Fault Detection in Solar Energy Systems: A Deep Learning Approach

Abstract

While solar energy holds great significance as a clean and sustainable energy source, photovoltaic panels serve as the linchpin of this energy conversion process. However, defects in these panels can adversely impact energy production, necessitating the rapid and effective detection of such faults. This study explores the potential of using infrared solar module images for the detection of photovoltaic panel defects through deep learning, which represents a crucial step toward enhancing the efficiency and sustainability of solar energy systems. A dataset comprising 20,000 images, derived from infrared solar modules, was utilized in this study, consisting of 12 classes: cell, cell-multi, cracking, diode, diode-multi, hot spot, hot spot-multi, no-anomaly, offline-module, shadowing, soiling, and vegetation. The methodology employed the exemplar Efficientb0 model. From the exemplar model, 17,000 features were selected using the NCA feature selector. Subsequently, classification was performed using an SVM classifier. The proposed method applied to a dataset consisting of 12 classes has yielded successful results in terms of accuracy, F1-score, precision, and sensitivity metrics. These results indicate average values of 93.93% accuracy, 89.82% F1-score, 91.50% precision, and 88.28% sensitivity, respectively. The proposed method in this study accurately classifies photovoltaic panel defects based on images of infrared solar modules.

Thermal Image and Inverter Data Analysis for Fault Detection and Diagnosis of PV Systems

“…Looking at the averages of the performance values obtained from the data analyzed for three different years and a total of 18 different months, the highest numerical values were obtained for the kNN model. Duranay [43] presented the performance metric results of the classification of PV faults and compared the results of different studies given in the literature using the same dataset [43]. The results reported in the study show that the average precision was in the range of 88.55-98.24%, and average F1-score was in the range of 84.45-97.51%.…”

Radiometric Infrared Thermography of Solar Photovoltaic Systems: An Explainable Predictive Maintenance Approach for Remote Aerial Diagnostic Monitoring

“…Different approaches have been recently proposed in the context of detecting and/or classifying SPV panel defects, as can be seen from Table 1 [43][44][45][46][47][48][49][50][51][52][53][54][55][56][57].…”

“…The tools and frameworks applied for the preprocessing and analysis include MATLAB/Simulink ® [43], Adobe Photoshop ® [44], C Sharp programming language, and OpenCVSharp (v. 3.4.1) [55], among others. The methodologies used for image processing and image (texture) feature extraction/mapping/selection are comprised of geometrical/statistical parameters [43,45], filtering techniques [44,53,54,58], temperature metrics [56], mathematical/perspective transforms [51,54,58], thresholding [54,56,58], masking/image binarization [54,57], augmentation [43][44][45][46][47][48][49][50][51][52][53][54][55][56][57], synthetic/generative oversampling [46,49], among others.…”

“…The algorithms applied for classification include mathematical/statistical learning techniques (random sample consensus-RANSAC [56], Robust Principal Component Analysis-RPCA [54]), Artificial Neural Network (ANN) [43,45], convolutional neural network (CNN) [50,58], pre-trained CNN (VGG16 [44], ResNet [46,52], AlexNet [47], Effi-cientNet [48], Ensemble Learning [46], machine learning-ML (K-NN [51], SVM [48], K-fold cross-validation for fewer data [50]), CNN-ML combination (U-Net with decision tree) [57], CNN-CNN combination (UDenseNet) [49], that have relatively resulted in a perfect binary classification accuracy of 100% and a highest multiclass (four-class) diagnosis reaching 99.8% [57,58]. For georeferencing and precise localization, either aluminum identification marks on SPV panel corners are used in IRT imaging due to less emissivity and better pointing, or matching/overlapping IRT-RGB orthomosaic maps are considered [55,59].…”