Applying fuzzy logic to improve the accuracy of Duval Triangle State Estimation Analysis for DGA based transformer fault classification

Abstract

In the electricity transmission and distribution network, power transformers are one of the most important and expensive equipment. This transformer equipment determines the availability and reliability of the power grid. Availability and reliability of transformer equipment will have a direct impact on the distribution of electricity to consumers so that it also has an impact on the financial sector. One of the most popular methods to keep the transformer from failing is to conduct dissolved gas analysis on transformer oil that is called Dissolved Gas-In-Oil Analysis (DGA). Duval's Triangle State Estimation Analysis is one of the popular DGA based methods. Because the behavior of DGA data is considered not linear, then a non-linear approach to improve the accuracy of this method is required. The purpose of this experiment is to demonstrate whether the non-linear approach based on fuzzy logic is able to provide an improvement in the accuracy of the Duval's method in diagnosing power transformer potential fault types based on DGA data. The result of this experiment shows that in overall the fuzzy logic approach increase the accuracy of 91.88% compared to the 82.90% of the original Duval's method.