Hybrid artificial intelligence model for reliable decision making in power transformer maintenance through performance index

Abstract

The preventive maintenance of power transformers is essential to ensure their reliability and is supported by efficient predictive techniques and accurate diagnostics. In this context, the objective of this work is to present a hybrid Artificial Intelligence (AI) model for reliable decision making in transformer maintenance based on performance index monitoring. The innovation lies in the application of Monte Carlo filters to monitor the operational state of transformers combined with a novel clustering strategy. The used methodology includes the development of an algorithm for outlier removal in the historical series of each predictive technique as well as the implementation of stochastic filters to forecast the overall operational condition. The results demonstrate the robustness and effectiveness of the developed model. This work contributes a new AI-based strategy for supporting preventive maintenance decisions, enabling precise and individualized actions for each piece of equipment, with broad applicability to companies in the electrical power sector.