Fractured carbonate reservoirs provide both high—quality reservoir space and efficient fluid flow pathways, serving as a crucial guarantee for achieving high productivity in deep and ultra-deep carbonate formations. However, existing fracture identification methods for carbonate reservoirs suffer from several drawbacks, such as the high cost of electrical imaging logging, susceptibility of conventional logging to interference, and strong subjectivity in manual interpretation. In this paper, a sensitivity analysis approach is adopted to systematically evaluate the sensitivity of conventional logging curves to fracture parameters in carbonate reservoirs of the study area, and a combination of highly sensitive logging curves is optimized. On this basis, the first derivatives and their products are introduced as fracture characteristic parameters, and an intelligent fracture identification model for carbonate reservoirs is constructed using the SVM algorithm. The results show that the model exhibits high accuracy in fracture identification of carbonate reservoirs in the Sichuan Basin. Compared with fracture identification results from FMI electrical imaging logging, the recall rate reaches 90.5% and the precision rate is 95%. The research results can provide methodological reference and technical support for efficient fracture identification in deep carbonate reservoirs in the Sichuan Basin.