Diagnosis of Loose Transformer Winding Faults in Pumped Storage Power Station Based on NGO-VMD and LSSVM

In order to more effectively diagnose and accurately classify winding loosening faults in pumped storage power station transformers, a method based on Northern Goshawk algorithm optimized variational mode decomposition (NGO-VMD) and least squares vector machine (LSSVM) is proposed. Diagnostic method for loose windings of pumped storage power station transformers. First, the vibration signals of the pumped storage power station transformer under different working conditions are decomposed using Variational Mode Decomposition Optimized by Northern Goshawk Algorithm (SCNGO-VMD), and the effective IMF components are screened based on the Pearson correlation coefficient index; then the sample entropy and approximate entropy of the Intrinsic Mode Function (IMF) component are calculated and selected as features; finally, introduce the least squares support vector machine to establish a classification model and input feature vectors to achieve fault diagnosis and classification. The results show that the decomposition method used can be used for fault diagnosis of pumped storage power station transformers, and has better diagnostic effect and accuracy than other methods.