Benchmarking of Multi-Modal Partial Discharge Sensors and a Cross-Modal Corroboration Framework for Noise Discrimination in Air-Insulated Medium-Voltage Metal-Clad Switchgear.

Reliable partial discharge (PD) assessment of air-insulated medium voltage (MV) metal-clad switchgear is challenged by the close proximity of adjacent units and the noise environment generated by control, relay, and mechanical or electromechanical equipment within the substation. Combined sensing approaches have been employed to identify primary PD sources and distinguish them from interference signals. Although individual sensing technologies have been extensively studied, no systematic framework has been established to corroborate readings across sensing modalities under controlled, reproducible noise conditions. This study presents a laboratory benchmark performed on an actual 15 kV metal-clad switchgear assembly, incorporating representative PD source configurations and interferences encompassing electrical, acoustic, and electromagnetic noise. Sensors spanning multiple detection classes were evaluated, including acoustic sensors (intrusive and remote airborne, and surface-contact types), high-frequency current transformers (HFCTs), transient earth voltage (TEV) sensors, and a wideband electromagnetic sensor, across both open-access and closed-panel configurations. Testing showed that electrical and electromagnetic noise primarily affects conducted-electrical and radiated-EM channels, while acoustic noise degrades acoustic sensors, leaving each channel type largely unaffected by the other's interference. This complementary behavior supports a two-path corroboration criterion that reduces false positives and false negatives and informs practical sensor selection for field deployment.