Error state prediction method of current transformer based on improved VMD-BP neural network

Abstract

In order to improve the fault diagnosis accuracy of voltage transformer, an online diagnosis method of current transformer error state based on improved variational mode decomposition (VMD) and BP neural network is proposed. Firstly, mutual information is used as the condition of stopping VMD iteration, so that the improved VMD algorithm can self-adaptively determine the decomposition number. Then the improved VMD algorithm is used to process the secondary circuit signal of the current transformer, and the IMF components in different scales are obtained. Finally, BP neural network optimized by genetic algorithm is used to predict different IMF components respectively, and the predicted results are superimposed to obtain the final predicted results. By comparing BP and VMD-ARMA models, the results show that the improved VMD-BP neural network model has higher prediction accuracy and more stable results.