An accurate and fast building load prediction model is critically important for guiding building energy system design, optimizing operational parameters, and balancing a power grid between energy supply and demand. A physics-based simulation tool is traditionally used to provide the building load demand; however, it is constrained by its complex model development process and requirement for engineering judgments. Machine learning algorithms (i.e., data-driven models) based on big data can bridge this gap. In this study, we used the massive energy data generated by a physics-based tool (EnergyPlus) to develop three data-driven models (i.e., LightGBM, random forest (RF), and long-short term memory (LSTM)) and compared their prediction performances. The physics-based models were developed using office prototype building models as baselines, and ranges were provided for selected key input parameters. Three different input feature dimensions (i.e., six-, nine-, and fifteen-input feature selections) were investigated, aiming to meet different demands for practical applications. We found that LightGBM significantly outperforms the RF and LSTM algorithms, not only with respect to prediction accuracy but also in regard to computation cost. The best prediction results show that the coefficient of variation of the root mean squared error (CVRMSE), squared correction coefficient (R²), and computation time are 5.25%, 0.9959, and 7.0 s for LightGBM, respectively, evidently better than the values for the algorithms based on RF (18.54%, 0.9482, and 44.6 s) and LSTM (22.06%, 0.9267, and 758.8 s). The findings demonstrate that a data-driven model is able to avoid the process of establishing a complicated physics-based model for predicting a building’s thermal load, with similar accuracy to that of a physics-based simulation tool.

A cooling tower is an important guarantee for the proper operation of a solar system. To ensure proper operation of the system and to maintain high-efficiency points, the cooling tower must operate year-round. However, freezing is a common problem that degrades the performance of cooling towers in winter. For example, the air inlet forms hanging ice, which clogs the air path, and the coil in closed cooling towers freezes and cracks, leading to water leakage in the internal circulation. This has become an intractable problem that affects the safety and performance of cooling systems in winter. To address this problem, three methods of freeze protection for cooling towers are studied: (a) the dry and wet mixing operation method—the method of selecting heat exchangers under dry operation at different environments and inlet water temperatures is presented. The numerical experiment shows that the dry and wet mixing operation method can effectively avoid ice hanging on the air inlet. (b) The engineering plastic capillary mats method—its freeze protection characteristics, thermal performance, and economics are studied, and the experiment result is that polyethylene (PE) can meet the demands of freeze protection. (c) The antifreeze fluid method—the cooling capacity of the closed cooling towers with different concentrations of glycol antifreeze fluid is numerically studied by analyzing the heat transfer coefficient ratio, the air volume ratio, the heat dissipation ratio, and the flow rate ratio. The addition of glycol will reduce the cooling capacity of the closed cooling tower.

The filler is the core component of the cooling tower, filler performance refers to both its thermal and flow resistance characteristics, which use empirical formulas of tower characteristic N, volumetric mass transfer coefficient β_xv, and pressure drop ΔP obtained through experimentation under specific conditions. However, the performance equations for identical countercurrent fillers can vary at different heights or seawater concentrations. Linear interpolation is the conventional method for obtaining the filler performance under different conditions, but its uncertainty limits the application. This paper explores the use of the radial basis function artificial neural network (RBF ANN) to analyze filler performance based on existing performance equations. The data set is generated by the filler performance equations. The results demonstrate that RBF ANN has a preferable prediction effect with high correlation (the determination coefficient R² > 0.99) and prediction accuracy (the proportion of relative error within 10% N₁₀ > 90%). Furthermore, the predicted results are consistent with the experimental results of the filler performance. Therefore, RBF ANN can accurately predict filler performance at varying heights and seawater concentrations, making it universal and providing a basis for cooling tower design.