The overheating of boiler heating surface seriously affects the safe operation of the power plant. It is of great significance for the safety of boiler to predict the tube temperature of heating surface and to take appropriate preventative measures. A data driven-based model for tube temperature prediction is proposed in this study. Firstly

the key variables affecting the tube temperature are selected by grey correlation analysis that affect the wall temperature of the heating surface

and a wall temperature prediction model based on long short term memory(LSTM) neural network is constructed. Then

the correlation feature coefficients under similar historical operating conditions are defined

and the predicted wall temperature obtained by the LSTM neural network is corrected to improve the model's prediction accuracy. Finally

an on-duty supercritical boiler with 600 MW capacity is taken as the case study. Results showed that the relative error of the proposed prediction model is within(-2.5%

2.5%). The average relative error is 0.40%

and the average tube temperature prediction error is 2.24 ℃. It indicates that the proposed model is helpful for the tube temperature prediction of the boiler under complex operating conditions.