High-gain DC/DC converters have a promising application prospect in new energy generation and DC microgrids, and good dynamic characteristics are the foundation of their applications. Compared with traditional switching converters, high-gain converters face challenges such as high computational complexity and high model order in modeling. A new modeling method is proposed for high-gain converters based on system identification. It analyzes the working principle of the three-winding Boost-Forward converter and establishes a small-signal model of the converter using the state-space method. The correctness of the established model is verified through simulation. The sources of modeling errors in the state-space method are analyzed. The small-signal model of the converter is initially extracted using the least squares method, with a system model accuracy of 91.43%. Subsequently, an improved Particle Swarm Optimization (PSO) algorithm is employed to accurately extract the small-signal model, achieving a system model identification accuracy of 94.62%. Numerical experimental results demonstrate the correctness of the proposed identification method. The results have high reference value for the modeling and control loop design of complex converters.