Application of Improved PNN in Transformer Fault Diagnosis

A transformer is an important part of the power system. Existing fault diagnosis methods are still limited by accuracy and efficiency solution excessively rely on manpower. In this paper, a novel neural network designed to overcome issue. Based traditional method judging ratio dissolved gas in internal insulation oil, fast model was built with improved probabilistic (PNN). The particle swarm optimization (PSO) algorithm used find global optimal smoothing factor improve PNN. based PNN not only eliminates influence human subjective factors but also significantly improves speed accuracy, meeting requirements for real-time application practical projects. feasibility effectiveness proposed paper illustrated case study actual data. Through analysis comparison, diagnostic 10% higher than that general BPNN 5% premise ensuring solution.