Multi-Step Ahead Prediction of Reheat Steam Temperature of a 660 MW Coal-Fired Utility Boiler Using Long Short-Term Memory

Peng Tan,Zhiyuan Jin,Gang Chen,Qingyan Fang,Hengyi Zhu,Cheng Zhang,Ziqian He

doi:10.3389/fenrg.2022.845328

Abstract

With increases in the penetration of renewables in grids, there is an increasing demand for coal-fired power plants to operate flexibly. Regulation of reheat steam temperature is of great importance for the safe and efficient operation of coal-fired power plants. However, the difficulty of reheat steam temperature regulation increases largely during flexible operation due to the large delay and nonlinear properties, especially those units designed to shoulder base load and with limited regulating strategy. A multistep prediction model on the reheat steam temperature of a 660-MW coal-fired utility boiler was developed based on long short-term memory. The results show that the multistep prediction model performs well. The average root mean square error and mean absolute percentage error values of the five-step prediction results are less than 0.52°C and 0.07%, respectively. The correlation coefficients of the five-step predictions are all greater than 0.95. With a sample interval of 30 s, the model provides an accurate prediction of reheat steam temperature within 2.5 min, which could supply an important reference for the reheat steam temperature regulation.

Highlights

As the largest contributor to global greenhouse gas emissions, the energy supply sector needs to make great changes to mitigate the climate change, for example, from traditional fossil fuel-based energy system to renewable energy-based energy system (Kang et al, 2020)
The look-back time step limits the depth of long short-term memory (LSTM) unfolding over time and determines the maximum length of the input data
It should be noted that the look-back time step in LSTM is different from the delay time order in conventional time-series forecasts

Summary

INTRODUCTION

As the largest contributor to global greenhouse gas emissions, the energy supply sector needs to make great changes to mitigate the climate change, for example, from traditional fossil fuel-based energy system to renewable energy-based energy system (Kang et al, 2020). According to the basic knowledge of the coal-fired boiler and the engineers’ suggestions, a total of 11 variables are used as the input of the multistep prediction model These variables include total fuel flow rate, burner tilt position, pressure difference between furnace and wind box, excess air coefficient, SOFA damper opening percentage, secondary air damper opening percentage, feed water flow rate, feed water temperature, superheat desuperheating water opening percentage, reheat desuperheating water opening percentage, and reheat steam temperature. As the five outputs of the model are predictive values of reheat steam temperature at different time steps, the RMSE of the output is in the same standard, and its mean value is significant. After adjusting the model parameters and structure, the cutoff time step is selected to be 390 s; the number of LSTM hidden layer nodes is 256, and the number of LSTM layers is 1

RESULTS AND DISCUSSION

CONCLUSION

DATA AVAILABILITY STATEMENT