Discrete State Space Fitting Algorithm for Digitally Recorded Impulse Parameter Evaluation

Abstract

Lightning impulse tests are performed on power apparatus to assess its insulation integrity when exposed to over voltage of atmospheric origin. For the evaluation of the test results it is imperative to determine the impulse parameters according to the standards. For the parameter evaluation of the digitally recorded impulse, the eigensystem realization algorithm (ERA) which produces a discrete state space model has been applied to the impulse curve fitting. With appropriate system order, the dynamic state space model is suitable for all types of impulse waveforms. Transforming the identified discrete state space model to the modal coordinates, it is possible to extract the oscillation or overshoot components from the original waveform in analytic function with high accuracy. Depending on the identified oscillation frequency or overshoot duration, the unwanted components can be removed for the construction of the mean curve without significant distortions in the original waveform. This paper demonstrates that ERA is a potentially valuable analytical tool for the evaluation of the impulse parameters.