This study suggested PV monitoring systems with lower maintenance costs and energy losses, and an accurate and efficient classifier for PV defects using normal and shading condition based on advanced ML techniques like Random Forest, K-Nearest Neighbors, Decision Tree, and SVM are proposed.
Abstract
Solar photovoltaic (PV) technology has become a major sustainable energy option in response to the increasing need for renewable energy. Nations are aiming for less fossil fuels, and promoting global solar PV capacity that was significantly expand to 710 GW in 2020. Some environmental factors which might render solar energy unusable include mud, trees and buildings. Alongside, hotspots, electrical imbalance and the possibility of damage to the PV modules by thermal reasons which reduces power output, especially partial and total shadowing will yield a power loss of 40–50%. These issues can be resolved through the advanced machine learning (ML) techniques which help to detect the defect on PV panel automatically and reduce downtime, and improve energy production reliability. In this work, an accurate and efficient classifier for PV defects using normal and shading condition based on advanced ML techniques like Random Forest (RF), K-Nearest Neighbors (KNN), Decision Tree (DT), Logistic Regression (LG), eXtreme Gradient Boosting (XGBoost) and Support Vector Machines (SVM) is proposed. A dataset including characteristics of voltage, current, power and irradiance is used to test the classification accuracy and the computational efficiency of these algorithms. The results suggested that, RF is the top algorithm with a classification accuracy of 99.7%. KNN, DT, XGBoost, and SVM are next in line with 99% classification accuracy but LR had the lowest performance of 95%. This study suggested PV monitoring systems with lower maintenance costs and energy losses.
This study aimed to develop and implement a cloud-enabled intelligent fault inference system for a photovoltaic installation, using locally acquired electrical and environmental data, preprocessed prior to cloud-based inference.
Claudemiro de Lima Júnior, Mariana da S. M. Sobral, Paulo F. C. Barbosa· Revista Brasileira de Engenh...· 0 citations
A hybrid PV forecasting framework that combines stacking ensemble learning with a targeted residual correction strategy, and demonstrates that analyzing error distribution and forecasting robustness provides valuable insights beyond conventional aggregate metrics, contributing to the development of more reliable photovoltaic forecasting systems.
Khawla Oufrit, A. Mouadili, M. Zazoui· EPJ Web of Conferences· 0 citations
The increasing deployment of solar photovoltaic (PV) systems has intensified the need for intelligent approaches that ensure reliable operation, minimize performance losses, and improve economic viability. Conventional maintenance strategies in solar farms are largely reactive or scheduled, often leading to delayed fault detection, reducing energy yield, and increased operational costs. This study presents a machine learning-based predictive maintenance and optimization framework aimed at enhancing the operational efficiency of a solar photovoltaic power plant, using Azari Farm as a case study. Operational data such as PV array voltage, current, irradiance, temperature, and power output are analyzed to identify performance degradation patterns and detect potential system faults before failure occurs. Machine learning algorithms are applied to monitor system behavior, predict anomalies, and optimize maintenance scheduling. The results demonstrate that integrating machine learning with predictive maintenance strategies significantly improves system reliability, reduces downtime, and enhances overall solar farm efficiency. The study highlights the potential of intelligent monitoring and optimization techniques to improve energy generation performance and sustainability in large scale solar photovoltaic installations.
Oyiogu Dennis, Nwokporo Sunday Celestine· International journal of re...· 0 citations
Accurate prediction of solar panel energy output is vital for managing power systems effectively and maintaining a stable electrical grid. This is especially important in regions that rely heavily on renewable sources. This research provides a direct comparison of five machine learning (ML) algorithms, that is, Decision Tree Regression (DTR), Multiple Linear Regression (MLR), Random Forest (RF), k‐Nearest Neighbors (kNN), and Extreme Gradient Boosting (XGBoost) for forecasting PV power output in Jazan, Saudi Arabia. Jazan has a tropical desert climate with high humidity, seasonal wind speed variation, and coastal proximity, unlike the arid inland regions typically studied in the KSA. This makes it an ideal testbed for evaluating model robustness under varied meteorological parameters. The models were trained on a 6‐year (2017–2022) dataset comprising hourly measurements of four different meteorological parameters and evaluated using coefficient of determination (
R
2
), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the Wilcoxon signed‐rank test. Among the models used, XGBoost achieved the highest accuracy (
R
2
= 0.93, MAE = 7.3, RMSE = 20.38), outperforming all others. The results highlight the effectiveness of ensemble methods in PV power forecasting and offer valuable guidance for improving forecast accuracy in regions with comparable climates.
Abdoalateef Alzhrani, A. Mas’ud, M. K. A. Kamarudin et al.· Energy Science & Enginee...· 0 citations
The nonlinear and dynamic nature of solar energy generation makes effective energy management, operational planning, and grid stability
challenging. Purely numerical and statistical approaches are often unable to pick up the nonlinearity in the photovoltaic power generation. The
present work tests the predictive capability of supervised regression algorithms for photovoltaic energy forecasting. This research is a comparative
analysis of five algorithms, which are Linear Regression, k-Nearest Neighbor, Support Vector Regression, Decision Tree, and Random
Forest. Unlike earlier studies that just targeted irradiation parameters and focused on short-term datasets, this study incorporates a long-term
dataset, consisting of 95,950 hourly observations collected from a rooftop photovoltaic plant found in the Western Australia region from 1990 to
2014. The dataset includes multiple variables that are categorized into solar-irradiance and meteorological variables. Solar-irradiance variables
are Direct-normal, Global-horizontal, and Diffused-horizontal. Meteorological variables are Wet-bulb temperature and Dew-point temperature.
The photovoltaic energy produced is the output variable. The model is trained strategically with hyperparameter tuning performed using a
hold-out k-fold cross-validation. From the first correlation heatmap, it is found that Direct-normal irradiance affects the photovoltaic energy the
most, followed by Global-horizontal, and Diffused-horizontal irradiance. Wet-bulb temperature and the Dew-point temperature are the least
influential parameters. The performance metrics evaluated to assess the developed models are Mean-Squared Error, Root Mean-Squared Error,
Mean Absolute Error, and the Coefficient of Determination. Overall, the ensemble models showed superior performance to the regression-based
models. Ensemble models pick up the non-linearity among the variables well. Compared to other models, Random Forest achieves the highest
value of the Coefficient of Determination. Also, the lowest values of Mean Absolute Error, Mean-Squared Error, and Root Mean-Squared Error.
Compared to the linear regression model, the random forest model has reduced the Mean-Squared Error by 45.51%, Mean Absolute Error by
41.95%, and increased the Coefficient of Determination by 1.34%. The average reduction in error metrics in all the models stays below 38%.
The proposed framework can support utility operators, photovoltaic system engineers, long-term energy planners, and policymakers in efficient
photovoltaic energy monitoring and forecasting to enhance societal energy security.
Bhushan Behede, M. Shahade, Siddharth Chakrabarti et al.· Journal of Thermal Engineeri...· 0 citations
This study investigates the critical impact of temperature in the performance of rooftop solar photovoltaic (PV) systems. Particularly focusing on energy generation efficiency losses when temperatures exceed 28°C. High temperatures increase the internal resistance of PV cells, reducing their ability to convert sunlight into electricity, the research identifies an optimal operating temperature range for solar panels between 24°C and 26°C, where efficiency is maximized. To predict these temperature-dependent performance losses, machine learning models, are employed. A comparation between Linear regression in capturing the complex, non-linear relationships between temperature, irradiance, and energy output. While Linear Regression showed a perfect R2 score of 1.0 indicative of overfitting. The Random Forest model achieved a robust R2 of 0.8987, demonstrating superior generalizability and accuracy of real- world applications. This research validates the effectiveness of machine learning techniques for more reliable solar energy forecasting and optimization and highlights the necessity of considering thermal management in solar system design, especially in hot climates.
Prastika Indriyanti, Nur Dinie Faqihah Jasmi, H. K. Adli et al.· International Journal of Inn...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
