Multifactor Aging Analysis and Lifetime Prediction Modeling of Polyimide Insulation under High-Frequency Electro-Thermal Stress

In solid-state transformers (SSTs), polyimide (PI) insulation operates under coupled kHz-level sinusoidal electrical stress and elevated ambient temperature, which can accelerate progressive degradation and promote partial-discharge (PD)-assisted failures. However, aging mechanisms and life assessment under high-frequency electro-thermal stresses remain insufficiently quantified. Here, a central composite design (CCD)-based response surface methodology (RSM) is combined with dielectric spectroscopy and finite-element (FE) electric-field modeling to systematically evaluate the coupled effects of voltage amplitude (2-4 kV), frequency (10–30 kHz), and ambient temperature ( 20-100 °C) on PD behavior and aging lifetime of PI films. The results show that voltage amplitude and frequency dominate PD severity and lifetime reduction, while temperature mainly acts as an accelerating factor. The number of PD pulses per cycle exhibits a non-monotonic dependence on frequency. Dielectric spectroscopy reveals a conduction-to-polarization loss transition within the kHz band, with dielectric loss increasing markedly with voltage amplitude and temperature. FE analysis confirms strong interfacial electric-field concentration and introduces the corresponding electric energy density ūeas an electrical stress descriptor representative of field concentration, incorporating both electric-field enhancement and dielectric properties. A multivariable lifetime model is then formulated within a thermally activated framework by coupling a stored-energy-driven barrier-lowering bias (ūe) with a frequency-dependent dissipation term (ωtanδ). The model reproduces the measured lifetimes withR2=0.9829 andMAPE=7.19%, and maintains stable accuracy under extended conditions (maximum error 7.60 %). Collinearity diagnostics confirm statistically distinguishable contributions of the two electrical drivers (r=0.316,VIF=1.11). The proposed methodology provides a mechanism-linked tool for insulation design and condition assessment in high-frequency power-conversion equipment.