Accuracy Improvement of Transformer Faults Diagnostic Based on DGA Data Using SVM-BA Classifier

The main objective of the current work was to enhance transformer fault diagnostic accuracy based on dissolved gas analysis (DGA) data with a proposed coupled system support vector machine (SVM)-bat algorithm (BA) and Gaussian classifiers. Six electrical thermal classes were categorized IEC IEEE standard rules. concentration five combustible gases (hydrogen, methane, ethane, ethylene, acetylene) utilized as an input two Two types vectors have been tested; first type considered in ppm, second introduced percentage sum gases. An extensive database 481 had used for training testing phases (321 samples 160 testing). SVM model conditioning parameter “?” penalty margin “C” adjusted through bat develop maximum rate. SVM-BA classifiers’ evaluated compared several DGA techniques literature.