Partial Discharge estimation in a high voltage transformer using multi signal decomposition technique

A practical technique is introduced to determine the possible high frequency oscillations generated by a partial discharge (PD) source within a high voltage transformer. The approximation of the discharge signal is difficult to achieve because of the fixed distribution components that are present as well as non-stationary components. To solve this, a Hilbert Huang transform is introduced to eliminate the presence of any non-stationary offset levels by employing a multi signal decomposition technique. The discharge signal is decomposed into different signal levels, and a least-square estimation technique is used to determine the harmonic decay equation parameters, that consist of signal gain (A_s), harmonic frequency (omega_s), damping factor (Upsi_s) and harmonic order (s). The estimated parameters are then used to reconstruct the possible original partial discharge signal oscillation using only the harmonic decay equation. Knowledge of the original discharge signal allows evaluation of the severity of the PD event and may assist in condition based maintenance of the transformer.