In recent years, the goal of becoming energy independent of fossil fuels has gained prominence. Solar energy is an excellent choice because it has a very less environmental impact in comparison to fossil fuels. The variability of solar radiation creates a divergence between energy demand and supply, making it necessary to implement efficient thermal energy storage systems to bridge the gap and enable solar thermal power plants to become viable solutions for uninterrupted power generation to meet the present and future energy needs. In this study, various data-driven machine learning (ML) models were used to analyze the design and performance of the packed-bed thermal energy storage (PBTES) system. Six different ML models, including linear regression (LR), support vector regression (SVR), K-nearest neighbors (KNN), decision trees (DT), random forests (RF), and extreme gradient boosting (XGB), were employed to evaluate the performance of PBTES. The models were trained and tested using experimental data reported in the literature on PBTES performance. The results of the study showed that the XGB ML model provided the best performance for PBTES analysis and prediction, with a maximum R2 value of 0.982, and minimum MAE, MAPE, and RMSE values of 0.057, 0.182, and 0.16, respectively. The study demonstrated the effectiveness of data-driven approaches in designing and analyzing PBTES performance, which can be useful for developing more efficient and sustainable energy systems. Overall, this study provides valuable insight into the potential of data-driven ML models for the design and performance analysis of PBTES, which could have significant implications for the utilization of renewable energy sources.
Read full abstract