Abstract:  Fractured-vuggy carbonate reservoirs in the Permian Maokou Formation of the central Sichuan Basin are highly heterogeneous. In such reservoirs, conventional logging responses from gas layers, water layers, and gas-water coexisting intervals commonly overlap, making fluid identification non-unique. To reduce this ambiguity, this study develops a fluid identification scheme by coupling principal component analysis (PCA) with Fisher discriminant analysis (FDA). Eleven fluid-sensitive logging variables were used to build the feature matrix. PCA retained the first three principal components, which together accounted for a cumulative contribution rate of 82.6%. These components were interpreted to represent reservoir-space development, combined variations in water-bearing response and conductive characteristics, and fluid occurrence state. The extracted components were then taken as FDA input variables to construct classification functions for gas layers, gas-water coexisting intervals, and water layers. Posterior probabilities were further used to separate determinate discrimination zones from ambiguous ones. Based on 1,209 calibrated samples from 14 appraisal wells, the model yielded an overall identification agreement rate of 92.6%. The agreement rates for gas layers, water layers, and gas-water coexisting intervals were 99.1%, 90.6%, and 72.6%, respectively. Applications to typical wells show that the method gives relatively stable results for pure gas and pure water layers. For gas-water coexisting intervals that may be misclassified by conventional interpretation under a tight-matrix high-resistivity background, it also provides supplementary discrimination evidence. The results can support logging-based fluid evaluation of fractured-vuggy carbonate reservoirs in the Maokou Formation, central Sichuan Basin. For complex intervals located in gas-water transition zones or with strong logging-response overlap, however, well testing and other logging data are still needed for integrated interpretation.