Prediction of Transformer Top Oil Temperature Based on Bayesian Optimization and LSTM Neural Network

As an indispensable power equipment in the power system, power transformers are widely used in various occasions. As an important hub in the system, its operation is related to people's production and life. As one of the influencing factors representing the condition of the transformer, predicting the top oil temperature of the transformer is critical to monitoring the transformer operation. In this paper, by using LSTM network for oil temperature prediction and increasing the hyperparameters in Bayesian optimization neural network, the optimized LSTM network can obtain smaller error ad prediction accuracy compared with the existing prediction strategy, which can perform more effective and fast prediction. Realize the early warning function for abnormal transformer status.