Transformer Temperature Prediction Method Based on Digital Twin Technology

The transformer is a key link in the power system, and its internal temperature has a decisive effect on the health status of the transformer and various decisions on it. Therefore, it is of great significance to study how to accurately predict the oil temperature of transformers. In this paper, a method for predicting transformer hot spot temperature based on digital twin technology is proposed. First, a transformer twin model is built in the virtual space, and multi-physics field coupling simulations are performed on it under various working conditions, and the transformer oil temperature data is saved as the twin body temperature database. Then combined with algorithms such as the extreme learning machine (ELM) in the neural network, the database data can be learned and the oil temperature can be actively predicted. Finally, the predicted temperature is compared with the actual temperature data to verify the accuracy of this method. The results show that the extreme learning machine algorithm has better prediction accuracy than other algorithms. This hot spot temperature prediction method based on digital twin technology can provide a certain reference value for the stable operation of the power system.