Cognitive Artificial-Intelligence for Rogers Ratio Dissolved Gas Analysis

Abstract

This paper discusses Cognitive Artificial Intelligence (CAI) method for Dissolved Gas Analysis (DGA) interpretation adopting Rogers Ratio method. CAI grows its knowledge through the interaction with its surroundings. Informations are extracted from multiple sources of data and are then fused to obtain new information with Degree of Certainty (DoC). The new information indicates the fault occurred before the observation took place. The proposed method CAI is validated using the IEC TC10 dataset and compared to the conventional Artificial Intelligence Fuzzy Inference System (FIS) and Artificial Neural Network (ANN). Compared to other methods, CAI performs the most accuracy in identifying Low-Energy Discharge (LE) and Thermal-High (TH), while FIS performs the most accuracy in identifying High-Energy Discharge (HE), and ANN performs the most accuracy in identifying Partial Discharge (PD) and Thermal-Low (TL).