Solar Cell Busbars Surface Defect Detection based on Deep Convolutional Neural Network (original) (raw)
Related papers
International Journal of Photoenergy, 2021
The defects of solar cell component (SCC) will affect the service life and power generation efficiency. In this paper, the defect images of SCC were taken by the photoluminescence (PL) method and processed by an advanced lightweight convolutional neural network (CNN). Firstly, in order to solve the high pixel SCC image detection, each silicon wafer image was segmented based on local difference extremum of edge projection (LDEEP). Secondly, in order to detect the defects with small size or weak edges in the silicon wafer, an improved lightweight CNN model with deep backbone feature extraction network structure was proposed, as the enhancing feature fusion layer and the three-scale feature prediction layer; the model provided more feature detail. The final experimental results showed that the improved model achieves a good balance between the detection accuracy and detection speed, with the mean average precision (mAP) reaching 87.55%, which was 6.78% higher than the original algorith...
Deep Learning-Based Model for Defect Detection and Localization on Photovoltaic Panels
Computer Systems Science and Engineering
The Problem of Photovoltaic (PV) defects detection and classification has been well studied. Several techniques exist in identifying the defects and localizing them in PV panels that use various features, but suffer to achieve higher performance. An efficient Real-Time Multi Variant Deep learning Model (RMVDM) is presented in this article to handle this issue. The method considers different defects like a spotlight, crack, dust, and micro-cracks to detect the defects as well as localizes the defects. The image data set given has been preprocessed by applying the Region-Based Histogram Approximation (RHA) algorithm. The preprocessed images are applied with Gray Scale Quantization Algorithm (GSQA) to extract the features. Extracted features are trained with a Multi Variant Deep learning model where the model trained with a number of layers belongs to different classes of neurons. Each class neuron has been designed to measure Defect Class Support (DCS). At the test phase, the input image has been applied with different operations, and the features extracted passed through the model trained. The output layer returns a number of DCS values using which the method identifies the class of defect and localizes the defect in the image. Further, the method uses the Higher-Order Texture Localization (HOTL) technique in localizing the defect. The proposed model produces efficient results with around 97% in defect detection and localization with higher accuracy and less time complexity.
Analysis Of Cracks In Photovoltaic Module Cells From Electroluminescence Images By Deep Learning
BOOK OF PROCEEDINGS OF 1st International Conference on Computing and Machine Intelligence , 2021
With the spread of solar power plants and investors becoming more conscious, the demand for quality and efficient solar panels is increasing day by day. Creating quality products is possible by minimizing the errors in production processes. The efficiency and quality of solar panels is directly proportional to the efficiency and quality of the solar cell used in the panel.In this study, it aims to provide useful contributions to 3 different steps in the solar panel production process: firstly, the quality control of the solar cell to be used before production, secondly, the detection and replacement of cells having cracks in the production process, and the classification of the panels, finally the detection of performance losses and cause after the panel production.In this study, the deep learning network models and datasets used in the literature were first examined and then the images obtained from the Electroluminescence devices were analyzed with deep learning. Alexnet model, one of the deep learning networks, was used. As a data set, a total of 876 60-cell solar panels with a minimum of 1 and a maximum of 28 cracks were obtained with a resolution of 4730x2883. As a result, 79.4 percent crack detection rate was achieved with the proposed method.
In this study, an automatic solar defect detection and classification system using deep learning was proposed. This study focuses on solar faults in photovoltaic systems identified through Electroluminescence (EL) images by employing a deep learning framework that utilizes both traditional Convolutional Neural Networks (CNNs) and a pre-trained VGG16 and VGG-19 network for feature extraction. This approach was designed to enhance the accuracy and efficiency of solar defect classification. The framework is structured into three main phases: image preprocessing, feature extraction using CNNs, Histogram of Oriented Gradients (HOG) and Artificial Neural Networks (ANN), and classification through a Deep Neural Network (DNN). During preprocessing, images are scaled down to uniform dimensions to ensure consistent learning. They adopted two classification strategies: binary classification (defective or non-defective) and multiclass classification; the class names are 0%, 33%, 67%, and 100% (here, % represents the percentage of defectiveness), which represents the defect likelihood. To refine the model's performance, a data augmentation technique has been utilized on the dataset. The effectiveness of the model was evaluated using various metrics, including the precision, recall, F1-score, and accuracy for two and four classes and obtained on, supported by confusion matrices. VGG-19 model outperformed other models and achieved precision, recall, F1-score and accuracy of 90% each for two classes respectively and similarly 94% for four classes. This study compares two classification methods to assess the ability of the deep learning framework to detect and classify solar defect images automatically.
2020
Luminescence images of solar cells show materialand process-related defects in solar cells, which are relevant for monitoring, optimization and processing. Convolutional neural networks (CNNs) allow the reliable segmentation of these defects in images of the solar cells. Nevertheless, the training of CNNs requires a large amount of empirical data, in which the defects have to be labeled expensively by experts. We introduce a method allowing efficient training by using Smart Labels. We show how this technique can be used for process monitoring to detect systematic errors. This approach differs from previous methods, which rely on human heuristics in the form of feature engineering or learning-based methods with human-annotated defects. However, this previous approach has some limitations and risks. These include label mistakes due to overlapping defect structures, poorly reproducible annotations and varying label quality. Furthermore, existing algorithms have to be adapted to new cel...
Deep Learning for Automatic Defect Detection in PV Modules Using Electroluminescence Images
Deep Learning for Automatic Defect Detection in PV Modules Using Electroluminescence Images, 2023
Solar energy, in the form of photovoltaic (PV) panels, is important for achieving clean energy solutions. The photovoltaic health index must be monitored and improved because of the high demand for green energy. Unfortunately, defective solar cells are a significant source of performance degradation in photovoltaic (PV) systems. Experts often manually analyze electroluminescence (EL) images by visually inspecting them, which is personal, time-consuming, and requires extensive expertise. This work presents a comparative analysis of YOLOv8 and an Improved YOLOv5 for an automatic PV defect detection system in EL images in which Global Attention Module (GAM) is incorporated into the traditional YOLOv5s model for better object representation. Adaptive Feature space fusion (ASFF) was added to YOLOv5's original structure for feature fusion. The Distance Intersection over Union (Non-Maximum) Suppression (DIoU-NMS) is aggregated to produce a more accurate bounding box. The ELDDS1400C5 dataset was used to train and evaluate the proposed system. Experiments on the ELDDS1400C5 test set revealed that the Improved YOLOv5 algorithm achieved a mean Average Precision of 76.3% (mAP@0.5), which is a 2.5% improvement over the standard YOLOv5 algorithm for detecting faults in PV modules in EL images. Furthermore, the experimental results demonstrated that Test Time Augmentation (TTA) significantly increased the mAP@0.5 to 77.7%, surpassing the YOLOv8 model, which achieved 77.5% under the same conditions.
Indonesian Journal of Electrical Engineering and Computer Science, 2023
In the last few years, the development of renewable energies has increased on a large scale. At least, to guarantee the security and stability of the photovolataic system's production, it is imperative that the photovoltaic modules exhibit a high level of reliability. Therefore, the development of an intelligent detection environment to enable the identification of defects in solar cells during manufacturing has become an important issue for the growth of the photovoltaic (PV) sector. This work proposed a fault diagnosis of surface solar cells using deep learning methods for computer vision, using the eighth version of the you only look once (YOLOv8) algorithm. This detection method was applied to a dataset of electroluminescence (EL) images containing twelve PV cell defects on a publicly available heterogeneous background. Then, using this dataset, we trained, validated, and tested the YOLOv8, YOLOv5 models. The results show that YOLOv8 provides a high level of accuracy in fault diagnosis compared with YOLOv5, and also improves the detection speed of the model. Indeed, the average precision achieves 90.5% This suggested approach ensures high accuracy in fault identification which demonstrates the effectiveness of computer vision to identify multi-object cell defects.
Detecting Defects in PCB using Deep Learning via Convolution Neural Networks
2018 13th International Microsystems, Packaging, Assembly and Circuits Technology Conference (IMPACT), 2018
In this paper we have deployed the concept of deep learning known as convolutional neural networks (CNN) as we can realize nowadays deep learning is growing in each and every field. Deep learning is executed in each and every platform and its outcome is impressive. On the other hand, the capability and accuracy of deep learning is somehow compared with human beings. We trained CNN to classify either defective or good printed circuit board(PCB). In this experiment we have used 41,387 images, which is divided into 3 different data sets i.e. training, validation and testing. The CNN, which has 60 million parameters and 500,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and two globally connected layers with a final 1000-way softmax. Hence, deep learning via convolution neural networks has been introduced in this paper, which will eventually increase the accuracy and reduce a lot of time and consumption of skilled manpower. According to this preliminary study, we can overall achieve accuracy of above 88% and minimize the count of defective PCB classifying as good. In the near future, we hope that over 95% accuracy can be achieved by using different CNN models like VGGNET, RESNET and GOOGLENET and collecting more PCB image data in order to reduce the consumption of time, manpower and increase the accuracy in quality inspection.
Classification of Solar Cells EL Images with Different Busbars Via Deep Learning Models
Sakarya University Journal of Computer and Information Sciences, 2024
Electricity generation from renewable energy sources such as solar energy has come to the forefront in the last decade. The solar energy cell is an indispensable part of the solar energy ecosystem of solar panels, and defective cells cause financial losses in energy production. Experienced experts are needed to detect defects on solar cells. Autonomous systems are important to accelerate the process. Classical image processing techniques are used to manually detect defects on cells. To use these techniques, many parameters are need to be entered into EL imaging software. However, in this study, these processes were carried out automatically without the need for external intervention. False detection/classification may occur during the processes performed by EL imaging devices due to weakness of the operator experience or EL imaging software. It is aimed to use automatic image processing and then deep learning techniques to achieve faster and higher performance than the results obtained from EL imaging devices using classic image processing techniques. AI algorithm and deep learning models can be an important solution. In this study, two AI algorithm and 10 different deep learning models were used to classify solar cells. EL images of defective and normal solar cells with 4 and 5 busbars were used in the study. The dataset, includes 9360 images of solar cells, 4680 of which are defective and 4680 are normal. Performance evaluation of the models made according to the confusion matrix. According to the results, Mobilenet-v2 and VGG-19 achieved the highest validation accuracy rate of 99.68%. According to F1-score, Mobilenetv2 achieved the highest performance of 99.73%. It has been shown that the Mobilenet-v2 is slightly more successful than other models in terms of validation and F1-score. The results show that trained DL models can be used as an inspection method in the production line of solar panels and cells.
Energies
The past two decades have seen an increase in the deployment of photovoltaic installations as nations around the world try to play their part in dampening the impacts of global warming. The manufacturing of solar cells can be defined as a rigorous process starting with silicon extraction. The increase in demand has multiple implications for manual quality inspection. With automated inspection as the ultimate goal, researchers are actively experimenting with convolutional neural network architectures. This review presents an overview of the electroluminescence image-extraction process, conventional image-processing techniques deployed for solar cell defect detection, arising challenges, the present landscape shifting towards computer vision architectures, and emerging trends.