Prediction Model of Dissolved Gas in Transformer Oil Based on EMD and BiGRU

Abstract

Oil-immersed transformer is one of the most important pieces of equipment in the power grid. Ensuring the safe and stable operation of transformer is of great significance to the power system reliability. Dissolved gas analysis is an effective method to detect defects and potential faults of oil-immersed transformers, making accurate prediction for dissolved gases in transformer oilcan give foundation to transformer early warning and predictive maintenance, but the non-stationary and nonlinear variation of dissolved gas concentration limits the accuracy of common prediction methods. Artificial intelligence is an effective means for prediction. In order to improve the accuracy of prediction, this paper proposed a prediction model of dissolved gas in transformer, which combines EMD (empirical mode decomposition) and BiGRU (bidirectional gate recurring unit). EMD is selected to stabilize the concentration series of dissolved gas, and BiRGU is used to predict the sub-sequence components. Finally, the prediction results are obtained by superposition and reconstruction. Compared with traditional models, this method effectively reduces the influence of the nonstationarity of the original data, and obviously improves the prediction accuracy, which is helpful to prolong the service life of equipment and improve the reliability of power grid.