Research on Temperature Trend Prediction of Hydrogenerator Thrust Bearings Based on ARIMA and Neural Network

The temperature prediction of the hydrogenerator thrust bearing is of great significance to the safe and reliable operation of the unit. A kind of temperature trend prediction based on ARIMA and neural network is proposed. The LSTM neural network is used to predict the thrust bearing temperature. The temperature predicted by the LSTM neural network and the actual temperature difference form a deviation series. The deviation series is modeled and predicted by using ARIMA, and finally the temperature prediction is realized. The example results indicate that the RMSE of the 5-minute prediction of thrust bearing temperature is 0.723°C, which effectively realizes the prediction of thrust bearing temperature trend.