Defect Shape Research Articles

Surface defect detection of CFRP materials based on infrared thermography and Attention U-Net algorithm

ABSTRACT Carbon fiber reinforced polymer (CFRP) materials have been widely used in aerospace and other fields because of their excellent properties such as high temperature resistance and corrosion resistance, so the nondestructive inspection technology for CFRP materials has become a hot research topic. In this paper, we propose a method based on infrared thermography and Attention U-Net algorithm to characterize the defect shape of CFRP material surface. Firstly, the CFRP surface is scanned by a line laser and the trend of its temperature distribution is recorded using an infrared thermography camera. Subsequently, temperature analysis and entropy value of image information are calculated for individual defect image blocks in order to select images with clear defect contours. Next, the Attention U-Net is used to segment the defect in the image blocks, and the defect shape is characterization. By calculating the evaluation indexes of image segmentation, the method in this paper can achieve 99.57% accuracy, 97.06% recall, 96.63% precision, and the processing time for a single image is 0.13s. Finally, the algorithm of this paper is compared with other algorithms to verify the advantages of this research in the task of detecting surface defects in CFRP materials with fast and high accuracy.

FDD: a deep learning-based steel defect detectors.

Surface defects are a common issue that affects product quality in the industrial manufacturing process. Many companies put a lot of effort into developing automated inspection systems to handle this issue. In this work, we propose a novel deep learning-based surface defect inspection system called the forceful steel defect detector (FDD), especially for steel surface defect detection. Our model adopts the state-of-the-art cascade R-CNN as the baseline architecture and improves it with the deformable convolution and the deformable RoI pooling to adapt to the geometric shape of defects. Besides, our model adopts the guided anchoring region proposal to generate bounding boxes with higher accuracies. Moreover, to enrich the point of view of input images, we propose the random scaling and the ultimate scaling techniques in the training and inference process, respectively. The experimental studies on the Severstal steel dataset, NEU steel dataset, and DAGM dataset demonstrate that our proposed model effectively improved the detection accuracy in terms of the average recall (AR) and the mean average precision (mAP) compared to state-of-the-art defect detection methods. We expect our innovation to accelerate the automation of industrial manufacturing process by increasing the productivity and by sustaining high product qualities.

Analysis of Plane Displacement of Pressure Vessels with Defects

Aiming at the problem that the internal defects of the pressure vessel are not easy to detect, the out-of-plane displacement method is used to study it. The out-of-plane displacement field of circular flat plate heads with defects is obtained by the method of elastic theory analysis and compared with the finite element method to verify the accuracy of the finite element method. Based on the finite element method, the effects of vessel material, pressure load, defect shape, and defect size on the first-order out-of-plane displacement derivative are studied. The results show that the material of the container has little effect on the out-of-plane displacement. The first derivative of out-of-plane displacement increases linearly with the increase of pressure, and this relationship will not change with the increase of pressure. With the increase of defect aspect ratio, the maximum value of the first derivative of out-of-plane displacement increases gradually. When δ ˃ 9, the defect is detectable. The research results lay a theoretical foundation for the internal defect detection of pressure vessels based on digital speckle shearing interferometry.

A Case Series of Prosthetic Rehabilitation of Post-COVID Rhinorbitocerebral Mucormycosis Ocular Defects: A Digital and Analog Experience.

Removal of a part of or the entire orbit results in facial defect, causing psychological trauma to the patient, apart from anatomic loss. This case series presents 6 clinical cases of prosthetic rehabilitation of ocular defects related to post-COVID-19 ROCM by both analog and digital workflow. The basic objective of this case series was to achieve a well retained, user-friendly, maxillofacial ocular prosthesis with esthetic accuracy. Unique size and shape of the ocular defect in each case, variety of skin tones, age range of patients and compromised neuromuscular control made each of the six cases challenging. This clinical series apart from proposing a digital & analog algorithm for rehabilitating ocular defects also illustrates analog workflow for 4 cases and digital workflow for 2 cases for fabrication of ocular prosthesis.

3D characterization of internal defects for fatigue performance of welded SA192 steel water walls

Weld defects in SA192 steel water walls have been characterized using 3D nano-tomography. The fatigue test specimens were extracted from different weld locations of the water walls. The internal welding defects number, size, shape, and location were measured before low-cycle fatigue tests. The weld defect size and shape were found to be the most prominent factors influencing fatigue crack initiation and fatigue life. Sharper defects significantly reduce the fatigue life of the specimen as compared to the rounder defects. The stress intensity factor for each type of defects was calculated using linear-elastic fracture mechanics and the Murakami methods. In the defect-free specimens, the weld fusion zone was the most preferential sites for fatigue crack initiation due to a higher hardness of the as-solidified microstructure.

Targeted expression of bgl23-D, a dominant-negative allele of ATCSLD5, affects cytokinesis of guard mother cells and exine formation of pollen in Arabidopsis thaliana.

Targeted expression of bgl23-D, a dominant-negative allele of ATCSLD5, is a useful genetic approach for functional analysis of ATCSLDs in specific cells and tissues in plants. Stomata are key cellular structures for gas and water exchange in plants and their development is influenced by several genes. We found the A. thaliana bagel23-D (bgl23-D) mutant showing abnormal bagel-shaped single guard cells. The bgl23-D was a novel dominant mutation in the A. thaliana cellulose synthase-like D5 (ATCSLD5) gene that was reported to function in the division of guard mother cells. The dominant character of bgl23-D was used to inhibit ATCSLD5 function in specific cells and tissues. Transgenic A. thaliana expressing bgl23-D cDNA with the promoter of stomata lineage genes, SDD1, MUTE, and FAMA, showed bagel-shaped stomata as observed in the bgl23-D mutant. Especially, the FAMA promoter exhibited a higher frequency of bagel-shaped stomata with severe cytokinesis defects. Expression of bgl23-D cDNA in the tapetum with SP11 promoter or in the anther with ATSP146 promoter induced defects in exine pattern and pollen shape, novel phenotypes that were not shown in the bgl23-D mutant. These results indicated that bgl23-D inhibited unknown ATCSLD(s) that exert the function of exine formation in the tapetum. Furthermore, transgenic A. thaliana expressing bgl23-D cDNA with SDD1, MUTE, and FAMA promoters showed enhanced rosette diameter and increased leaf growth. Taken together, these findings suggest that the bgl23-D mutation could be a helpful genetic tool for functional analysis of ATCSLDs and manipulating plant growth.

Removal of GAGs Regulates Mechanical Properties, Collagen Fiber Formation, and Alignment in Tissue Engineered Meniscus.

The complex fibrillar architecture of native meniscus is essential for proper function and difficult to recapitulate in vitro. In the native meniscus, proteoglycan content is low during the development of collagen fibers and progressively increases with aging. In vitro, fibrochondrocytes produce glycosaminoglycans (GAGs) early in culture, in contrast to native tissue, where they are deposited after collagen fibers have formed. This difference in the timing of GAG production hinders the formation of a mature fiber network in such in vitro models. In this study, we removed GAGs from collagen gel-based tissue engineered constructs using chondroitinase ABC (cABC) and evaluated the effect on the formation and alignment of collagen fibers and the subsequent effect on tensile and compressive mechanical properties. Removal of GAGs during maturation of in vitro constructs improved collagen fiber alignment in tissue engineered meniscus constructs. Additionally, removal of GAGs during maturation improved fiber alignment without compromising compressive strength, and this removal improved not only fiber alignment and formation but also tensile properties. The increased fiber organization in cABC-treated groups also appeared to influence the size, shape, and location of defects in these constructs, suggesting that treatment may prevent the propagation of large defects under loading. This data gives another method of modulating the ECM for improved collagen fiber formation and mechanical properties in tissue engineered constructs.

Real meso-scale model reconstruction and tensile failure prediction of 3D braided PIP-SiCf/SiC composites considering pore defects

• A combination of parametric and statistical methods is proposed for the real meso-scale model reconstruction. • The finite element model of 3D6d braided SiCf/SiC composite considering yarn shape and pore defects is established. • To predict well the material tensile properties and failure mechanisms, the effect of the fiber damage and pore defects is considered. • The simulation values and damage modes are consistent with the experimental ones when the damage factor of the fiber bundle is 0.8. This paper presents the real meso-scale model reconstruction and tensile failure prediction of three-dimensional six-directional (3D6d) braided precursor impregnation and pyrolysis (PIP) SiC f /SiC composites. A combination of parametric and statistical methods is proposed, and the real meso-scale finite element model considering yarn shape and pore defects is built. The damage factor is introduced into the mixing law and the progressive damage model is developed. The results reveal that the simulation values and damage modes are consistent with the experimental ones when the damage factor is 0.8. In detail, the predicted errors of modulus, stress and strain are 3%, 9% and 5%, respectively. The main damage modes of 3D6d braided SiC f /SiC composites are braiding yarn shear fracture, fifth yarn tensile fracture, and matrix cracking. Furthermore, the initial damage firstly appears in the braiding yarn, and when the load is near the maximum value, the fifth yarn and matrix also occur the damage.

Atomistic Investigation of Interface Edge Defect in CoFeB/MgO Ferromagnetic Nano-Dots

The spin transfer torque (STT) based magnetic random access memory (MRAM) device has been seen as the next general storage device which can realize non-volatile storage and in-memory computing. However, the shape defect that comes from the fabrication process will influence the performance of the MRAM device. In this work, the relationship between the critical current and switching time and the interface edge defect in a CoFeB MRAM is studied by the atomistic spin model. The atomistic simulations that the effect of the interface edge defect depends on the thickness of the magnetic layer, the operating temperature and the driven current density. Our finding should also benefit the control of the fabrication process of the STT-based technologies.

A Material View on Extrinsic Magnetic Domain Wall Pinning in Cylindrical CoNi Nanowires

Speed and reliability of magnetic domain wall (DW) motion are key parameters that must be controlled to realize the full potential of DW-based magnetic devices for logic and memory applications. A major hindrance to this is extrinsic DW pinning at specific sites related to shape and material defects, which may be present even if the sample synthesis is well controlled. Understanding the origin of DW pinning and reducing it are especially desirable in electrochemically deposited cylindrical magnetic nanowires (NWs), for which measurements of the fascinating physics predicted by theoretical computation have been inhibited by significant pinning. We experimentally investigate DW pinning in CoxNi100–x NWs by applying quasi-static magnetic fields. Wire compositions were varied with x = 20, 30, and 40, while the microstructure was changed by annealing or by varying the pH of the electrolyte for deposition. We conclude that pinning due to grain boundaries is the dominant mechanism, decreasing inversely with both the spontaneous magnetization and grain size. Second-order effects include inhomogeneities in lattice strain and the residual magnetocrystalline anisotropy. Surface roughness, dislocations, and impurities are not expected to play a significant role in DW pinning in these wire samples.

Effect of Defects in Graphene/Cu Composites on the Density of States.

The process of handling and bonding copper (Cu) and graphene inevitably creates defects. To use graphene/Cu composites as electronic devices with new physical properties, it is essential to evaluate the effect of such defects. Since graphene is an ultrathin anisotropic material having a hexagonal structure, an evaluation of graphene/Cu composites containing defects was conducted taking into account the inherent structural characteristics. The purpose of this study is to evaluate defects that may occur in the manufacturing process and to present a usable basic method for the stable design research and development of copper/graphene composites essential for commercialization of copper/graphene composites. In the future, when performing analytical calculations on various copper/graphene composites and defect shapes in addition to the defect conditions presented in this paper, it is considered that it can be used as a useful method considering defects that occur during application to products of desired thickness and size. Herein, density functional theory was used to evaluate the behavior of graphene/Cu composites containing defects. The density of states (DOS) values were also calculated. The analysis was implemented using three kinds of models comprising defect-free graphene and two- and four-layered graphene/Cu composites containing defects. DOS and Fermi energy levels were used to gage the effect of defects on electrical properties.

Mesenchymal Stem Cell-Derived Exosomes Enhance 3D-Printed Scaffold Functions and Promote Alveolar Bone Defect Repair by Enhancing Angiogenesis.

The reconstruction of severe alveolar bone defects remains a complex and challenging field for clinicians. Three-dimensional-printed scaffolds can adapt precisely to the complicated shape of the bone defects, which is an alternative solution to bone tissue engineering. Our previous study constructed an innovative low-temperature 3D-printed silk fibroin/collagen I/nano-hydroxyapatite (SF/COL-I/nHA) composite scaffold with a stable structure and remarkable biocompatibility. However, the clinical translation of most scaffolds is limited by insufficient angiogenesis and osteogenesis. In this study, we investigated the effects of human umbilical cord mesenchymal-stem-cell-derived exosomes (hUCMSC-Exos) on bone regeneration, especially from the perspective of inducing angiogenesis. HUCMSC-Exos were isolated and characterized. In vitro, the effect of hUCMSC-Exos on the proliferation, migration, and tube formation of human umbilical vein endothelial cells (HUVECs) was examined. Moreover, the loading and release of hUCMSC-Exos on 3D-printed SF/COL-I/nHA scaffolds were evaluated. In vivo, hUCMSC-Exos and 3D-printed SF/COL-I/nHA scaffolds were implanted into alveolar bone defects, and bone regeneration and angiogenesis were investigated by micro-CT, HE staining, Masson staining, and immunohistochemical analysis. The results showed that hUCMSC-Exos stimulated HUVEC proliferation, migration, and tube formation in vitro, and the effect increased with increasing exosome concentrations. In vivo, the combination of hUCMSC-Exos and 3D-printed SF/COL-I/nHA scaffolds promoted alveolar bone defect repair by enhancing angiogenesis and osteogenesis. We constructed an elaborate cell-free bone-tissue-engineering system by combining hUCMSC-Exos with 3D-printed SF/COL-I/nHA scaffolds, potentially providing new ideas for treating alveolar bone defects.

PAEDID: Patch Autoencoder-based Deep Image Decomposition for pixel-level defective region segmentation

Unsupervised pixel-level defective region segmentation is an important task in image-based anomaly detection for various industrial applications. The state-of-the-art methods have their own advantages and limitations: matrix-decomposition-based methods are robust to noise, but lack complex background image modeling capability; representation-based methods are good at defective region localization, but lack accuracy in defective region shape contour extraction; reconstruction-based methods detected defective region match well with the ground truth defective region shape contour, but are noisy. To combine the best of both worlds, we present an unsupervised Patch AutoEncoder-based Deep Image Decomposition (PAEDID) method for defective region segmentation. In the training stage, we learn the common background as a deep image prior by a patch autoencoder network. In the inference stage, we formulate anomaly detection as an image decomposition problem with the deep image prior and sparsity regularizations. By adopting the proposed approach, the defective regions in the image can be accurately extracted in an unsupervised fashion. We demonstrate the effectiveness of the PAEDID method in simulation studies and an industrial dataset in the case study.

Tiny Machine Learning for High Accuracy Product Quality Inspection

The quality inspection of industrial products is a fundamental step in large-scale production as it boosts the yield and reduces the costs. Intelligent embedded platforms with built-in tiny machine learning (tinyML) algorithms and cameras can automate quality inspection; however, running complex deep learning algorithms in low-cost and low-power embedded devices is still challenging because of limited memory and energy resources. This article presents an innovative sensor system with three microcontroller unit (MCU)-based tinyML cameras capable of automatic artifact and anomaly detection in plastic components. The system consists of a top camera responsible for identifying shape defects and two side cameras for color anomalies. Data processing is executed locally with tinyML reducing data transmission to a few bytes. Two state-of-the-art convolutional neural network (CNN) architectures are evaluated, namely, MobileNetV2 and SqueezeNet. Results show how both the architectures—with appropriate compression techniques—are suitable to be evaluated by resource-constrained microcontrollers. The networks achieve 99% classification accuracy while maintaining suitable real-time performance, respectively, equal to 5 and 2 frames/s.

Inferential-Statistical Reevaluation of Spent Fuel Zircaloy Cladding Integrity

From the published results of experiments investigating the effects of delayed hydride cracking (DHC) on spent fuel Zircaloy cladding integrity, relevant data have been extracted and re-analyzed, taking advantage of inferential statistics and an information-theoretic model selection criterion. Statistical tolerance intervals, the method of maximum likelihood estimation, and the Akaike information criterion, corrected for small sample size, were applied to a small sample of measured values of the threshold stress-intensity factor . The purpose was to create a well-grounded probability density function for use in a mathematical model correlating random variates of with important conditions for the initiation of crack growth by DHC, specifically, cladding hoop stress and the depth and shape of surface flaws. A selection criterion purposely designed for small sample sizes and the robust nature of inferential statistics were ideally suited for the intended reevaluation. The fidelity of the mathematical model was protected by the exclusion of any simplifying approximations, e.g., substitution of constants or single-valued descriptive statistics for variables. The probabilistic effect of the random variable was thereby precisely mapped onto the linearly correlated variable, threshold cladding hoop stress, as a function of surface flaw depth and shape. Contour plots of the results constitute significant improvements over previous quantitative single-point estimates of the effects of DHC on spent fuel Zircaloy cladding integrity.

Effect of defects in adhesive layer on the interfacial bond behaviors of externally bonded CFRP-to-concrete joints

The structural strengthening technique using fiber-reinforced polymer (FRP) composites to strengthen concrete structures has been widely used. For this reinforcement technique, the stress transfer mechanism at the FRP-concrete interface is one of the key factors affecting the reinforcement effectiveness. In practice, the bond layer is prone to defects due to uneven concrete surface, environmental deterioration or external loading during service life, which can weaken the bonding performance of the FRP-concrete system and even lead to unexpected structural failure in severe cases. In this paper, an experimental study on the influence of interface defects on FRP-concrete interface bonding behavior is designed. Based on the digital image correlation (DIC) technique, the effects of defect size, shape and location on the FRP interfacial bonding behavior were investigated separately. It was found that when the defect area was less than 5% of the effective bond area, the reduction of debonding load was similar to the fluctuation of the conventional interfacial debonding process, and the effect of defects on the interfacial bonding performance was relatively limited. The results also show that the degree of influence of defects on bonding performance varies for different configurations. The higher the concentration of defects, the higher the decrease in the FRP-concrete bearing capacity, especially those defects closer to the FRP boundary. Meanwhile, the proposed loading capacity reduction model considering the defect area is compared with the experimental results to assess the severity degree of defects.

Highly elastic and self-healing nanostructured gelatin/clay colloidal gels with osteogenic capacity for minimally invasive and customized bone regeneration

Extrusible biomaterials have recently attracted increasing attention due to the desirable injectability and printability to allow minimally invasive administration and precise construction of tissue mimics. Specifically, self-healing colloidal gels are a novel class of candidate materials as injectables or printable inks considering their fascinating viscoelastic behavior and high degree of freedom on tailoring their compositional and mechanical properties. Herein, we developed a novel class of adaptable and osteogenic composite colloidal gels via electrostatic assembly of gelatin nanoparticles and nanoclay particles. These composite gels exhibited excellent injectability and printability, and remarkable mechanical properties reflected by the maximal elastic modulus reaching ∼150 kPa combined with high self-healing efficiency, outperforming most previously reported self-healing hydrogels. Moreover, the cytocompatibility and the osteogenic capacity of the colloidal gels were demonstrated by inductive culture of MC3T3 cells seeded on the three-dimensional (3D)-printed colloidal scaffolds. Besides, the biocompatibility and biodegradability of the colloidal gels was proved in vivo by subcutaneous implantation of the 3D-printed scaffolds. Furthermore, we investigated the therapeutic capacity of the colloidal gels, either in form of injectable gels or 3D-printed bone substitutes, using rat sinus bone augmentation model or critical-sized cranial defect model. The results confirmed that the composite gels were able to adapt to the local complexity including irregular or customized defect shapes and continuous on-site mechanical stimuli, but also to realize osteointegrity with the surrounding bone tissues and eventually be replaced by newly formed bones.

Numerical study of the effect of infilling material in a hole-like flaw on acoustic emission characteristics and damage evolution in sandstone

Per-existing flaws are wide existence in rock mass, which have a significant effect on mechanical properties and fracturing behaviors. To investigate the effects of flaw type, infilling material, and flaw shape on the mechanical properties and fracture evolution of rock, a series of uniaxial compression tests were conducted on sandstone specimens. The bonded-particle model (BPM) and acoustic emission (AE) techniques were adopted to study the cracking process and AE characteristics of sandstone. The results indicate that relatively rigid infilling can significantly improve the mechanical properties of the rock. The AE simulation in the BPM revealed its fracture evolution, and there is a good consistency between fracture evolution and AE events. The cracking process and crack type around the hole-like flaw are influenced by the type of infilling material and its shape due to changes in the stress state around the hole-like flaw. Tensile cracks in nature usually initiate from the top or the bottom of a hole-like flaw. The shear and compaction cracks in nature prefer to initiate within the infilling material or from the lateral sides of the hole-like flaw. The cracking process of rocks can be characterized by the b-value, the variation of which is comparable with fracture behaviors and energy dissipation.

LFRNet: Localizing, Focus, and Refinement Network for Salient Object Detection of Surface Defects

Salient object detection of surface defects is one of the surface defect detection tasks, which aims at highlighting the defect regions from the surface of strip steel, magnetic tale, road, and so on. However, the performance of existing methods degrades dramatically when dealing with complex scenarios such as low-contrast of defect regions and various defect shapes. Therefore, in this paper, we propose a novel saliency model, namely localizing, focus, and refinement network (LFRNet), which consists of the semantic-guided localizing module, the context-driven focus module, and the edge-aware refinement module. Firstly, the semantic-guided localizing module deploys the graph reasoning (GR) unit and global attention (GA) unit to localize the potential defect regions from a global view. Secondly, the context-driven focus module employs the split context (SC) unit and the mutual attention (MA) unit to perform the identification process via the introduction of spatial detail features. Lastly, to further improve the accuracy of the detection results, we deploy the edge-aware refinement (ER) module, which introduces the boundary cues via the edge generation (EG) unit and aggregates the localizing result, the focus results, and the edge information into the high-quality detection map. Extensive experiments on four public defect datasets clearly show the effectiveness and superiority of the proposed LFRNet, where the LFRNet obtains an improvement of 4.1%, 5.7%, 1.0%, and 0.8% on FM, WF, EM, and SM respectively when compared with the top-level method AEP.

Some specifics of the billet open and submerged stream casting

Experimental series of steel casting were carried out to determine the effect of the continuous casting method on the fre-quency of defects formation in billets. At the same time, the 125x125 mm square billets production was carried out on a six-strand caster with casting speeds of 2.8-4.3 m/min. In the process of research, a series of 5 melts of steel 60 (St60) was initially casted with an open stream. After that, using the same molds, 5 melts of steel 40X were sequentially casted with a submerged stream. As a result of an industrial experiment, it was found that because of the open stream continuous casting the number of rhomboidal billets was 4.5 %. In the case of using the submerged stream technology defects in the billet shape were not observed. The use of the "stopper rod” — "tundish nozzle" — "submerged entry nozzle" system allows to control metal consumption during casting more accurate and timelier in automatic mode without changing the billet casting speed. At the same time, the position of the metal level in the mold is constantly ensured, which most accurately considers the features of the billet solid shell formation over the entire mold length at the current casting speed. The absence of the need to control the casting speed by the level of metal in the tundish makes it possible to maintain the optimal amount of melt in it, which sta-bilizes the casting temperature. The submerged casting prevents bubbling at the phase boundary in the mold and provides the possibility of a significant increase in the nozzle inner diameter (compared to the open stream) that reduces the rate of its clogging with corundum deposits several times. In addition, submerged casting allows the use of special mold powders, which improve the surface quality of the billets. It is possible to additionally improve billets quality using electromagnetic stirring of liquid steel during continuous casting.