Partial Discharge Recognition in Real Switchgear Based on Soft Scoring Mechanism and Multi-Mode Fusion

Partial discharge (PD) in switchgear is closely related to its premature insulation degradation, and accurate identification of PD types is an important means to improve its operation condition and to ensure safety of power system. In this paper, four typical PD types are designed in real switchgear: point discharge, suspension potential discharge, strong vertical component along the surface discharge and weak vertical component along the surface discharge. UHF sensor, TEV sensor, AE sensor and HFCT are combined for detection of switchgear PD, and the detection sensitivity of different sensors under the same defect and the spectral characteristics of different defect types are analyzed. Five characteristic parameters of discharge density distribution and correlation coefficient of contour of the extraction spectrum are combined to form a new feature system. A soft scoring method is proposed to identify the PD pattern of switchgear. This method solves the issues associated with traditional recognition methods that rely solely on single spectral characteristics, resulting in low recognition accuracy and inaccurate recognition types. Moreover, it offers a theoretical framework for online monitoring and fault diagnosis of switchgear equipment.