Naïve Bayes Model for Combining Results from Multi-Methods Fault Identification of Power Transformer

This research paper presents a development of a Naive Bayes model to combine the results of multi-methods fault identification in power transformers based on dissolved gas analysis (DGA). The proposed model incorporates four commonly used DGA methods: IEC Ratio Method, Rogers’ Ratio Method, Duval Pentagon Method, and Duval Triangle Method. The model uses the results from each method as input features and predicts the presence of fault types in power transformers. To validate the effectiveness of the proposed model, a dataset containing DGA data with known actual fault from power transformers was used. The results indicate that the simple majority vote increase the accuracy of prediction compared to single method. This approach is easy to implement. However, by using proposed naive bayes model, the increase in accuracy is more pronounced. Overall, the proposed Naïve Bayes model provides a promising approach for combining multi-method fault identification in power transformers based on DGA. The model has the potential to improve the accuracy and reliability of fault diagnosis and facilitate proactive maintenance of power transformers.