-frequency transform and optimizing sensitive frequency bands based on seismic signatures; (2) applying edge-preserving filtering to the single-frequency phase volumes within the sensitive bands to enhance fault imaging; (3) performing artificial intelligence-based fault identification by integrating multiple filtered volumes; (4) fusing the resultant attributes for a comprehensive interpretation. The application of this method in Yaha fault zone of the Tarim Basin demonstrates the successful delineation of multi-scale fault distribution in the main target layers. Its exceptional performance in mapping low-grade faults, characterized by robust noise resistance and structural regularity, provides the foundation for future high-precision high-efficiency development of hydrocarbon reservoirs.