Prony Algorithm for Noise Filteration in Partial Discharge Signals Captured by High Frequency Current Transducer

This paper presents application of Prony algorithm for de-nosing of partial discharge (PD) signals captured using a high frequency current transducer (HFCT). A laboratory setup was developed to generate individual and mixed PDs due to corona, surface and cavity type discharges. The generated PD signals were captured by HFCT inductive coupling method. Then Prony algorithm was utilized to de-noise the captured PD signals. PD signals from corona and surface discharges were exponentially varying single spikes but with different phase information indicating the ability of usage of first order Prony filter for denosing. On the contrary, the cavity PD signals were oscillatory shape and effectively get filtered by a third order filter. It was verified that a Prony algorithm based filter designed according to PD shape, can be used as an effective de-noising method for HFCT based PD detection in presence of multiple PD sources.