Fault identification of power transformers using Proximal Support Vector Machine (PSVM)

The diagnosis of incipient fault is very important for power transformer condition monitoring. The incipient faults are monitored by conventional and artificial intelligence (AI) based models. In this paper, the Proximal Support Vector Machine (PSVM) has been utilized to identify the incipient type of faults in an oil-immersed power transformer. Its performance is compared with traditional IEC/IEEE and AI methods (i.e. ANN and SVM). The juxtaposition of fault classification of ANN and SVM method notify that proposed approach is much swiftly. Simultaneous identification of oil immersed power transformer incipient faults has never been identified formerly by using Multi-PSVM. The desired test analysis of experimental data from working transformers in the Northern Power Grid of India has been executed to present the robustness of evaluated incipient faults for large variation in loading and operational conditions perturbations.