Role of AI in Thermal Control for Power Plants: Techniques, Challenges and Opportunities

INTRODUCTION: Power plants require efficient thermal control to reduce operational costs and improve energy performance. Conventional methods may have limited adaptability to changing operating and environmental conditions. AI-based techniques offer a promising solution for improving thermal efficiency and predictive maintenance. OBJECTIVES: This study aims to develop an AI-based framework for optimizing power plant thermal control, improving energy efficiency, and predicting equipment failures using Reinforcement Learning (RL) and Artificial Neural Networks (ANNs). METHODS: The Coal Fired Power Plant Thermal Performance Dataset is used to train the proposed models. RL is applied to optimize boiler temperature and turbine pressure, while ANNs are employed for failure prediction. The Multiverse Optimizer is used to optimize model hyperparameters and improve performance. RESULTS: The proposed framework achieves an R² of 0.9528, MAE of 0.0216, RMSE of 0.0237, and MAPE of 0.023%, demonstrating high prediction accuracy and improved thermal efficiency compared with conventional approaches. CONCLUSION: The proposed AI-driven framework provides an adaptive and efficient approach to power plant thermal management. The integration of RL, ANN, and the Multiverse Optimizer enhances thermal optimization and predictive maintenance, supporting more reliable and efficient power plant operations.