Research on Partial Discharge Ultra-High Frequency Time-Domain Signal Separation Algorithm Based on SSA-PCA-IF

With the increasing demand for reliability in power systems, the condition monitoring and fault diagnosis of electrical equipment have become increasingly important. Partial Discharge (PD) is an early indicator of insulation failure in electrical equipment, and the accurate separation of multi-source PD signals is crucial for fault diagnosis. This paper proposes a method for separating partial discharge signals in the ultra-high frequency (UHF) time domain. The method first utilizes Singular Spectrum Analysis (SSA) to extract feature parameters from the PD signals, then applies Principal Component Analysis (PCA) for dimensionality reduction, and finally employs the Isolation Forest (IF) algorithm to cluster and separate the reduced data. The results of processing experimental data with this method demonstrate its effectiveness in separating multi-source PD signals with a high degree of accuracy, which holds significant practical value for the condition monitoring and fault diagnosis of electrical equipment.