Dynamic Incipient Fault Forecasting for Power Transformers Using an LSTM Model

Dissolved gas analysis (DGA) is a traditional approach for power transformer fault diagnostics based on the measurement of contamination. Hydrocarbon gases generated and dissolved in oil during operation can increase density as conditions predominate. Critical determination concentration changes assessment trending prediction prevention damage essential. In this article, dynamic proposed using long short-term memory (LSTM) model with intelligent classification to determine running state avoidance potential damage. LSTM processed DGA data collected from real on-site field measurements predicts future concentrations time sequence. Four artificial intelligence (AI) diagnostic models [support vector machine (SVM), ${k}$ -nearest neighbors (KNN), decision tree, neural network (ANN)] were rendered used comparative assessment. By comparing experimental results different LSTM-based models, article asserts that LSTM-KNN provides highest most reliable accuracy transformers.