Signal Processing for Space Charge Measurement Using Laplace Deconvolution

This paper deals with the space charge measurement on full-size HVDC insulation systems using the pulsed electroacoustic (PEA) method. The PEA method gives a sharp electric pulse onto an insulation system and detects its acoustic response due to an interaction between the electric field and space charge. Such a time resolved measurement normally requires the deconvolution process, because the impulse response of the measurement system is not perfectly flat in frequency spectrum. The deconvolution process has conventionally been performed in frequency domain through the Fourier transform. However, in the measurement for a practical full-size insulation system the response lasts for a long time because the impulse is affected by some factors. In such a case, the way that extracting main component by a window function before the deconvolution process may not be appropriate, because it modifies the signal spectrum. As Laplace transform applied to a pulse response in time domain may emphasize the main component of the signal and suppress its tail, the Laplace deconvolution may be a solution to the above problem. In this study we propose to apply the Laplace deconvolution in the PEA method. An XLPE cable with thickness of about 20mm was employed as a specimen. A differential signal was acquired by applying different bias voltages and was employed as the reference signal. A charge and electric field distribution profiles at 10 kV in bias voltage were successfully calculated through numerical Laplace deconvolution. The result was briefly compared with that calculated through the Fourier deconvolution to exhibit the advantage of the Laplace deconvolution.