A novel hybrid intelligent classification system for PV fault detection is proposed by integrating Fuzzy C-Means (FCM) clustering and Deep Learning (DL) techniques such as Multi-Layer Perceptron (MLP), Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM) and Gated Recurrent Unit (GRU).
Abstract
Photovoltaic (PV) systems have proven themselves to be a viable alternative energy source; however, there are multiple faults related to PV systems which cause energy losses and low efficiencies. Manual or rule-based algorithms are traditionally used for fault diagnosis, which are not efficient and unsuitable for real-time applications. In this paper, a novel hybrid intelligent classification system for PV fault detection is proposed by integrating Fuzzy C-Means (FCM) clustering and Deep Learning (DL) techniques such as Multi-Layer Perceptron (MLP), Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM) and Gated Recurrent Unit (GRU). The dataset consists of 102,400 samples collected from a real-time 10 kW solar PV system operating under varying irradiance conditions ranging from 600 W/m2 to 1,000 W/m2 and temperature conditions ranging from 25 °C to 40 °C. The FCM technique is used to enhance the extracted features by clustering the membership functions, and the obtained features are used for model training. The performance of proposed models is evaluated using the classification metrics and confusion matrices. The proposed FCM + GRU model achieved 91.13% accuracy 0.79 precision, 0.76 recall, and F1-score of 0.78. The obtained results confirm the effectiveness of the proposed hybrid framework by improving fault classification performance under various operating environments.
The efficient and safe operation of Solar Power Plants (SPP) depends on the autonomous detection of photovoltaic (PV) faults. In this thesis, an original multimodal Decision Support System (DSS) based on electrical analytics and computer vision was designed by processing multidimensional chronological sensor data obtained from a 10 kWp domestic PV system installed in the Van/Başkale region under varying seasonal conditions and operational scenarios.
In the first stage, the fault detection performances of Machine Learning (ML) and Deep Learning (DL) architectures were compared on sensor data. A lightweight ML layer was constructed through statistical residual analysis, utilizing the Isolation Forest algorithm for unsupervised pre-filtering and the HistGradientBoosting (HGB) model for supervised power estimation. This infrastructure was benchmarked against a 1D-CNN-LSTM Hybrid Autoencoder architecture, which hierarchically learns temporal and spatial features, based on accuracy, F1-Score, and computational cost criteria. To clarify the "black box" nature of the deep learning model, an Explainable AI (XAI) layer was established by integrating the SHAP algorithm into the system.
In the second stage, to autonomously verify the physical causes of electrical anomalies (such as micro-cracks, fractures, soiling, and shading), panel images collected from the field were labeled on Roboflow and trained using the Transformer-based RF-DETR object detection architecture. Physical deformations were successfully mapped with a confidence score exceeding 95%.
In the final stage, all analytical and visual models were integrated into a unified user interface using the Antigravity infrastructure. The developed DSS fuses two different data sources to provide the operator with instantaneous "Predicted Percentage of Efficiency Loss (%)" and action-oriented "Maintenance Recommendations." Empirical findings demonstrate that the proposed DSS infrastructure significantly enhances predictive maintenance reliability in domestic SPP projects while substantially reducing operational downtime and technical maintenance costs.
Abstract. The explosive growth of grid-connected renewable energy systems (RES) has increased the complexity of the operation of the modern power infrastructure, making the detection of the faults reliably an inevitable condition of the stable functioning and safety. Traditional single-algorithm and rule-based monitoring systems are not sufficiently flexible or discriminatory to distinguish between the many varieties of faults that occur in photovoltaic (PV) arrays, wind turbines, battery management systems, and grid-tie inverters. The paper suggests a new stacking-ensemble hybrid machine learning model that will combine a one-dimensional convolutional neural network with a bidirectional long short-term memory (CNN-BiLSTM) module, a Random Forest classifier, and an XGBoost gradient booster as base learners under the guidance of a logistic regression meta-learner. Publicly available SCADA and lab bench data were used to create a curated multi-source dataset of 8,400 labelled samples to represent five operational states. The proposed framework achieved an accuracy of 97.8, a macro-averaged F1-score of 97.1, and a Matthews Correlation Coefficient (MCC) of 0.972, outperforming all the compared baseline methods at least by 3.3 percentage points. The findings verify the effectiveness of the hybrid stacking paradigm in identifying faults in real-time and multiple classes in heterogeneous renewable energy settings.
A. Gopalakrushna· Materials Research Proceedin...· 0 citations
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.
Aafaque Ali, M. A. Raza, M. Altayeb et al.· Discover Sustainability· 0 citations
ABSTRACT Photovoltaic solar energy plays an important role in renewable energy, but its performance can be affected by operational issues such as partial shading and soiling accumulation on the modules. In this context, intelligent monitoring strategies are essential for identifying faults and assessing performance. 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. The model was trained using machine learning and artificial neural network techniques with real experimental data collected at the photovoltaic plant of the Laboratory of Physics and Renewable Energy at the University of Pernambuco, Petrolina Campus. The neural network predictions were integrated into a web-based platform for data visualization, parameter analysis, and automated operational alerts. The system showed high precision in identifying the evaluated conditions, with Mean Square Error on the order of 10⁻2, Mean Absolute Error between 5.33 and 5.40%, and coefficients of determination (R2) ranging from 99.74 to 99.76%. These results indicate the potential of the proposed approach to support monitoring and predictive maintenance of photovoltaic systems.
Claudemiro de Lima Júnior, Mariana da S. M. Sobral, Paulo F. C. Barbosa· Revista Brasileira de Engenh...· 0 citations
Nowadays, a big photovoltaic (PV) farm is operating to use solar energy as a source of electricity. Finding and estimating electrical problems on these farms is crucial to ensure the system is reliable, extract the maximum energy from it, and minimize maintenance costs. Machine learning algorithms are the tools that enable the detection of faults in the panel, thereby minimizing downtime. However, changes in PV technologies or environmental conditions make model use difficult because models must be updated frequently to be accurate. This paper presents an ensemble approach of machine learning to tackle this issue. A dataset obtained from a 25 KW PV power farm is used to categorize panels in four classes, including three fault types: string fault, string-ground fault, and string–string fault, and forth one is without fault. Initially, a feature reduction technique is employed, reducing the feature size from 30 to 4. Subsequently, a Bayesian-optimized ensemble approach utilizing the bagging method is applied to identify three types of faults, as well as normal conditions. Experimental evaluation suggested that even with just 4 features, the overall classification rate is maintained at 100% accuracy on the training dataset and at 95% on the test dataset.
H. Mewada, Miral Desai, L. S. Syam Sundar· Engineering Research Express· 0 citations
Deep learning frameworks, such as Simple Recurrent Neural Network, Long Short-Term Memory, Long Short-Term Memory, Bidirectional LSTM, Bidirectional LSTM, and Gated Recurrent Unit, are employed for the detection and categorization of IM.
R. Sooraj, S. Ramu, R. Sitharthan et al.· Scientific Reports· 0 citations
