Aging Characteristics And Lifetime Estimation Of Epoxy Resin Insulation Under High-frequency Electrical-thermal Stress

Power electronic transformer (PET) is a kind of new intelligent power transformer. It can flexiblyrealize the voltage’s transmission and transformation and satisfy significant demand of modern and future power system.The safe and reliable operation of PET is an important guarantee for power grid.The insulation suffers from high frequency electrical and thermal aging stresseswhich lead to the premature failure.The failure mechanism is far more complex than the power-frequency condition. Based on the former research under power-frequency condition, this paper focuses on the aging characteristicsof epoxy resin insulationunder high frequency electro-thermal stress.In order to study the aging of PET insulation, a high frequency high voltage test platwas built,and high frequencycurrentsensor was used to measure partial discharge.Based on the characteristics of measured signal, this paper developed a method to extractpartial discharge signal. A high-pass FIR digital filter was designed to reduce the low-frequency interference. Also, a wavelet thresholding denoising method was used to filter high-frequency interference.Then effective data of discharge was gained by selecting a suitable length of time window.Aging experiments on epoxy resins under electrical-thermal stress was conductedto study PET’s insulation aging process. The partial discharge in each stage was extracted to study the aging characteristics. The experiment results showed that, high-frequency voltage would broaden the distribution of PD phase beyond 90°, the rabbit ears-like and triangle-like form discharge could be easily found. The partial phaseand skewnessat different aging time could be considered as references of insulation’s aging degree.Based on the accelerating tests, the lifespan of the insulation was calculated by Weibull distribution, and the least square fitting was used to analyze its relationship with different aging factors.As the results indicate,the relationship between lifespan and voltage is exponential function model, and between lifespan and frequency and temperature are both power function model.Then through multiple statistic analysismethod considering all of related data, amulti-factors insulation aging lifespantime model was proposed.