Research on Transformer Fault Diagnosis Based on Fused Sparrow Search Algorithm

The proposed method aims to address problems associated with transformer fault recognition rate and stability by introducing the Chaotic Pareto Sparrow Search Algorithm (CPSSA). The CPSSA approach fused fractional order chaotic mapping and improved lens imaging backward learning strategies to optimize the fault diagnosis model of Support Vector Machine (SVM). These strategies were used to increase sparrow population diversity, convergence speed, and accuracy by reducing the impact of local extremes. CPSSA was employed to optimize kernel function parameters and penalty factors of SVM to create a CPSSA-SVM based model for diagnosing transformer faults. Experimental results demonstrate that the CPSSA-SVM model can identify faults with a high accuracy of 92.36%. These findings highlight the potential of this approach for real-time monitoring of transformer operations and ensuring safe grid operation.