AI-Based Inline Metrology for Defect Detection in Solar Cells

A system for detecting defects in images of a specimen, comprising an ensemble of deep learning models and a pseudo-loss function, wherein the ensemble is trained with a training dataset and parameters are adjusted until the pseudo-loss function reaches a threshold of approximately 0.5, and wherein the trained ensemble is used to detect defects in runtime specimen images.

Source

A substrate inspection method and device that improves defect detection performance by acquiring multiple images of a substrate under different optical conditions, generating difference images and difference-of-difference images, and detecting low-SNR defect candidates through a multi-image analysis approach.

Source

A system for determining information for a specimen, such as classifying defects or predicting structure characteristics, that uses a combination of local and global texture features. The system determines a global texture characteristic of an image of the specimen and one or more local characteristics of a localized area in the image, and then inputs these features into a machine learning model to determine the specimen information. The global texture characteristic captures patterns that are present across the entire image, while the local characteristics capture patterns in specific regions. The system can be used for both metrology and inspection applications, and can improve classification performance by incorporating global texture information that is typically neglected by traditional feature generation techniques.

Source

Solar cells are playing a significant role in aerospace equipment. In view of the surface defect characteristics in the manufacturing process of solar cells, the common surface defects are divided into three categories, which include difficult-detecting defects (mismatch), general defects (bubble, glass-crack and cell-crack) and easy-detecting defects (glass-upside-down). Corresponding to different types of defects, the deep learning model with different optimization methods and a classification detection method based on multi-models fusion are proposed in the paper. In the proposed model, in order to solve the mismatch problem between the default anchor boxes size of YOLOv5s model and the extreme scale of the battery mismatch defect label boxes, the K-means algorithm was adopted to re-cluster the dedicated anchor boxes for the mismatch defect label boxes. In order to improve the comprehensive detection accuracy of YOLOv5s model for the general defects, the YOLOv5s model was also improved by the methods of image preprocessing, anchor box improving and detection head replacing. In ord... Read More

Source

A method and system for evaluating defective areas of semiconductor substrates, comprising: enhancing the substrate image through a multi-level neural network; segmenting the enhanced image into clusters based on pixel features; and identifying defect locations through probe-based excitation and marking.

Source

A solar wafer defect detection system using deep learning that detects various defects including cracks, scratches, LCOs, hotspots, injectors, and LDSEs in solar wafers. The system employs a defect detection deep learning model and a black spot classification deep learning model to analyze electroluminescence images of the wafers. The defect detection model is trained on labeled images of defective wafers, while the black spot classification model identifies black spots in the images. The system provides real-time detection and classification of defects, enabling early detection and prevention of wafer failures.

Source

A semiconductor fabrication inspection system that uses a classification CNN to determine wafer defects, trained with a novel encoder-decoder architecture that generates soft labels through clustering of feature vectors. The system captures wafer images and classifies them into multiple defect types, with the training process utilizing multiple images to optimize the encoder, decoder, and classification CNN.

Source

A method for defect inspection in semiconductor fabrication that compensates for noise in current process steps by subtracting a noise image from a previous process step. The method involves generating images of sample regions after each process step, identifying a test region and comparison regions, and generating a multi-step difference image by weighted subtraction of previous process step images from the current process step image. Defects are then identified in the test region based on the multi-step difference image.

Source

Abstract The increasing production of solar cells, resulting from the rapid development of new energy sources, necessitates their inspection during both solar cell production and photovoltaic power plant inspection. Target detection algorithms are widely utilized for defect detection in solar cells. To achieve more accurate detection of minor defects in electroluminescent solar cells, an improved algorithm called CSR-YOLOv5s is proposed in this paper. The CSR-YOLOv5s combines Decoupled Head and CSRBlock with the YOLOv5s baseline model. The CSR-Y OLOv5s demonstrates a 1.1% increase in accuracy and a 2.1% increase in F1-score compared to the YOLOv5s baseline model, resulting in improved accuracy and recall. The algorithm effectively identifies minor defects in electroluminescent solar cells.

Source

Renewable energy resources such as solar energy, biomass, tidal, geothermal, and hydroelectric energy are becoming increasingly important due to their potential to mitigate the negative impacts of climate change and reduce our dependence on finite and polluting fossil fuels. Solar power can provide a clean, sustainable, and reliable source of renewable energy. Important component of solar power generation is the silicon panel and its surface quality is highly related to its robustness and power generation efficiency. Cell breakages resulting from micro-cracks, degradation and shunted areas on cells are proven to cause major issues and these affect the photovoltaic module efficiency and performance. Solar cell defect identification is important because defects in solar cells significantly reduce their efficiency, which in turn affects their power output and lifespan. By identifying and classifying defects during the production of these cells, engineers and researchers can improve the quality control of solar cells, leading to more reliable and efficient solar energy systems. The propo... Read More

Source

This paper proposed an improved model based on YOLOv7, with the aim of tackling the issues of low detection accuracy and heavy model weight of current defect detection algorithms for photovoltaic cells. Ghost convolution is introduced in the original backbone network and neck to reduce the weight of the model. Secondly, the DenseBlock module is improved in two parts to enhance feature extraction ability and feature fusion ability. The introduction of a Wise-IOU loss function to further enhance the detection accuracy of the model was finally successful. Experiments with four defect datassets featuring cracks, broken gates, thick lines and black cores revealed that the weight size of the improved model in this paper was reduced by 24.4%, and the defect detection accuracy and mean average detection accuracy were both increased by 9.7% and 3.7% respectively, in comparison to the original YOLOv7 network.

Source

In the production process of solar cells, inevitable faults such as cracks, dirt, dark spots, and scratches may occur, which could potentially impact the lifespan and power generation efficiency of solar cells. Addressing this issue, this paper combines neural networks with photoluminescence detection technology and proposes a novel neural network model for the classification and grading of defects in solar cells. Firstly, the YOLOv5 model is optimized and adjusted for algorithm and network structure. The optimization process is divided into three parts: global optimization of the network structure, optimization of the neck network structure, and optimization of the head structure, each addressing specific issues in recognition, detection, and classification. The impact of the optimized network model on recognition and detection speed is analyzed, and solutions are proposed to address any observed effects. Additionally, an iterative update of neural network hyperparameter combinations is performed for solar cell defect identification. Finally, using the ultimately optimized model str... Read More

Source

One way of examining surface defects on photovoltaic modules is the Electroluminescence (EL) imaging technique. The data set used in this work is an open data set for fault detection and classification of photovoltaic cells. In this article, we have used various deep learning (DL) techniques to ensure fault detection and diagnosis of photovoltaic modules. A binary classification model was developed that highlighted defective PV modules and normal modules. The subset of defective PV modules was used to design a multi-class model of default detection (light, moderate, and severe). Evaluation results (confusion matrix, mean square error (mse)) showed that methods based on deep learning performed exceptionally well, making it possible to solve the problem of detecting and diagnosing faults in photovoltaic modules with good overall precision (mse = 0.060).

Source

The importance of solar energy as a renewable power source has led to increased adoption of solar modules for electricity generation. However, faults in solar cells can significantly impact their performance and efficiency. Manual defect detection is time-consuming and subjective, hence the need for an intelligent and efficient detection solution. In this study, we propose a method for detecting defective solar cells in electroluminescence imaging using an advanced object detection algorithm, specifically YOLO5 version. An important step in the algorithm is to formulate the detection problem in terms of real-time detection of defects. We evaluate our method on a dataset of different types of solar modules containing a total of 240 solar cells with various defects, including finger interruptions, microcracks, electrically separated or degraded cell parts and material defects. Experimental evaluation on solar cell images extracted from high-resolution electroluminescence images of photovoltaic modules datasets reveals that the proposed framework successfully mitigates the influence of ... Read More

Source

This paper presents a novel hybrid model employing Artificial Neural Networks (ANN) and Mathematical Morphology (MM) for the effective detection of defects in solar cells. Focusing on issues such as broken corners and black edges caused by environmental factors like broken glass cover, dust, and temperature variations. This study utilizes a hybrid model of ANN and K-Nearest Neighbor (KNN) for temperature prediction. This hybrid approach leverages the strengths of both models, potentially opening up new avenues for improved accuracy in temperature forecasting, which is critical for solar energy applications. The significance lies in the interconnectedness of temperature fluctuations and solar cell efficiency, leading to defects. The proposed model aims to predict temperatures accurately, providing insights into potential solar cell efficiency problems. Subsequently, this work studies the transitions to defect detection using Fuzzy C-Means (FCM) clustering and MM techniques. The hybrid model demonstrates accurate temperature prediction with Mean Absolute Percentage Error (MAPE) values ... Read More

Source

Silicon has been one among a few key technologically important materials especially in our modern world. Silicon, especially in its current most useful forms (thin layers), is a brittle material by nature. Thin silicon cells crack easily during manufacturing assembly. Every crack is a defect in any PV manufacturing lines, and if let to grow/propagate it will lead to reliability and quality issues with time. Nevertheless, the global silicon solar PV industry is aggressively reducing the cost of solar PV systems through cutting down the thickness of silicon solar cells. Cell cracks immediately lower PV module efficiency and can lead to premature aging of the entire module. The Crack Catcher AI was our research joint collaborations entry in the Department of Energy (DOE)s American-Made Solar Innovation competition in 2022 (Round 6) which later was selected as the national semifinalists and won an award announced by DOE in December 2022. It uses smart stress sensing and smart fracture prediction approaches utilizing fundamental fracture mechanics and big data analytics to reduce crac... Read More

Source

This study presents an advanced defect detection approach for solar cells using the YOLOv10 deep learning model. Leveraging a comprehensive dataset of 10,500 solar cell images, annotated with 12 distinct defect types, our model integrates Compact Inverted Blocks (CIB) and Partial Self-Attention (PSA) modules to enhance feature extraction and classification accuracy. Training on the Viking cluster with state-of-the-art GPUs, our model achieved remarkable results, including a mean Average Precision ([email protected]) of 98.5%. Detailed analysis of the model's performance revealed exceptional precision and recall rates for most defect classes, notably achieving 100% accuracy in detecting black core, corner, fragment, scratch, and short circuit defects. Even for challenging defect types such as thick line and star crack, the model maintained high performance with accuracies of 94% and 96%, respectively. The recall-confidence and precision-recall curves further demonstrate the model's robustness and reliability across varying confidence thresholds. This research not only advances... Read More

Source

Renewable energy resources can play an important role in significantly offsetting the power demand in most of the world. Being the lowest-cost renewable resource, investment in photovoltaic solar can be prioritized if the reliability of continuous generation is guaranteed. Micro defects in PV cells can reduce the electrical output and if not detected can lead to large-scale power distributions. Therefore, electroluminescence (EL) image is commonly used these days to detect these micro defects and classify the various types of these defects. Hence, a hybrid approach consisting of Alibi detection algorithm is proposed, which makes use of a deep learning-based approach to detect abnormal solar cells. And decision tree is employed for the previously obtained outlier instance score to further classify it into the different kinds of defects present in the EL image of the solar cells. The proposed model is tested on the ELPV image dataset consisting of mono conventional crystalline silicon cells with above 97% accuracy.

Source

A method for detecting defects in semiconductor wafers during processing, comprising capturing a test image of the wafer, generating a simulated integrated circuit layout by analyzing the test image, and identifying defects in the wafer based on the simulated layout. The simulated layout is generated using a machine learning process trained on reference layouts, enabling detection of defects in complex wafer features through comparison with a reference layout.

Source

Photovoltaic (PV) fault detection is crucial because undetected PV faults can lead to significant energy losses, with some cases experiencing losses of up to 10%. The efficiency of PV systems depends upon the reliable detection and diagnosis of faults. The integration of Artificial Intelligence (AI) techniques has been a growing trend in addressing these issues. The goal of this systematic review is to offer a comprehensive overview of the recent advancements in AI-based methodologies for PV fault detection, consolidating the key findings from 31 research papers. An initial pool of 142 papers were identified, from which 31 were selected for in-depth review following the PRISMA guidelines. The title, objective, methods, and findings of each paper were analyzed, with a focus on machine learning (ML) and deep learning (DL) approaches. ML and DL are particularly suitable for PV fault detection because of their capacity to process and analyze large amounts of data to identify complex patterns and anomalies. This study identified several AI techniques used for fault detection in PV systems... Read More

Source

At different stages of solar panel production, there may be shadows, microcracks and other defects. This will have a influence on the conversion efficiency of turning light energy into electric energy of the battery. For the sake of enhancing the efficiency of solar cell, it is certainly worth detecting the defects of the battery. For the sake of accurately detecting the defects of solar panels, a detection method based on Dense U-Net is came up with. This method can prevent the problem of gradient from disappearing in the process of feature transfer by introducing dense connection network. The results show that this method can produce good results in defect detection.

Source

In this article, the loss analysis of silicon solar cells with polysilicon on locally-etched dielectric passivating contacts with V <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">oc</inf> =729.0 mV and efficiency=22.6% has been presented. Experimentally, nano-pinholes were introduced in SiO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</inf> (2.2 nm) and SiO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</inf> /SiN <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">y</inf> (2.2 nm/8nm) stack using metal-assisted chemical etching (MACE). SunSolve and Quokka3 were used to simulate the experimental solar cell and investigate the optical and electrical power losses. Simulations suggest maximum power loss occurs due to recombination and resistive losses in the bulk (~0.76 mW/cm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ) followed by power ... Read More

Source

A deep learning-based defect detection method for semiconductor elements that uses a generative adversarial network (GAN) algorithm to classify defects based on color, area, and aspect ratio. The method generates segmentation images from reference and defect SEM images, converts them to different colors, combines the images, and trains a GAN model to predict defect types. The trained model is then applied to inspection SEM images to detect defects and predict their types.

Source

Method for analyzing 3D data of semiconductor devices using machine learning to detect anomalies and measure structures, comprising: detecting target objects using a first machine learning logic, applying a transformation to feature space to the classified voxels, and extracting structural features from the transformed voxels.

Source

It is a very important method in nondestructive testing to detect defects using infrared or electroluminescent (EL) images of solar cell modules. Traditionally, this work is completed by skilled technicians, which is not only time-consuming but also vulnerable to subjective factors. The surface defect detection method of solar cells based on machine learning has become one of the main research directions for its high efficiency and convenience. Therefore, this paper proposes a defect detection method based on machine learning for solar cell EL image defect detection. First, the EL image is preprocessed to provide a clearer image with more obvious defects for subsequent feature extraction; Then, the gray, contrast, brightness and edge gradient features of the preprocessed image are extracted, and the edge gradient information of the defect is extracted using the improved LBP algorithm. Combined with the gray value, contrast and brightness information of the image, the training and testing data set of the classifier model is formed. Finally, the optimized decision tree algorithm - C4.5... Read More

Source

Automated defect inspection system for semiconductor manufacturing that uses optical detectors to scan substrate surfaces and generate images, which are then analyzed by a processor to detect defects in hole arrays based on predetermined criteria. The system can be integrated into a manufacturing line between processing stations to enable real-time quality control.

Source

A nondestructive detection method that combines convolutional neural network (CNN) and photoluminescence (PL) imaging was proposed for the multi-classification and multi-grading of defects during the fabrication process of silicon solar cells. In this paper, the PL was applied to collect the images of the defects of solar cells, and an image pre-processing method was introduced for enhancing the features of the defect images. Simultaneously, the defects were defined by 13 categories and three divided grades of each under the definition rules of defects that were proposed in accordance with distribution and characteristics of each defect category, and expand data were processed by various data augmentation. The model was therefore improved and optimized based on the YOLOv5 as the feature extractor and classifier. The capability of the model on distinguishing categories and grades of solar cell defects was improved via parameter tuning and image pre-processing. Through experimental analysis, the optimal combination of hyperparameters and the actual effect of data sample pre-processing ... Read More

Source

A deep learning model trained without labeled data generates a reference for a specimen from one or more inputs including a specimen image or data generated from the image. The reference is used to determine information for the specimen, such as defects or metrology data, from the specimen image or generated data. The model can be applied to inspection, metrology, and defect review applications in semiconductor manufacturing.

Source

Using PV systems, solar energy may be used to create electricity. Every year, the proportion of solar energy in the electric system increases significantly. On the other hand, photovoltaic cells are susceptible to malfunctions that diminish their efficiency and profitability. Due to the severity of the defects, fault detection and diagnosis (FDD) in the PV system have become difficult. Thus, the primary objective of the proposed study is to detect and diagnose particular types of PV system problems using an artificial neural network (ANN). This early operation is more effective for avoiding errors during PV installation and minimizing PV system power losses. A new solar cell model is created in the MATLAB/SIMULINK environment and is used to identify the fault dataset. The solar cell consists of three parallel strings and three series modules, with each module containing 20 series-connected photovoltaic cells. This model determines four parameters (V-load in Volts, I-load in Amps, Irradiance in W/m2, and Temperature in Celsius) under varying situations (five temperature values, three ... Read More

Source

Solar-cell panels use sunlight as a source of energy to generate electricity. However, the performances of solar panels decline when they degrade, owing to defects. Some common defects in solar-cell panels include hot spots, cracking, and dust. Hence, it is important to efficiently detect defects in solar-cell panels and repair them. In this study, we propose a lightweight inception residual convolutional network (LIRNet) to detect defects in solar-cell panels. LIRNet is a neural network model that utilizes deep learning techniques. To achieve high model performance on solar panels, including high fault detection accuracy and processing speed, LIRNet draws on hierarchical learning, which is a two-phase solar-panel-defect classification method. The first phase is the data-preprocessing stage. We use the K-means clustering algorithm to refine the dataset. The second phase is the training of the model. We designed a powerful and lightweight neural network model to enhance accuracy and speed up the training time. In the experiment, LIRNet improved the accuracy by approximately 8% and per... Read More

Source

The effect of process-induced defects on the photo-generated charge-carrier lifetime and solar cell performance is critical, which will help optimize the process recipe. In this work, we attempt to quantify the effects of process-induced defects during boron implantation on the n-type silicon wafer in different annealing ambiences. We have evaluated the role of defects that can be formed during oxygen and inert ambience annealing on n-type bifacial passivated emitter rear totally diffused solar cells using a recombination current prefactor (J0). The numerically calculated J0 is calibrated with the reported experimental J0 values using two different methods: (i) ShockleyReadHall lifetime and (ii) effective trap-density method. In the latter method, we used the simulated defect density profiles. Both methods capture the process-induced degradation. We observed that the process-induced defects could deteriorate by almost 1% absolute efficiency for the considered annealing conditions. We found that dislocation loops alone cause an ignorable effect on terminal characteristics, but other... Read More

Source

Fast and non-destructive analysis of material defect is a crucial demand for semiconductor devices. Herein, we are devoted to exploring a solar-cell defect analysis method based on machine learning of the modulated transient photovoltage (m-TPV) measurement. The perturbation photovoltage generation and decay mechanism of the solar cell is firstly clarified for this study. High-throughput electrical transient simulations are further carried out to establish a database containing millions of m-TPV curves. This database is subsequently used to train an artificial neural network to correlate the m-TPV and defect properties of the perovskite solar cell. A Back Propagation neural network has been screened out and applied to provide a multiple parameter defect analysis of the cell. This analysis reveals that in a practical solar cell, compared to the defect density the charge capturing cross-section plays a more critical role in influencing the charge recombination properties. We believe this defect analysis approach will play a more important and diverse role for solar cell studies.

Source

This paper proposes an automated defect detection method for photovoltaic (PV) modules using a developed convolutional neural network (CNN) architecture. By examining the electroluminescence (EL) image, the system determines whether the solar cell/PV module will be accepted or rejected based on the presence and extent of the defect. The system detects defects at the cell and module level, identifying cracks, minicracks, potential induced degradation (PID), and shaded areas. The proposed system achieves a validation accuracy of 98.07% and is effective in accurately evaluating the health of PV modules. The tool reduces the demand for manual inspection, minimizes human error, saves time, and increases efficiency, ultimately leading to high-quality standards and reduced costs. The system was validated in a case study for PV installation faulted with PID, where it identified all defective modules with a high degree of precision 96.6%. Furthermore, a significant aspect of the study is that the proposed system outperforms existing studies, and this can have profound implications for the PV ... Read More

Source

The surface defects such as cracks, broken cells and unsoldered areas on the solar cell caused by manufacturing process defects or artificial operation seriously affect the efficiency of solar cell. For the surface defects of solar cell, which have the characteristics of various shapes, large-scale changes, and difficult to detect, a surface defect detection algorithm based on similarity non-maximum suppression mechanism is proposed by improving the Faster region-based convolution neural network in this paper. In the proposed algorithm, a similarity non-maximum suppression mechanism is used, and the effectiveness of prediction frame screening is improved by introducing the cosine similarity of candidate box aspect-ratio. In addition, the cross-layer connection based on Shuffle operation and the three-branch dilated convolution block are introduced in the main feature extraction channel, which improves the network's ability to express features through multi-scale feature fusion. The experimental results show that, compared with the latest deep learning target detection models, the pro... Read More

Source

The effective passivation material (ITIC) for PIN type perovskite solar cells is selected by machine learning. In the verification experiment, the defect density of the perovskite layer is significantly decreased after treatment with ITIC.

Source

The common defects observed on the photovoltaic cells during the manufacturing process include chipping, tree crack, micro-line, soldering, and short circuits. Most of the defects mentioned above are not directly visible, making it harder for visual inspection. An appropriate method for defect classification would be by performing electroluminescence (EL) imaging, which helps reveal the defects, making it possible to visualize cracks and helps evaluate the quality of photovoltaic (PV) modules. The phenomenon where light emission occurs when current passes through PV cells is electroluminescence. The manual analysis of these electroluminescence images can be time-consuming and needs expert knowledge of various defects. This paper explains the automatic defective solar mono-cell classification task executed with different classifiers of machine learning and deep learning along with necessary image preprocessing techniques used to enhance the detection results. In comparison with the machine learning approach, deep learning offers better results on dataset of 1840 solar cell images. CNN... Read More

Source

Passivating contacts based on poly-Si have enabled record-high c-Si solar cell efficiencies due to their excellent surface passivation quality and carrier selectivity. The eventual existence of pinholes within the ultra-thin SiOx layer is one of the key factors for carrier collection, beside the tunneling mechanism. However, pinholes are usually believed to have negative impact on the passivation quality of poly-Si passivating contacts. This work studied the influence of the pinhole density on the passivation quality of ion-implanted poly-Si passivating contacts by decoupling the pinhole generation from the dopants diffusion process by means of two annealing steps: (1) a pre-annealing step at high temperature after the intrinsic poly-Si deposition to visualize the formation of pinholes and (2) a post-annealing step for dopants activation/diffusion after ion-implantation. The pinhole density is quantified in the range of 1106 to 3108 cm2 by the TMAH selective etching approach. The passivation quality is discussed with respect to the pinhole density and the post-annealing thermal bud... Read More

Source

Nowadays, the rapid development of photovoltaic(PV) power stations requires increasingly reliable maintenance and fault diagnosis of PV modules in the field. Due to the effectiveness, convolutional neural network (CNN) has been widely used in the existing automatic defect detection of PV cells. However, the parameters of these CNN-based models are very large, which require stringent hardware resources and it is difficult to be applied in actual industrial projects. To solve these problems, we propose a novel lightweight high-performance model for automatic defect detection of PV cells in electroluminescence(EL) images based on neural architecture search and knowledge distillation. To auto-design an effective lightweight model, we introduce neural architecture search to the field of PV cell defect classification for the first time. Since the defect can be any size, we design a proper search structure of network to better exploit the multi-scale characteristic. To improve the overall performance of the searched lightweight model, we further transfer the knowledge learned by the existin... Read More

Source

Renewable energies have an increasing share in the energy supply. In order to ensure the security of this supply, the reliability of the systems is therefore increasingly important. In photovoltaic modules or in manufacturing, defective solar cells due to broken busbars, cross-connectors or faulty solder joints must be detected and repaired quickly and reliably. This paper shows how the magnetic field imaging method can be used to detect defects in solar cells and modules without contact during operation. For the evaluation of the measurement data several neural networks were used, which were trained with the help of results from finite element simulations. Different training data sets were set up in the simulation model by varying the electrical conductivities of the different parts of the solar cell. The influence of the neural network type and the variation of the training data sets as well as an advantage of a combination of simulated and experimental training data are presented and discussed.

Source

A deep learning-based method for denoising images of semiconductor specimens to improve defect detection accuracy. The method employs a convolutional neural network (CNN) to remove noise from images generated by optical inspection tools, enabling more sensitive detection of critical defects. The denoised images are then analyzed to determine specimen information, such as defect presence and location.

Source

The detection of micro cracks on the surface of solar cells is very important to improve the durability of photovoltaic modules. In this paper, Haar feature extraction and kernel fuzzy c-means clustering algorithms are proposed to detect the defects of solar cells. Haar extended template is used to extract the edge features as training samples, combined with kernel fuzzy c-means clustering (KFCM) algorithm and improved Xie Beni index to detect the surface defects of solar cells. The recognition rate of no defects is 98%, and the recognition rate of vertical finger defects is 97%, The recognition rate of microcrack is 93%, and that of fracture is 92%.

Source

A system and method for determining information from a specimen image without a reference image, using a semantic segmentation model to assign labels to each pixel in the image based on its content, and then extracting relevant information from the labeled pixels. The model is trained to recognize specific defects or features, enabling defect detection and characterization without the need for a reference image.

Source

A system and method for improved semiconductor inspection using tensor decompositions and singular value decomposition (SVD) processes. The system generates a new type of reference image for inspection comparison operations by decomposing target images into tensors and extracting reference images from the tensor components. The system then performs SVD on the difference images to identify defects, and selectively modifies the singular vectors to enhance defect detection sensitivity.

Source

A literature review is presented addressing the major characteristics to assess photovoltaic modules electroluminescent emission in order to detect defects as a complementation of the traditional photovoltaic characterization methods. General aspects are presented regarding the image acquisition and the methods for pre-processing are mentioned. A review of the works on the subject of defect detection shows a strong tendency for automatic processing and the development and application of artificial intelligence techniques. Machine learning methods are a promising alternative in terms of computational requirements and assertivity on the detection.

Source

A method of intelligent detection and precise localization of EL compositive defects in PV modules based on machine learning has been proposed. In the previous methods, the individual cell is taken as the inspection unit, however, this can not detect the holistic defects in PV modules. In this paper, two series-connected YOLOv5 networks with upstream and downstream process-dependent inspection model is established, which take a PV module as the detecting unit instead of a single cell. It can finish the detection and localization of 13 kinds of defects commonly found currently. These defects include not only those in a single cell, but also holistic defects that may appear after the cell assembling as a module. The experimental results show that both of detecting accuracy and speed are improved effectively for each PV module than before.

Source

A technique for accurately identifying image defects using segmentation and compression to improve the reliability of AI-based defect detection systems. The technique involves generating two learning models, one using segmented images and another using compressed and segmented images, to identify the type of image defect. This allows the system to capture both high-frequency and low-frequency defects. The segmentation and compression steps are applied to the training data to generate the learning models. One model learns segmented images to detect high-frequency defects, while the other learns compressed and segmented images to detect low-frequency defects. These models are then used to identify defect types in inspected images.

Source

A defect inspection device that automatically adjusts judgment parameters for separating defects from noise in images. The device uses a pre-trained identification part to extract features from inspection images, and then employs a post-processing part to specify defect areas based on the extracted features. The device calculates image scores based on pixel density within and outside the specified defect areas, and updates the judgment parameters to maximize the difference between these scores. This enables the device to automatically optimize its defect detection performance without manual parameter tuning.

Source

A system and method for inferring defects in semiconductor wafers using machine learning, where a scanning electron microscope captures images of the wafer's circuit patterns and a mask layout image is combined with the wafer image to enable defect detection through machine learning analysis.

Source

A method for characterizing coating surface defects using machine learning. The method involves acquiring digital images of coating surfaces, processing the images to detect and classify defects, and generating quantitative characterizations of the defects. The processing includes image classification, contour detection, edge detection, clustering, and morphological transformations. The method also includes training a predictive model using labeled training images to recognize defect patterns, and using the model to predict coating composition properties based on observed defects.

Source

More than half of the total renewable addictions correspond to solar photovoltaic (PV) energy. In a context with such an important impact of this resource, being able to produce reliable and safety energy is extremely important and operation and maintenance (O&M) of PV sites must be increasingly intelligent and advanced. The use of Artificial Intelligence (AI) for the defects identification, location and classification is very interesting, as PV plants are increasing in size and quantity. Inspection techniques in PV systems are diverse, and within them, electroluminescence (EL) inspection and current-voltage (I-V) curves are one of the most important. In this sense, this work presents a classifier of defects at the PV cell level, based on AI, EL images and cell I-V curves. To achieve this, it has been necessary to develop an instrument to measure the I-V curve at the cell level, used to label each of the PV cells. In order to determine the classification of cell defects, CNNs will be used. Results obtained have been satisfactory, and improvement is expected from a greater number of s... Read More

Source