Key points are not available for this paper at this time.
Power transformer monitoring is an essential asset management routine for reliable operation of transmission and distribution networks. Over the last two decades, several mathematical and soft computing -based health assessing approaches have been developed based on various power transformer diagnostic tests. However, the tests data correlation analysis and attributes prioritization are not yet adequately established in the current transformer health estimation methods. Therefore, this paper presents a hybrid analytic hierarchy process-based grey relational analysis as a potential multi-criteria decision-making tool for transformer health assessment. The proposed approach facilitates a comparative correlation between transformer samples and attributes prioritization with authenticated hierarchal analysis. The closely relatable attributes are grouped into three distinct grades to ensure transparency. These grades are further correlated using an artificial intelligence fuzzy logic technique. Through this integration, reliable numerical health indices can be obtained to help classify the transformer health condition. Moreover, the general complexity issues observed in forming expert rules with high number of fuzzy logic inputs are completely nullified with reduced three input grades and maximum input-output feasible fuzzy combination. The proposed hybrid model is validated using pre-known health conditions of 730 transformers assessed by Saudi electric company in Gulf region and compared with some transformer health prediction methods published in the literatures. Comparison reveals the superiority of the proposed model that achieved a prediction accuracy more than 96%. Contrary to other techniques published in the literature, the proposed model does not call for extensive training data and can facilitate real-time condition monitoring of power transformers.
Manoj et al. (Tue,) studied this question.