Fault Diagnosis of Surge Arresters Based on Swarm Optimization Support Vector Neural Network

As the requirements for reliability and safety in power systems continue to rise, accurate assessment of the fault status of critical equipment such as surge arresters becomes increasingly important. Traditional fault assessment methods usually rely on physical inspections or simple data analysis techniques, which often fail to comprehensively reflect the real-time fault status of surge arresters and have limitations in data processing and prediction accuracy. In response to these challenges, a fault status assessment method for surge arresters using swarm optimization support vector neural networks (SOSVNN) is proposed. This method integrates the high accuracy of support vector machine (SVM) with the sequential data processing capabilities of varying parameter recurrent neural networks (VPRNN), and utilizes genetic algorithm (GA) to optimize kernel function parameters and penalty factors, effectively enhancing the accuracy and efficiency of fault status assessments for surge arresters.