Fault Diagnosis and Prediction Method of Distribution Transformer in Active Distribution Network Based on Data Augmentation

As the key hub equipment of the distribution network, the operation status of the distribution transformer is directly related to the operational safety of the whole distribution network. Aiming at the current situation that the faults of traditional operation and maintenance methods are not well targeted, this paper optimizes the probabilistic neural network (PNN) based on the sparrow search algorithm (SSA), and uses variational autoencoder (VAE), non-coding ratio method and principal component analysis (PCA) method to process the fault data in turn to train the fault diagnosis and prediction model of the distribution transformer, which can quickly and efficiently diagnose and predict the operating status of the distribution transformer, so that the maintenance plan can be arranged in time to reduce the power supply shortage and major economic loss caused by the power failure of the distribution transformer.