Time–Frequency Analysis of Acoustic Partial Discharge Signals in a 20 kV Switchgear: A Case Study Using an Acoustic Camera

Abstract

This study presents an acoustic camera-based approach for characterizing partial discharge (PD) signals in a 20 kV switchgear using time-frequency analysis. Acoustic signals were acquired using an acoustic camera and processed through audio extraction, bandpass filtering, and segmentation. Subsequently, Short-Time Fourier Transform (STFT) and Melspectrogram representations were employed to analyze the time-frequency characteristics of the recorded signals. Several features, including Root Mean Square (RMS), Spectral Centroid, Band Energy Ratio (BER), Entropy, Mel Energy, and Mel Entropy, were extracted to characterize the energy and frequency distributions associated with different PD conditions. Experimental measurements were conducted under five operating conditions, namely normal, corona, void, surface, and arc discharges. The results reveal distinct spectral patterns and feature distributions for each discharge type, demonstrating the capability of time-frequency analysis to capture characteristic acoustic signatures of PD activity. The proposed approach provides a systematic framework for acoustic PD characterization and contributes to a better understanding of discharge-related acoustic behavior in medium-voltage switchgear applications.