<p>Addressing the inaccurate model predictions in high-frequency wireless power transfer systems caused by the electro-magneto-thermal coupling effect, this paper proposes a semi-analytical coupled model based on the partial element equivalent circuit method. This model is tailored to characterize the nonlinear behavior of ferrite cores under high-frequency, non-uniform magnetic flux excitation. The proposed model enhances the accuracy of heat source calculation by integrating a loss separation model that accounts for the skin effect. Furthermore, it introduces an asymmetric Fano model to precisely quantify the dual dependence of the core material’s magnetic permeability on both frequency and temperature. Through an iterative solution of the coupled electromagnetic and temperature fields, the model’s predictions for system self-inductance, mutual inductance, and peak temperature demonstrate high fidelity with experimental and simulation results, with all errors contained within 4.3%. Compared to a foundational model that neglects coupling effects, the proposed model achieves a prediction accuracy improvement of over 13.53%. This work provides an analytical tool for the optimal design of high-frequency WPT systems, offering a balance between physical accuracy and computational efficiency.</p>