Dielectric response measurement of oil-paper insulation based on system identification and its time-frequency-domain conversion method

Abstract

Dielectric response testing is a non-destructive insulation testing method that is widely used for assessing insulation properties. Time-and frequency-domain dielectric spectra are closely dependent on insulation conditions. In comparison with the time-domain spectrum, the frequency-domain dielectric response under sweep sinusoidal excitation has a stronger anti-interference capability and a more comprehensive insulation information. However, the wide application of the measurement using the frequency-domain dielectric response under sweep sinusoidal excitation is hindered by its long testing time. A possible solution is obtaining the frequency-domain spectrum by transforming the tested time-domain spectrum, such that the time- and frequency-domain spectra can be obtained in one measurement. Polarization depolarizing current measurement under DC excitation is a common time-domain testing method for time-frequency-domain conversion. However, such conversion is valid only at extremely low frequencies (less than 0.05 Hz), indicating a slight overlap with the frequency range of the measurement using the frequency-domain dielectric response under sweep sinusoidal excitation. Moreover, the effective conversion frequency range cannot be controlled manually, and the measurement using time-domain dielectric response under DC excitation is hindered by the anti-interference capability due to micro polarization-depolarization current. In addition, scholars have found nonlinearity in a partial insulation system. As the bridge of the traditional time-frequency-domain conversion, a linear extended Debye model cannot accurately reflect insulation materials that exhibit nonlinear characteristics, thereby further limiting the application scope of time-frequency-domain conversion. This study introduces the system identification theory to demonstrate the problems existing in the traditional time-frequency-domain conversion and proposes an improved time-domain testing method based on the previous work. The improved method extends the effective frequency range of time-frequency-domain conversion and emphasizes the comprehensive analysis of the insulation by combining the measured time-domain spectrum and the frequency-domain spectrum conversed from time-domain data.