2019 21st International Middle East Power Systems Conference (MEPCON) | 2019
A Decision Transformer Fault Diagnostics System Based on Dissolved Gas Analysis
Abstract
Dissolved gas analysis (DGA) is considered one of the simplest and most widely used tools to diagnose incipient faults of power transformers. Several methods have been proposed for DGA in power transformers, such as Roger’s ratio method, IEC method and Duval Triangle method. However, these methods suffer from the low accuracy or no decision problems. Accordingly, in this paper, a combined method is designed and built using MATLAB/SIMULINK to construct enhanced DGA diagnostics of transformer faults. This method will be referred as Decision Transformer Fault Diagnostics (DTFD) method. DTFD method is built based on the conventional methods, but, with a combined framework in order to give an enhanced accuracy. First, each individual method is built separately with an output representing the fault in a matrix form. Then, the outputs of all individual methods are used as the input to DTFD system. Through this method, the fault states of each matrix are compared giving a combined matrix. The fault state that has the highest number within the matrix represents the dominant fault. 386 dataset samples are used to validate the proposed method. A comparison of DTFD method with conventional ones indicated that the proposed method has the highest classification.