Prediction of Dissolved Gas Content in Transformer Oil Based on BWO-VMD-IWTD-GRU Model

Aiming at the difficulty of concentration prediction caused by the intermittency and non-stationarity of dissolved gas concentration in oil in oil-immersed transformers, a gas concentration prediction model based on the combination of variational modal decomposition (VMD) and gated recurrent unit (GRU) is proposed. Firstly, the real-time data singularities are processed by Teager energy operator and so on; then the VMD parameters are optimized by using the beluga whale optimization (BWO), and the data are decomposed by using the BWO-VMD to obtain a set of smoothed components; the components are processed by noise reduction through the improved wavelet thresholding noise reduction (IWTD) algorithm; and finally, the prediction of components is reconstructed by using the GRU. By simulating the prediction of H2 content in oil-immersed transformers and comparing with other algorithms, it is concluded that the proposed prediction model has an average absolute percent error (MAPE) of 1.16%, a root mean square error (RMSE) of 0.47 L/L, and a coefficient of determination R^2 of 0.996, which effectively improves the prediction accuracy. The reliability of the model was proved.