Accurate forecasting of solar energy is essential for balancing supply and demand, enhancing energy planning, and supporting the integration of renewable resources into modern electricity grids. While recent research has heavily focused on machine learning-based models such as Long Short-Term Memory networks for solar energy forecasting, these approaches often lack transparency and interpretability. This study presents an interpretable by design photovoltaic (PV) forecasting framework that couples hierarchical factor analysis (HFA) with ridge regression. HFA compresses high dimensional meteorology into three physics meaningful second order factors after which a single parameter ridge model provides coefficient level transparency and regularization in this compact space. Using 15 min measurements from a 93.6 kWp plant in Adıyaman, Türkiye (May 17, 2021–Jan 12, 2025), we evaluate under a unified chronological split (0.64/0.16/0.20). The model combines strong generalization with clear insights into how meteorological variables affect solar power generation, ensuring transparency and verifiability. These results highlight regression-based methods as robust, explainable alternatives to complex deep learning models in photovoltaic forecasting.Since development and forecasting using highly multivariate models is typically not an easy task, our approach is designed to provide a more streamlined model through which future prediction is easier. Simplifying complexity and making it easier to understand how parameters affect the result, our proposed model simplifies finding the most important drivers of solar power generation.

Precisely predicting photovoltaic (PV) output is crucial for reliable grid integration; so far, most models rely on site-specific sensor data or treat large meteorological datasets as black boxes. This study proposes an explainable machine-learning framework that simultaneously ranks the most informative weather parameters and reveals their physical relevance to PV generation. Starting from 27 local and plant-level variables recorded at 15 min resolution for a 1 MW array in Çanakkale region, Türkiye (1 August 2022–3 August 2024), we apply a three-stage feature-selection pipeline: (i) variance filtering, (ii) hierarchical correlation clustering with Ward linkage, and (iii) a meta-heuristic optimizer that maximizes a neural-network R² while penalizing poor or redundant inputs. The resulting subset, dominated by apparent temperature and diffuse, direct, global-tilted, and terrestrial irradiance, reduces dimensionality without significantly degrading accuracy. Feature importance is then quantified through two complementary aspects: (a) tree-based permutation scores extracted from a set of ensemble models and (b) information gain computed over random feature combinations. Both views converge on shortwave, direct, and global-tilted irradiance as the primary drivers of active power. Using only the selected features, the best model attains an average R² ≅ 0.91 on unseen data. By utilizing transparent feature-reduction techniques and explainable importance metrics, the proposed approach delivers compact, more generalized, and reliable PV forecasts that generalize to sites lacking embedded sensor networks, and it provides actionable insights for plant siting, sensor prioritization, and grid-operation strategies.

Solar simulators are significant in testing solar cells and photovoltaic (PV) modules since they provide spectrum and optic characteristics close to natural sunlight. This research introduces two fundamental novelties to the literature on light-emitting diode (LED) light source solar simulators. One of these innovations is designing and developing a solar simulator to achieve I-V characterization on a PV module basis instead of cell-based contrary to the previous studies. For this purpose, a unique board with a modular structure is designed on which the LEDs are positioned. In the research, LEDs of at six different wavelengths are used. Spectral match, spatial nonuniformity, and temporal instability performance criteria tests conducted on the illuminated area is performed based on IEC 60904-9 and ASTM 927E-10 standards. Class AAA was obtained according to the large-scale solar simulator standard specified in ASTM 927E-10. The second novelty the study introduces is developing an active cooling system that can prevent an increase in LED junction temperature when the number of LEDs increases in an LED solar simulator with a wide illuminated area. In addition to the fans, this solution is suggested to prevent any damage to the LEDs during the test, as well as high temperatures from affecting the wavelengths and at the same time achieve 25 °C, one of the standard test conditions. Also, the device dimensions are optimized ergonomically for ease of use. With this system designed and devised, it becomes possible to test commercially available PV modules successfully. The active cooling system minimizes the heating problem and the impact of wavelengths that might invalidate performance criteria classes.

The use of electrical energy from renewable energy sources has increased considerably in recent years. Photovoltaic (PV) solar energy, which is one of the renewable energy sources, takes its source from the sun, which is considered an unlimited source. With the widespread use of photovoltaic systems, some legal regulations by governments have been made for their installation. Simulation programs are used for the design, layout, technical and economic analysis of PV systems. In this study, the design was made using the PVsyst, PV*SOL, Solarius programs, and the PVGIS website used in PV systems. The projects are designed to have an output power of 25 kW. According to the simulation results, the PVGIS website provides approximate data, since technical details cannot be entered. In the results of PVGIS and all simulation programs, it has been calculated that the system can produce 42-48 MWh of electrical energy annually. Power Ratio (PR) results of the system were in the range of 78-85%. Annual electrical energy production per installed kW has been calculated to be 1436-1635 kWh/year. It is stated that the annual saved CO2 will be 18.66-22.226 tons /year. When the simulation results are analyzed monthly, the differences between the months are remarkable. The reason for these deviations can be explained by the annual solar radiation (kWh/m2) data and databases used by the programs. It has been observed that the simulation programs consider the monthly average temperatures and sunshine durations of the region. It is thought that there is a need to compare real-time data with simulation results in future studies. It is considered that this study will be helpful for future studies.

Photovoltaic (PV) power generation plays a critical role in the global transition toward sustainable energy systems. However, accurate PV power forecasting remains challenging due to the non-stationary nature of PV power time series and the presence of structural zeros. Most existing studies rely on parametric models or treat zero inflation as a secondary issue. In this study, a hybrid hurdle modeling framework is implemented to explicitly account for both structural zeros and continuous positive power values through a two-part structure. The zero component is modeled using Bayesian logistic regression (BLR), random forest classifier (RFC), and support vector classifier (SVC). The positive values are analyzed using the parametric models Gamma regression (GR), Log-normal regression (LNR), and Weibull regression (WR) and the non-parametric machine learning (ML) techniques random forest regression (RFR), extreme gradient boosting (XGBoost), and support vector regression (SVR). The proposed framework is validated using the target variable active power (AP, kW) based on a real-world data from a 110 kWe (129.6 kWp) PV plant located in Çaycuma, Zonguldak, Türkiye. Four metrics were used to compare the predictive performances of the models: mean squared error (MSE), mean absolute (scaled) error (MAE and MASE), and the coefficient of determination ( $$\:{R}^{2}$$ ). Among the 18 distinct hurdle models, the lowest error values and the highest $$\:{R}^{2}$$ were obtained when the positive component was modeled by either RFR, such as BLR-RFR (MSE = 67.425 kW, MAE = 3.974 kW, MASE = 0.615, $$\:{R}^{2}$$ = 0.899) or XGBoost, such as RFC-XGBoost (MSE = 69.252 kW, MAE = 4.075 kW, MASE = 0.630, $$\:{R}^{2}$$ = 0.898). The findings indicate that hybrid modeling approaches where the zero component is modeled using BLR, RFC, or SVC, and the positive component is modeled with either RFR or XGBoost provide a reliable framework for PV power forecasting, when predicting continuous data characterized by excessive structural zeros.