Aiming at the problems of large buried depth, weak magnetic response and insufficient recognition accuracy of concealed coal mines, the key technologies of high-precision magnetic exploration were studied in the deep concealed area of Xiyu Mine Field of Xishan Coalfield, Taiyuan, Shanxi. Based on the layout of graded survey network and the data collected by high-sensitivity magnetometer, combined with daily change correction, IGRF deduction and trend surface separation, the effective stripping of regional field and local anomaly was realized. Multi-parameter enhancement methods including vertical derivative, analytical signal and normalized gradient were introduced, and a comprehensive identification index with anomaly amplitude, gradient change rate and spatial continuity as the core was constructed to quantitatively identify concealed structures. The results showed that multiple primary and secondary faults and fold structures were identified in the study area. The abnormal characteristics had a good correspondence with the structural distribution, and the fault intersection area and local abnormal bodies were effectively described. The research showed that the multi-parameter fusion and comprehensive identification method could significantly improve the identification ability of weak magnetic anomalies, and provide a reliable basis for the analysis of coal seam occurrence and mining layout in complex structural areas.