The frequency response analysis (FRA) method, frequently integrated with equivalent circuit models of transformer windings, is used to recognize the physical state of the windings by analyzing frequency response characteristics. Therefore, accurate winding modeling is the foundation for transformer winding fault diagnosis. This study proposes an intelligent gray-box modeling approach for transformer windings. This study introduces an equivalent circuit structure first based on a single-phase ladder network. Subsequently, the state space method is employed to characterize the equivalent circuit. Finally, the FRA curve is then simulated using a circuit solver to validate the proposed method. The simulation results show that the proposed method can intelligently derive the winding parameters of the transformer’s equivalent circuit from the simulated FRA curve, thereby providing critical data for subsequent fault diagnosis.