Crack Region Research Articles

Characteristics of Solidification Cracking for Ni-P Coated Cu-Al5052 Single-Mode Fiber Laser Welds

This study fundamentally examines the characteristics of solidification cracking in Cu-Al5052 single-mode fiber laser welding by utilizing uncoated and Ni-P coated Cu plates for the high durability busbar welding of pouch-type lithium-ion battery packs. Weld solidification cracking was confirmed regardless of Ni-P coating for both weld materials. However, there were differences in the regions of crack occurrence. Unlike the uncoated Cu-Al5052 welds where cracks occurred randomly within the fusion zone, for the Ni-P coated Cu-Al5052 welds, solidification cracks were concentrated near the faying surface of the upper and lower materials, particularly near the fusion line. The effect of the Ni and P components on the solidification cracking susceptibility, i.e., the welding solidification temperature range, was considered. Diffusion-controlled Scheil’s solidification calculations revealed that the mushy zone range enlarged with the use of a Ni coating layer. It was confirmed that an extremely wide mushy zone range of 1060 K was calculated at the solidification cracking region. Furthermore, P was also identified as an element that expands the solidification temperature range based on the Cu-P binary system. Consequently, during the laser welding of Cu-Al5052 dissimilar materials, the behavior of solidification cracking in the welds is influenced by the composition of the coating material and the incorporated composition within the weld fusion zone. Hence, it is necessary to select the optimal coating material and layer thickness considering weld solidification cracking susceptibility.

Dadnet: dual-attention detection network for crack segmentation on tomb murals

Many tomb murals have punctate losses, cracks, and craquelure due to underground subsidence and changes in their physical support. Visual non-destructive detection techniques enable rapid assessment of how much tomb murals are affected by cracking, providing suggestions for their conservation. However, tomb murals are typically created by sketching outlines and then colored. Detailed sketches can easily interfere with crack detection work, requiring the use of deep learning network to better learn crack features and improve detection accuracy. At the same time the limited data of tomb mural presents a challenge to build a deep learning network. To address these issues, this paper introduces a novel dual-attention detection network (DADNet) for crack segmentation of tomb murals. In this work, a customized dataset is first constructed by collecting mural images from the Tang Dynasty tombs. Then the ConvNeXt framework serves as the basis for feature extraction, enhancing the process. Lastly, a dual-attention module utilizing neighborhood attention and biaxial attention is employed to accurately identify the crack regions. Neighborhood attention performs a local self-attention operation around the pixel point, addressing the limitations of self-attention. This approach significantly reduces computational demands as the image size increases. Biaxial attention performs attention calculations in the horizontal and vertical directions. This compensates for the limitation of neighborhood attention in capturing global dependencies. Our DADNet outperformed the competing methods, achieving the highest recorded scores of 78.95% for MIoU and 61.05% for the Jaccard index.

New insights into fracture and cracking simulation of quasi-brittle materials based on the NMMD model

The simulation of crack-induced failure is of vital importance for the safety and integrity assessment of solids and structures but still challenging. Being the foremost approach, fracture mechanics is plagued by the logical inconsistency in dealing with crack and non-cracked region, and to make matters worse, it has been contradicted by the ongoing experiments that the fracture energy is not a material constant irrelevant to geometry even for the unsophisticated situations. In this paper, based on the recently proposed nonlocal macro-meso-scale consistent damage (NMMD) model, it is demonstrated that the logical consistency of governing equations either on crack-tip or off crack-tip plays a crucial role in the complete solution of cracking problems. To this end, the central themes of fracture mechanics, including the definition of an idealized crack, the separated equilibrium laws and the cracking criteria attached to crack-tip, are firstly revisited. Then, after a concise description of NMMD model with emphasis on the geometrical transmission form micro- to macro-scale and the physical–mechanical translation from geometry to energy, it is theoretically proved that the geometry-based damage zone reproduces idealized crack as a limit when the nonlocal characteristic length, i.e., the influence radius in NMMD model, tends to zero. As a result, the separated equilibrium laws of fracture mechanics can be covered by NMMD model with a unified set of governing equations that does not distinguish between points at crack surface and at non-cracked region. This advantage permits NMMD model to be well-suited to analyze both crack-induced failure problems with and without pre-existing cracks in the absence of additional cracking criteria. In contrast, owing to its macro-meso-scale consistent feature, a new criterion for kinking angle of crack, the maximum stress intensity factor (SIF) of mode-I, can be derived from NMMD model. Some representative numerical examples are also presented to further confirm the above intrinsic link to fracture mechanics. In particular, the superiority of the NMMD model is demonstrated by capturing not only the non-constant fracture parameters (especially the fracture energy) but also the cracking initiation not from the pre-existing crack-tips, which are revealed by experimental results but contradict fracture mechanics. These investigations imply that the NMMD model could provide a potential avenue to understand when and how cracks nucleate and propagate based on a two-scale perspective.

Analytical solution of nano-concrete-epoxy interaction area considering static equilibrium

This research proposes an analytical solution of the nano-concrete-epoxy interaction area within nano crack region of the reinforced concrete beam by applying Newton’s third law in static equilibrium. For deriving the governing equation, the imaginary beam with free ends (no support) is considered within nano crack region. This imaginary beam is acted along the imaginary line of concrete-epoxy interface. Newton’s third law is applicable for deriving the governing equation because of assuming the absence of frictional and other external forces. The parametric study is performed for implementing the proposed formula of nano interactive area considering variable nano crack depths and thicknesses. The nano interactive area is increased gradually with the increment of depths and thicknesses based on the parametric study because of linear functionality of interactive area and geometry of nano crack region. The maximum interactive area is found to be 314 nm2 at 0.6 ratio of depths and thicknesses of the nano crack. The incremental differences in interactive area between the crack depth or thickness ratios of 0.1 and 0.6 are found to be 25.4% and 1.6% for variations of the crack depth and thickness ratios, respectively. So, the crack depth shows higher impact on the interaction area compared to the thickness of the crack. However, there is a scope for enhancing this research in future by deriving closed-formed analytical formulations to consider appropriate boundary conditions.

Concrete Crack Detection and Segregation: A Feature Fusion, Crack Isolation, and Explainable AI-Based Approach.

Scientific knowledge of image-based crack detection methods is limited in understanding their performance across diverse crack sizes, types, and environmental conditions. Builders and engineers often face difficulties with image resolution, detecting fine cracks, and differentiating between structural and non-structural issues. Enhanced algorithms and analysis techniques are needed for more accurate assessments. Hence, this research aims to generate an intelligent scheme that can recognize the presence of cracks and visualize the percentage of cracks from an image along with an explanation. The proposed method fuses features from concrete surface images through a ResNet-50 convolutional neural network (CNN) and curvelet transform handcrafted (HC) method, optimized by linear discriminant analysis (LDA), and the eXtreme gradient boosting (XGB) classifier then uses these features to recognize cracks. This study evaluates several CNN models, including VGG-16, VGG-19, Inception-V3, and ResNet-50, and various HC techniques, such as wavelet transform, counterlet transform, and curvelet transform for feature extraction. Principal component analysis (PCA) and LDA are assessed for feature optimization. For classification, XGB, random forest (RF), adaptive boosting (AdaBoost), and category boosting (CatBoost) are tested. To isolate and quantify the crack region, this research combines image thresholding, morphological operations, and contour detection with the convex hulls method and forms a novel algorithm. Two explainable AI (XAI) tools, local interpretable model-agnostic explanations (LIMEs) and gradient-weighted class activation mapping++ (Grad-CAM++) are integrated with the proposed method to enhance result clarity. This research introduces a novel feature fusion approach that enhances crack detection accuracy and interpretability. The method demonstrates superior performance by achieving 99.93% and 99.69% accuracy on two existing datasets, outperforming state-of-the-art methods. Additionally, the development of an algorithm for isolating and quantifying crack regions represents a significant advancement in image processing for structural analysis. The proposed approach provides a robust and reliable tool for real-time crack detection and assessment in concrete structures, facilitating timely maintenance and improving structural safety. By offering detailed explanations of the model's decisions, the research addresses the critical need for transparency in AI applications, thus increasing trust and adoption in engineering practice.

Measurement on the fatigue-healing performance of SARAs fractions in bitumen and its characterization by molecular simulations

The objective of this research was to investigate the healing characteristics of bitumen through the assessment of fatigue-healing performance of SARAs fractions and its characterization at the molecular level with SARAs fractions as the intermediate. Firstly, the flow behaviors and fatigue-healing performance of fractions were characterized by rheometer. Then, the diffusion systems were constructed and the diffusion with preset cracks was carried out, the evaluation of which was identified by density distribution, relative concentration in crack region, mean square displacement and interaction effect. The results indicate that aromatics exhibit the closest similarity to original bitumen in flowing behavior, which also reflected in the healing performance. The healing performance of resins and asphaltenes are extremely limited while that of saturates was comparable to fluid. Molecular simulation was further supporting to investigate the healing process. A platform period was found in the healing period when the molecules would move to seek the lowest energy state which was considered as healing in the alternative perspective. During the period, heavy components were more difficult to traverse while weak interaction between light components would shorten the period and identify the lowest energy state more easily, thereby greatly affecting the healing performance of fractions. Furthermore, heavy components are analogous to anchor points, among which resins were more readily aggregated and asphaltenes were relatively dispersed, while light components diffuse flexibly within them. The diffusion has the potential to drive the movement of heavy components which also kept the capture of other components, ultimately resulting in the formation of a stable healing state. The outcome with a discernible interaction effect would be important theoretical significance for the refinement of bitumen of better healing performance and the potential to influence the development of rejuvenators.

Chemical composition and bioactivity of Codium dwarkense collected from the earthquake crack region of Ras Muhammad National Park, Red Sea, Egypt

Ras Muhammad National Park, located at the southern tip of the Sinai Peninsula, Egypt, is well-known for its unique marine biodiversity. One interesting region of the park is the Earthquake Crack, a tidal-influenced inland cave formed by a 1968 earthquake, that known for its rich seaweed community, with dominance of Codium dwarkense. This study evaluates the bioactive compounds and bioactivities of C. dwarkense from the crack region. The sample were collected, processed, and assessed for hydrographic parameters, nutritional composition, minerals, bioactive compounds, and bioactivity. Results showed high levels of carbohydrates and significant mineral content, with potassium and calcium being predominant. C. dwarkense also showed considerable amounts of vitamin C (32.4 mg/100g), vitamin E (400 mg/100g), phenolics (250.08 mg GAE/g), and flavonoids (85.4 mg rutin/g). The HPLC profile was dominated with quercetin, apigenin, pyrogallol, zeaxanthin, and benzoic acid. The GC-MS analyses identified several fatty acids, which were dominated with Hexadecanoic acid and Oleic acid. Bioactivity assays demonstrated significant antimicrobial, antioxidant, antidiabetic, anti-inflammatory, and anticancer activities. This study highlights the influence of the unique features of the Earthquake Crack region on C. dwarkense ability to produce valuable bioactive compounds with diverse health benefits. These findings contribute to the recognition of C. dwarkense as a functional food and nutraceutical resource.

Acoustic and thermal response characteristics and failure mode of gas-bearing coal–rock composite structure under loading

Exploring the characteristics of the instability and failure processes of gas-bearing coal and rock is crucial for monitoring and predicting mine gas accidents. Thus, a real gas environment was simulated based on a self-developed gas–solid coupling infrared observation system. The acoustic–thermal response characteristics and failure mode of the gas-bearing coal–rock composite structure were studied. The results showed the following: (1) From the plastic stage, the average infrared radiation temperature of the coal increased significantly. The variances of differential infrared temperature (VDIRT) of the combination and coal started to mutate approximately 30 s before the peak, and the b value of the combination began to fluctuate frequently, while the VDIRT of rock remained approximately 2.128 × 10−4 throughout the process. (2) When stress was about to peak, a clear temperature boundary formed between coal and rock. Acoustic emissions with high energy were mainly concentrated at the interface and inside the coal. (3) The early plastic stage was dominated by high-frequency, low-amplitude events. In the post-peak stage and late plastic stage, the proportion of events with 80–90 dB amplitude rose, and there was a significant increase in low-frequency, high-amplitude events. (4) As the loading proceeded, the density and area gradually increased and tended to move toward the shear crack region. The distribution range of the rise time/amplitude expanded from 0–12 ms/V at the beginning of the loading to the range of 0–60 ms/V in the post-peak stage.

Pavement Crack Detection Using Fractal Dimension and Semi-Supervised Learning

Pavement cracks are crucial indicators for assessing the structural health of asphalt roads. Existing automated crack detection models depend on large quantities of precisely annotated crack sample data. The irregular morphology of cracks makes manual annotation time-consuming and costly, thereby posing challenges to the practical application of these models. This study proposes a pavement crack image detection method integrating fractal dimension analysis and semi-supervised learning. It identifies the self-similarity characteristics within the crack regions by analyzing pavement crack images and using fractal dimensions to preliminarily determine the candidate crack regions. The Crack Similarity Learning Network (CrackSL-Net) is then employed to learn the semantic similarity of crack image regions. Semi-supervised learning facilitates automatic crack detection by combining a small amount of labeled data with a large volume of unlabeled image data. Comparative experiments are conducted on two public pavement crack datasets against the HED, U-Net, and RCF models to comprehensively evaluate the performance of the proposed method. The results indicate that, with a 50% annotation ratio, the proposed method achieves high-precision crack detection, with an intersection over union (IoU) exceeding 0.84, which is close to that of U-Net. Visual analysis of the detection results confirms the method’s effectiveness in identifying cracks in complex environments.

Mechanism of healing, strength, and durability concepts of bacteria concrete - a review

Abstract Research work is going on to explore the concept of self-healing concrete. Outcome of the investigation reveals potential of bacteria concrete. Self-healing capacity imparted to the concrete by incorporating bacteria. Traditional methods reported in the literature such as epoxy injection, grouting, drilling, and plugging and gravity filling method have environmental impact, require more maintenance and are time consuming. The self-healing mechanism can be a better alternative to compensate these defects due to ecofriendly nature. Self-healing mechanism involves the production of CaCO3 precipitate deposites in crack region. Review paper focused on self-healing efficiency, water permeability, chloride penetration, compressive strength, flexural strength and recovery of water tightness, percentage weight loss of both conventional and bacterial concrete. For good environment tomorrow and better sustainability, self-healing conceptin concrete structure contributes promising technology.

Understanding the influence of boron in additively manufactured GammaPrint®-700 CoNi-based superalloy

Boron is commonly added to superalloys in small amounts to enhance creep resistance, but can lead to cracking at high concentrations, especially during the additive manufacturing process. Two variants of CoNi-based GammaPrint®-700 superalloy with different B contents (0.08 at% vs 0.16 at%) were printed via laser powder bed fusion (LPBF) with the same printing parameters, with only the high B alloy exhibiting solidification cracking. Atom probe tomography (APT) revealed stronger segregation behaviors in the high B alloy compared to the low B alloy at both the inter-dendritic regions and grain boundaries (GBs). The segregation behavior at inter-dendritic regions was well captured with Scheil simulation and can correlate with the existing cracking susceptibility index (CSI) on cracking tendencies, although high angle GBs are where cracking occurs according to electron backscatter diffraction (EBSD) measurements. Additionally, the extent of GB segregation was compared between the high B and low B alloy. Higher B additions led to significantly more GB B segregation in the high B alloy compared to the low B alloy. For the high B alloy, the cracked region of one GB exhibited higher levels of B compared to the uncracked region of the same GB. However, much higher B contents were also found in two other uncracked GBs in the high B alloy, which demonstrates that higher GB B concentrations are not fully responsible for the cracking. A much larger variance in GB B segregation content was found in the high B alloy compared to the low B alloy. These phenomena were explained with a solidification model with the GB segregation content expressed explicitly by a modified Langmuir-McLean equation. This model linked the GB segregation content with solidification undercooling, which can be used as quantitative cracking criteria for future builds.

Advanced Parallel Laser Line-Camera System for Automatic and Precise Crack Width Measurement

Accurate quantification of cracks is crucial in evaluating building conditions. However, current non-contact measurement technologies encounter challenges in measuring crack widths, particularly in hard-to-reach areas. This paper presents a novel portable parallel laser line-camera system that is specifically designed for automatic and precise measurement of crack widths from different angles and over long distances. The system is composed of a digital camera, parallel line laser diodes, and rigid positioning arms, forming a triangulation-based configuration. A newly developed image processing algorithm is at the core of this innovation, allowing for the extraction of accurate pixel scale information from laser strips, thus overcoming the challenge of pixel distortion in non-perpendicular photography. The U-Net algorithm is utilized to efficiently identify crack regions, followed by the Canny algorithm for meticulous extraction of crack features. By utilizing the pixel scale data obtained from the laser strips, the actual width of cracks can be accurately determined. The system's efficacy and precision have been validated through laboratory and field tests. This automatic, precise, and portable solution significantly advances the field of on-site crack measurement, offering substantial potential for practical applications in building condition assessments.

Automated intelligent measurement of cracks on bridge piers using a ring-climbing vision scanning operation robot

Detecting cracks on the surface of bridge piers is critical to the health condition of bridges. Aiming at the limitations of existing unmanned aerial vehicles (UAVs) and climbing robots, this research proposes a ring-climbing visual scanning operation robotic disease acquisition system, which achieves entire region, high-definition, and fast acquisition of cracks on bridge piers surfaces. In addition, a crack segmentation network based on an efficient self-attention mechanism and a refined segmentation algorithm are proposed to achieve accurate detection and quantification of cracks on bridge piers. Finally, an experimental prototype was built, and experiments were conducted. The results show that the maximum error of crack quantification is within 0.1 mm. The designed ring-climbing visual scanning operation robotic disease acquisition system can be applied to the surface of any pier with a 1.2 to 1.5 m diameter, which is of great significance for the automatic detection of the health condition of bridge piers.

Fatigue behavior of an airfoil with tolerable exfoliation cracks induced by dynamic corrosion: Experimental and numerical assessment

Pitting corrosion creates preferential sites for cracks formation by the exfoliation process, which grow due to the continuous action of flight loads, even those of constant amplitude (CA). Whether the find-fix protocol must repair all defects, some may go unnoticed due to their sizes or locations in the aircraft. For this reason, it is necessary to study the fatigue behavior of airfoils with defects (cracks), analyzing their growth rates and the affectation on residual strength. This research work studied experimental and numerically fatigue performance of an airfoil with initial defects caused by dynamic corrosion. Intentionally, single-edge cracks (SECs) formation was induced in three wing skin regions: leading and trailing edges and central zone. The pre-corroded airfoil was subjected to alternate drag and lift forces to assess the crack growth and take validation data as deflection and the load level. The low stress intensity factor (SIF) estimated by means of a 3D finite element (FE) numerical model indicated that cracks did not propagate. Even though the stress level in the crack tip remained under the fatigue limit, giving rise to infinity life, the degraded material residual strength was considerably reduced. Also, the estimations obtained from an FE 1D model considering the damage factor and the mechanical properties loss in the crack region corroborated the infinity life behavior. Numerical and experimental results could determine an allowable flaw size under CA load conditions. However, the load variation play an important role increasing crack growth rate up to six times.

Implantable sensing technology for civil engineering structures

Providing crack location and stiffness variation information using practical structural health monitoring technology is highly significant for in-service structures. Inspired by the fast development of implantable devices, this paper introduced an implantable sensing technology (IST) for health monitoring of a typical concrete member known as shear wall undergoing loading progression. Two collaborative monitoring steps were developed to acquire targeted damage information at different loading stages, the one was, a cylindrical concrete implantable module (CCIM) sensor-enabled IST was applied for monitoring and imaging the early cracks of the shear wall at the initial loading stage; and the other was, a concrete implantable bar (CIB) sensor-enabled IST was employed for monitoring and assessing the damage evolution at the mid-late loading stage. Experimental results showed that these two collaborative ISTs can accurately identify the early cracking regions of the shear wall and reveal its stiffness degradation processes throughout the loading.

Pavement Crack Detection Based on the Improved Swin-Unet Model

Accurate pavement surface crack detection is crucial for analyzing pavement survey data and the development of maintenance strategies. On the basis of Swin-Unet, this study develops the improved Swin-Unet (iSwin-Unet) model with the developed skip attention module and the residual Swin Transformer block. Based on the channel attention mechanism, the pavement crack region can be better captured while the crack feature channels can be assigned more weights. Taking advantage of the developed residual Swin Transformer block, the encoder architecture can globally model the pavement crack feature. Meanwhile, the crack feature information can be efficiently exchanged. To verify the pavement crack detection performance of the proposed model, we compare the training performance and visualization results with the other three models, which are Swin-Unet, Swin Transformer, and Unet, respectively. Three public benchmarks (CFD, Crack500, and CrackSC) have been adopted for the purpose of training, validation, and testing. Based on the test results, it can be found that the developed iSwin-Unet achieves a significant increase in mF1 score, mPrecision, and mRecall compared to the existing models, thereby establishing its efficacy in pavement crack detection and underlining its significant advancements over current methodologies.

Study on improving the crack self-healing width and depth based on microbial mineralization modified by Ca–Al bimetallic hydroxides

In-depth research was conducted on the application of an inorganic mineral composite microbiological remediation agent in mortar specimens with varying crack widths. Firstly, the remediation agent's efficacy in repairing cracks was validated through surface observations and macroscopic performance tests, including methods such as crack surface image capture, crack water permeability recovery rate testing, and crack ultrasonic velocity recovery testing. Subsequently, this study investigated whether the inorganic mineral and microbiological components in the remediation agent played a role by examining the repair products within the cracks. Material composition was identified using techniques such as Scanning Electron Microscopy (SEM), Energy Dispersive Spectroscopy (EDS), X-Ray Diffraction (XRD), Thermogravimetric Analysis (TGA), and Fourier Transform Infrared Spectroscopy (FTIR). The primary minerals identified were layered double hydroxides and calcium carbonate. Furthermore, it was found that CI− and SO42− ions were immobilized within the interlayers of layered double hydroxides, aligning with the expected objectives of the experiment, thus confirming the feasibility of this study. Finally, the self-healing mechanism of cracks in this experiment was elucidated through a schematic diagram, demonstrating that the use of layered double hydroxides as inorganic minerals improved the microenvironment of the crack area, lowered the pH in the crack region, providing a favorable environment for microbial growth and mineralization. Additionally, these minerals could adsorb erosive ions such as CI− and SO42−, thereby enhancing the durability of cement-based materials.

In situ observation of fatigue crack propagation in soda‐lime glass with initial crack and residual stress induced by Vickers indentation under four‐point bending

AbstractThe present study examined the four‐point bending fatigue properties of glass with an initial crack and residual stress induced by Vickers indentation. The fatigue properties of glass with an initial crack are influenced by the residual stress induced by Vickers indentation. Regardless of the presence of residual stress, the failure time under static fatigue is shorter than that under cyclic fatigue. In the region of a small crack, the fatigue crack aspect ratio for glass is smaller than that for metallic materials. The crack growth rate in glass with an initial crack initially decreases and then increases with an increase in the maximum stress intensity factor, Kmax. To investigate the effect of residual stress on the fatigue crack propagation behavior of soda‐lime glass, the in situ observation of fatigue crack propagation was conducted. Crack closure occurs in glass in which tensile residual stress induced by Vickers indentation was released by annealing.

Microstructure evolution of V-containing low carbon ultra-fine bainitic steel during medium-temperature tempering and strengthening-toughening mechanism

In the present study, the microstructure evolution and the corresponding mechanical properties of V-containing low carbon ultra-fine bainitic steel during tempering were studied. The results show that during the initial stages of tempering, part of retained austenite in V-containing ultra-fine bainite steel transforms into bainite. The stability of the retained austenite improves when the tempering period is increased to 5 h. Simultaneously, nano-sized VC precipitates are precipitated in the microstructure. On the one hand, these phases can give full play to the transformation-induced plasticity (TRIP) effect by modulating the carbon content of the retained austenite. On the other hand, they can also reduce the lattice distortion in the matrix and combine it with the lower dislocation density, thereby improving the plasticity of the specimen. At the same time, the fine precipitated phase can also effectively improve the toughness of the cleavage crack region in the transition zone, preventing the micro-cracks from generating at the interface of the precipitated phase and then extending to the adjacent matrix to cause fracture, so that the uniform elongation, total elongation, impact toughness at room temperature and impact toughness at low temperature of the 5 h specimen are as high as 8.8 ± 0.2 %、16.2 ± 0.7 %、113 ± 4 J/cm2 and 76 ± 2 J/cm2, respectively. Furthermore, the primary contribution of the strengthening mechanism in the microstructure has shifted from fine grain strengthening and dislocation strengthening to precipitation strengthening, and part of the retained austenite undergoes a martensitic transformation during cooling, resulting in the highest yield strength of 1086 MPa for the 5 h specimen. The tensile strength is effectively maintained at 1402 MPa by increasing the tempering period, which has good tempering resistance, to achieve the comprehensive performance optimization of bainitic steel.

Research on Computer Classification Algorithm of Concrete Crack Based on Deep Learning

In combination with transfer learning under the framework of Unet semantic partitioning, VGG16 pre-trained neural network enhanced encoder is used to extract multi-level high-level semantic information. The cross-entropy loss function is used to eliminate the imbalance between samples, and finally the crack shape is accurately semitone. Combining with the theory of computer vision, a quantitative calculation method of the crack region, length, width and other geometric characteristic parameters based on the binary segmentation template is proposed. Finally, the self-developed dam concrete crack image is taken as an example to carry out simulation and comparison test to check the correctness and superiority of the research results of this project. The research results will reveal that the crack recognition based on deep neural network can achieve a high recognition rate. The calculation results of fracture characteristic parameters meet the requirement of detection accuracy. The research results of this paper are expected to provide a new technical means for the quality control of dam concrete structure.