Abstract: Reservoir fluid identification plays a crucial indicative role in the exploration and development of carbonate reservoirs, including hydraulic fracturing and sweet spot prediction. Carbonate reservoirs with a set of directional fractures typically exhibit anisotropic characteristics, and the complexity of the medium significantly increases the difficulty of fluid identification. This paper, starting from anisotropy theory, parameterizes carbonate reservoirs as transversely isotropic media containing a set of vertical fractures, with the fracture symmetry axis along the horizontal direction. Based on this, a new anisotropic reflection coefficient equation is derived, including fluid indicator factors, quasi-fracture tangential weakness parameters, and quasi-fracture normal weakness parameters. The accuracy and precision of the equation are compared and analyzed. Simultaneously, within the Bayesian theoretical framework, the expectation-maximization (EM) algorithm is used to identify fluid factors in carbonate reservoirs. Combined with actual well model testing and actual seismic data inversion in the work area, the accuracy and reliability of the proposed carbonate reservoir fluid identification method are verified.