With the rapid development of new electric power systems

large-scale integration of renewable energy sources such as photovoltaic and wind power via power electronic converters has led to increased harmonic content in the grid. A high proportion of harmonics significantly exacerbates capacitor losses

accelerates dielectric aging

and shortens service life. To address the challenge of accurately evaluating capacitor losses under complex harmonic conditions

this study develops a loss model considering nonlinear electric field effects and proposes a small-sample parameter fitting method guided by physical model. Firstly

the loss components under harmonic excitation are analyzed

and physical models are established for conductor and dielectric losses. Secondly

experimental data on polypropylene dielectric under varying field strengths are used to analyze the thermionic polarization mechanism via a double-well potential model

and to establish a capacitor loss model after taking into account the nonlinear effects of field strength. Besides

a hybrid optimization strategy combining differential evolution (DE) and L-BFGS-B is employed to improve parameter identification stability and accuracy. Finally

experimental validation shows that the proposed model achieves a maximum error below 0.11W and a relative error under 5% under multi-harmonic conditions. The method outperforms conventional empirical models in accuracy and generalization

and remains effective under limited sample conditions

providing a reliable basis for capacitor loss and thermal evaluation in new electric power systems.