Transformer Winding Vibration Prediction Based on VMD- DTW - Informer

The winding, as a part with frequent faults in power transformers, is an important basis for analyzing its health state. In order to reveal the trend of winding vibration in operation transformers, this paper proposes a winding vibration prediction method based on variational mode decomposition-dynamic time warping (VMD-DTW)-Informer. Firstly, according to the principle of winding vibration generation in transformers, an extraction method of surface vibration features of transformers based on VMD-DTW is designed. Secondly, considering the intrinsic relationship between the operating conditions of transformers and the winding vibration, an Informer transformer winding prediction model is established, and the winding vibration prediction in operation transformers is completed. The experimental results show that the winding vibration extraction method based on VMD-DTW can effectively extract the vibration features of transformer windings, and the Informer transformer winding prediction model considering the operating conditions of transformers can significantly improve the accuracy of vibration prediction, which can provide important reference for the in-operation winding state monitoring technology.