Artificial intelligence based forecasting and optimization model for concentrated solar power system with thermal energy storage

Power tower concentrated solar power systems integrated with thermal energy storage systems offer promising solutions for reliable and cost-effective energy production. This research applies Artificial Intelligence techniques to enhance the operational efficiency, reliability, and economic performance of a power tower system. A comprehensive real-time data-driven optimization model was developed incorporating an AI-based machine learning technique - Random Forest Regressor combined with grid search cross-validation to accurately predict output power. Furthermore, an interdependent dual-parameter optimization was conducted to optimize critical system parameters, including mirror angles and heat transfer fluid flow rates. The proposed model facilitates energy forecasting, performance optimization, and operational decision-making, as well as economic, weather impact, and sensitivity analysis. Economic feasibility was evaluated using Net Present Value and Levelized Cost of Energy calculations, while sensitivity analysis highlighted the system's resilience to variations in fuel prices, discount rates, and technology cost. The results indicate a highly accurate prediction, with a Mean Squared Error of 0.0676 and an R2 score of 0.9999, featuring the model's robustness. Additionally, a weather impact and correlation analysis was conducted to analyze the system's operational capabilities under varying weather conditions. Moreover an environmental impact assessment illustrated the sustainability advantages of integrating thermal energy storage (TES) with the concentrated solar power (CSP) system, particularly in improving energy dispatch and reducing emissions. Overall, integrating the TES significantly enhanced dispatch capabilities, particularly under varying weather scenarios.