Speaker
Description
Solar water heating systems (SWHS) provide a convenient means of reducing energy consumption for both residential and commercial applications, space conditioning and domestic hot water production. Although physics-based models accurately describe thermal systems’ behavior and performance, they require system parameters and heat transfer mechanisms knowledge, making them complex and computationally demanding, especially under changing climatic and operating conditions. In contrast, data-driven models learn the system's behavior from data and do not require detailed models, thus offering flexibility, a low computational burden and while capturing complex nonlinear behaviors. In this work, a data-driven model for an SWHS is developed which is first emulated using MATLAB/Simulink, and generates 1097 operating points to cover a wide range of environmental and operational conditions. These operating points were varied with respect to the system’s collector efficiency and included solar irradiance, ambient temperature, wind speed, storage tank volume, heat exchanger effectiveness, and storage inlet and outlet temperatures. Second, several models were created to forecast collector efficiency using Multiple Linear Regression, Nonlinear Regression, and machine learning techniques including Support Vector Regression (SVR), Random Forest (RF), Decision Tree (DT), and Extreme Gradient Boosting (XGBoost), in addition to models based on Artificial Neural Networks (ANN). The performance of these methods was analyzed using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and the coefficient of determination (R²). The results indicated the highest forecasting ability for SVR, followed by the ANN model, demonstrating the effectiveness of machine learning techniques to forecast SWH performance accurately.
| Topic Area | Computational Intelligence for Power Systems |
|---|