Hydraulic concrete freeze-thaw damage microstructure segmentation method, electronic equipment and medium

Abstract

The invention provides a hydraulic concrete freeze-thaw damage microstructure segmentation method, electronic equipment and a medium. The method comprises the following steps: acquiring concrete two-dimensional CT images of different freeze-thaw cycle times; and taking the concrete two-dimensional CT image as the input of the U-Net network model, and training the U-Net network model to obtain a concrete microstructure segmentation model. According to the method, the problem that high-density groups of hydrated silicate, calcium hydroxide, unhydrated clinker and the like belonging to a mortar phase are misclassified into coarse aggregates is effectively avoided, and the problems that a traditional threshold method is not obvious in dividing boundary and unstable in dividing result are solved. The method provides an effective means for quantifying the influence of freeze-thaw damage on a multi-scale pore structure in the concrete and establishing damage evolution of the concrete under the action of freeze-thaw cycle.

Description

Technical Field

[0001] The present invention belongs to the technical field of concrete microstructure analysis, and particularly relates to a method for segmenting the microstructure of frozen-thawed damaged hydraulic concrete, an electronic device, and a medium. Background Art

[0002] After hydraulic concrete undergoes different freeze-thaw cycles, its internal microstructure will change. Accurately capturing these changes is crucial for evaluating the frost resistance of hydraulic concrete. Currently, when using traditional threshold segmentation methods for microstructure analysis of hydraulic concrete, there is a problem that the threshold segmentation boundaries of each phase are not obvious. Specifically, when determining the threshold of each phase using the trough in the gray histogram of the CT image, the trough range is relatively large. Using the trial-and-error method to determine the threshold is not only inefficient, but also when segmenting the pore / crack phase, although a higher threshold can detect microcracks, some microcracks are still not successfully identified. At the same time, some pixels belonging to the mortar phase will be misclassified as the pore / crack phase, resulting in significant errors in the segmentation of the internal pores of hydraulic concrete and affecting the analysis results of the evolution law of the microstructure of frozen-thawed damaged hydraulic concrete. In addition, similar problems exist in the mortar phase and the aggregate phase. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for segmenting the microstructure of frozen-thawed damaged hydraulic concrete, an electronic device, and a medium to improve the accuracy of analyzing the evolution law of the microstructure of frozen-thawed damaged hydraulic concrete in view of the deficiencies of the prior art.

[0004] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0005] A method for segmenting the microstructure of frozen-thawed damaged hydraulic concrete includes the following steps:

[0006] S1. Obtain two-dimensional CT images of concrete with different numbers of freeze-thaw cycles;

[0007] S2. Preprocess the two-dimensional CT images of the concrete;

[0008] S3. Use the preprocessed two-dimensional CT images of the concrete as the input of the U-Net network model, and train the U-Net network model to obtain a concrete microstructure segmentation model.

[0009] The present invention creatively combines CT technology with U-Net image segmentation, effectively avoiding the problem of misclassifying high-density components belonging to mortar, such as hydrated silicate, calcium hydroxide, and unhydrated clinker, as coarse aggregates. At the same time, it can accurately identify capillary pores and microcracks that cannot be segmented by traditional threshold methods, solving the problems of unclear segmentation boundaries and unstable segmentation results of traditional threshold methods. The method of the present invention can accurately capture the changes in the internal microstructure of hydraulic concrete after different numbers of freeze-thaw cycles, providing an effective means for quantifying the influence of freeze-thaw damage on the multi-scale pore structure inside concrete and establishing a damage evolution model of hydraulic concrete under freeze-thaw cyclic loading.

[0010] Furthermore, the concrete microstructure includes an aggregate phase, a mortar phase, and a pore / crack phase.

[0011] In order to obtain clear CT images, in S2, a median filtering method is used to denoise the two-dimensional CT image of the concrete, and the denoised image is binarized.

[0012] Furthermore, the two-dimensional CT images of concrete with the same number of freeze-thaw cycles are input into the concrete microstructure segmentation model to obtain the aggregate phase, mortar phase, and pore / crack phase of the concrete with that number of freeze-thaw cycles.

[0013] Furthermore, the aggregate phase, mortar phase, and pore / crack phase of the concrete with the same number of freeze-thaw cycles are sequentially superimposed to obtain three-dimensional images of the aggregate phase, mortar phase, and pore / crack phase of the concrete.

[0014] Furthermore, based on the three-dimensional images of the aggregate phase, mortar phase, and pore / crack phase of concrete with different numbers of freeze-thaw cycles, the evolution process of the concrete microstructure after freeze-thaw damage is analyzed.

[0015] Furthermore, in S3, the Adam algorithm is used to optimize the concrete microstructure segmentation model; k-Fold cross-validation is used to evaluate the concrete microstructure segmentation model.

[0016] Based on the same inventive concept, the present invention also provides an electronic device, including:

[0017] One or more processors;

[0018] A memory storing one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement the steps of the method for segmenting the microstructure of freeze-thaw damaged hydraulic concrete.

[0019] Based on the same inventive concept, the present invention also provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps of the method for segmenting the microstructure of hydraulic concrete freeze-thaw damage are realized.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0021] The present invention effectively avoids the problem of misclassifying high-density groups belonging to the mortar phase, such as hydrated silicate, calcium hydroxide, and unhydrated clinker, as coarse aggregates.

[0022] The present invention can accurately identify capillary pores and microcracks that cannot be segmented by the traditional threshold method, and solves the problems of unclear segmentation boundaries and unstable segmentation results of the traditional threshold method.

[0023] The present invention can accurately capture the changes in the internal microstructure of hydraulic concrete after different numbers of freeze-thaw cycles, and provides an effective means for quantifying the influence of freeze-thaw damage on the multi-scale pore structure inside the concrete and establishing a damage evolution model of hydraulic concrete under freeze-thaw cyclic loading. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a schematic flow chart of the method for segmenting the microstructure of hydraulic concrete of the present invention;

[0025] Figure 2 is a schematic diagram of the three-dimensional reconstruction of the concrete microstructure of the present invention;

[0026] Figure 3 is a schematic diagram of the U-Net model structure;

[0027] Figure 4 is a comparison diagram of the segmentation results of the concrete microstructure by different methods;

[0028] Figure 5(a) is a schematic diagram of the internal pore structure of hydraulic concrete after 0 freeze-thaw cycles;

[0029] Figure 5(b) is a schematic diagram of the internal pore structure of hydraulic concrete after 100 freeze-thaw cycles;

[0030] Figure 5(c) is a schematic diagram of the internal pore structure of hydraulic concrete after 200 freeze-thaw cycles;

[0031] Figure 5(d) is a schematic diagram of the internal pore structure of hydraulic concrete after 300 freeze-thaw cycles;

[0032] Figure 5(e) is a schematic diagram of the internal pore structure of hydraulic concrete after 400 freeze-thaw cycles;

[0033] Figure 6 is a schematic diagram of the evolution of internal pores of hydraulic concrete under freeze-thaw cycles;

[0034] Figure 7 It is a schematic diagram of the evolution of the normalized pore content of hydraulic concrete with the number of freeze-thaw cycles. Specific implementation manners

[0035] The present invention will be described in detail below in conjunction with embodiments. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. For convenience of description, words such as "upper", "lower", "left", and "right" hereinafter only indicate the same directions as the upper, lower, left, and right directions of the drawings themselves, and do not limit the structure.

[0036] Embodiment

[0037] As Figure 1 shown, the method for segmenting the microstructure of hydraulic concrete in this embodiment includes the following steps:

[0038] (1) Prepare high-strength hydraulic concrete specimens and perform standard curing. Conduct rapid freeze-thaw tests on the concrete specimens, and regularly measure the mass, dynamic elastic modulus, etc. Perform CT scans on the hydraulic concrete specimens after different numbers of freeze-thaw cycles. Use the German Vtomexs microfocus X-ray CT system to scan the target specimens, set the voltage of the X-ray source to 120 kV, the tube current to 150 μA, the power to 18 W, the spatial resolution of the scanned image to 28.9 μm / voxel, and the number of pixels to 1024×1024.

[0039] The method for fabricating the hydraulic concrete specimens is as follows:

[0040] 1) Prepare the specimens. In terms of material selection, the cementitious materials are CEMI 42.5R Portland cement, class I fly ash, and silica fume, with densities of 3120 kg / m 3 , 2300 kg / m 3 and 2160 kg / m 3 respectively; the fine aggregate is basalt artificial sand with a density of 2740 kg / m 3 , and the particle size distribution conforms to the ASTM C33(2004) standard; the coarse aggregate is basalt crushed stone with a density of 2860 kg / m 3 , and is divided into continuous gradations of small stones (5-20 mm) and medium stones (20-40 mm) according to different particle sizes; the water-binder ratio is 0.30, and a polycarboxylate superplasticizer with a solids content of 25.5% and a water reduction rate of 25% and an air-entraining agent are incorporated.

[0041] 2) When preparing concrete, the slump and spread of the fresh concrete are 180 mm and 420 mm respectively, and the air content is 3.5%. A concrete beam with dimensions of 90×30×30 mm is cast using a plastic mold. Then it is cured. All specimens are demolded 24 hours after casting and cured in a standard curing room at 20±2°C and RH>90% for 28 days before relevant tests are carried out.

[0042] 3) A rapid freeze-thaw test is carried out. According to DL / T5150 - 2017 "Test Code for Hydraulic Concrete", a DDR-2 type rapid freeze-thaw testing machine is used to test the cast concrete specimens. The center temperature of the specimen is controlled between -20°C and 7°C, and the duration of one freeze-thaw cycle is 4 hours. The temperature control process strictly follows a specific curve. Every 50 freeze-thaw cycles, the concrete specimen is taken out of the rapid freeze-thaw box, the surface moisture is wiped off, and the mass, dynamic elastic modulus and microstructure (obtained by CT scanning) of the specimen are measured in turn. The relative mass loss and normalized dynamic elastic modulus are calculated respectively.

[0043] (2) The median filtering method is used to denoise the two-dimensional CT image, and the denoised image is binarized to obtain clear CT slices.

[0044] Median filtering is mainly used for noise reduction in image processing. Median filtering is a non-linear signal processing technology based on sorting statistics theory that can effectively suppress noise. The basic principle of median filtering is to replace the value of a point in a digital image or digital sequence with the median of the values of all points in a neighborhood of that point, making the surrounding pixel values closer to the true values, thereby eliminating isolated noise points. The method is to use a two-dimensional sliding template of a certain structure, sort the pixels in the template according to the pixel values, and generate a two-dimensional data sequence that monotonically increases (or decreases). The output of two-dimensional median filtering is g(x,y) = med{f(x - k,y - l),(k,l∈W)}, where f(x,y) and g(x,y) are the original image and the processed image respectively, and W is a two-dimensional template, usually a 3×3 or 5×5 area, and can also be of different shapes, such as linear, circular, cross-shaped, circular ring-shaped, etc.

[0045] Median filtering has a good effect on filtering impulse noise. Especially when filtering noise, it can protect the edges of the signal and prevent them from being blurred.

[0046] Image binarization is the process of setting the gray value of the pixel points on the image to 0 or 255, that is, making the whole image show an obvious black and white effect.

[0047] A binary image that can still reflect the overall and local features of the image is obtained by appropriately selecting the threshold for a grayscale image with 256 brightness levels; all pixels with grayscale greater than or equal to the threshold are determined to belong to a specific object, and their grayscale values are represented by 255. Otherwise, these pixel points are excluded from the object area, and the grayscale value is 0, indicating the background or an exceptional object area.

[0048] (3) Perform the segmentation of the concrete microstructure based on U-Net. The U-Net model is as Figure 3 shown. Manually label 20 CT images with a resolution of 1024×1024, of which 14 are used for training and 6 are used for testing. During the training and testing processes, these images are cut into 512×512, and data augmentation methods are adopted.

[0049] The U-Net model consists of an encoding part and a decoding part. During the encoding process, image features are extracted through iterative convolution and subsampling. During the decoding process, the target segmentation result is generated through iterative upsampling and convolutional feature mapping. The decoder provides information for decoding to repair detailed features through skip connections between the corresponding upsampling layer and downsampling layer.

[0050] (4) During the training process, the Adam algorithm is used to optimize the model, and k-Fold cross-validation is used to evaluate the model. For the segmentation of the target phase (such as aggregate, mortar, or pore / crack), the pixels belonging to the target phase are defined as positive (P o ), and the rest are negative (N e ). If the segmentation result of a pixel is correct, it is set to true (T r ); otherwise, it is set to false (F a ). The pixels in the labeled image and the predicted image can be divided into four categories, namely true positive (T r P o ), true negative (T r N e ), false positive (F a P o ) and false negative (F a N e ). Precision (P r ), recall (R), and the harmonic mean of precision and recall (F1) are used as evaluation indicators, calculated through formula (1), formula (2), and formula (3), and then the accuracy (A) of the prediction result of a certain phase is determined through formula (4).

[0051]

[0052] As shown in Table 1, the comparison of the segmentation accuracies of different phases of hydraulic concrete by the traditional threshold method and the U-Net segmentation method is listed. As Figure 4The results of the three-phase segmentation of hydraulic concrete based on the marking method, threshold method, and U-Net segmentation method are presented. The first row represents the pore / crack phase, the second row represents the aggregate phase, and the third row represents the mortar phase. The method of the present invention can effectively avoid misclassifying high-density components belonging to the mortar phase, such as hydrated silicate, calcium hydroxide, and unhydrated clinker, as coarse aggregates. The segmentation accuracy is approximately 30% higher than that of the threshold method. In addition, the method of the present invention can also accurately identify capillary pores and microcracks that cannot be segmented by the traditional threshold method, and the segmentation accuracy of the pore / crack is more than 10% higher than that of the threshold method.

[0053] Table 1 Comparison of the segmentation accuracy of different phases of hydraulic concrete by the traditional threshold method and the U-Net segmentation method

[0054]

[0055] (5) As Figure 2 shown, the accurately segmented three-phase (aggregate, mortar, pore / crack) slices are stacked in sequence to three-dimensionally reconstruct the three phases of concrete and quantitatively analyze the evolution of the microscopic structure of concrete after freeze-thaw damage.

[0056] Figures 5(a), 5(b), 5(c), 5(d), and 5(e) are the three-dimensional reconstruction images of concrete with different numbers of freeze-thaw cycles. As shown in Figures 5(a), 5(b), 5(c), 5(d), and 5(e), as the number of freeze-thaw cycles continues to increase, the freeze-thaw damage suffered by hydraulic concrete continues to intensify, and the number and volume of internal pores both increase significantly, and microcracks continue to develop and penetrate each other to form cracks with larger sizes. This is the main reason for the significant influence of freeze-thaw cycles on the performance of hydraulic concrete. In Figures 5(a), 5(b), 5(c), 5(d), and 5(e), blue represents pores and red represents cracks.

[0057] As Figure 6 and Figure 7 shown, as the number of freeze-thaw cycles increases, the growth rate of the internal pore / crack content of hydraulic concrete shows three stages, namely the initial stage I, the stable increase stage II, and the rapid increase stage III. From the above rules, it can be seen that freeze-thaw damage can be regarded as a special type of fatigue damage, where the cyclic load applied inside the concrete is caused by the freeze-thaw cycle of the pore water contained.

[0058] The method for segmenting the microscopic structure of hydraulic concrete based on U-Net in this embodiment can accurately segment the three phases of hydraulic concrete (aggregate, mortar, pore / crack), and can quantitatively analyze the evolution of the microscopic structure of concrete after freeze-thaw damage using the obtained accurate pore / crack information. This method is an effective tool for accurately segmenting each phase of hydraulic concrete and detecting changes in the internal pore structure caused by freeze-thaw damage.

[0059] This method can accurately capture the changes in the internal microstructure of hydraulic concrete after undergoing different freeze-thaw cycles, providing an effective means for quantifying the influence of freeze-thaw damage on the multi-scale pore structure inside the concrete and establishing a damage evolution model of hydraulic concrete under freeze-thaw cycle loading.

[0060] Another embodiment of the present invention provides an electronic device, including:

[0061] One or more processors;

[0062] A memory, on which one or more programs are stored. When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the microscopic structure segmentation method for freeze-thaw damage of hydraulic concrete.

[0063] In some implementations, the memory can be a high-speed random access memory (RAM: Random Access Memory), and may also include non-volatile memory, such as at least one disk memory.

[0064] In other implementations, the processor can be various types of general-purpose processors such as a central processing unit (CPU), a digital signal processor (DSP), etc., which are not limited herein.

[0065] Another embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the microscopic structure segmentation method for freeze-thaw damage of hydraulic concrete are implemented.

[0066] The content clarified in the above embodiments should be understood that these embodiments are only used to illustrate the present invention more clearly, rather than to limit the scope of the present invention. After reading the present invention, various equivalent forms of modification made by those skilled in the art fall within the scope defined by the appended claims of this application.

Claims

1. A microscopic structure segmentation method for freeze-thaw damage of hydraulic concrete, characterized in that It includes the following steps: S1. Obtain two-dimensional CT images of concrete with different numbers of freeze-thaw cycles; S2. Preprocess the two-dimensional CT images of the concrete; S3. Take the preprocessed two-dimensional CT images of the concrete as the input of the U-Net network model, train the U-Net network model, and obtain a concrete microstructure segmentation model.

2. The microscopic structure segmentation method for freeze-thaw damage of hydraulic concrete according to claim 1, wherein The concrete microstructure includes aggregate phase, mortar phase, and pore / crack phase.

3. The microscopic structure segmentation method for freeze-thaw damage of hydraulic concrete according to claim 2, characterized in that Input the two-dimensional CT images of the concrete with the same number of freeze-thaw cycles into the concrete microstructure segmentation model to obtain the aggregate phase, mortar phase, and pore / crack phase of the concrete with this number of freeze-thaw cycles.

4. The method for segmenting the microstructure of freeze-thaw damaged hydraulic concrete according to claim 3, characterized in that, Stack the aggregate phase, mortar phase, and pore / crack phase of the concrete with the same number of freeze-thaw cycles in sequence to obtain three-dimensional images of the aggregate phase, mortar phase, and pore / crack phase of the concrete.

5. The microscopic structure segmentation method for freeze-thaw damage of hydraulic concrete according to claim 4, wherein Analyze the evolution process of the concrete microstructure after freeze-thaw damage based on the three-dimensional images of the aggregate phase, mortar phase, and pore / crack phase of the concrete with different numbers of freeze-thaw cycles.

6. The method for segmenting the microstructure of freeze-thaw damaged hydraulic concrete according to any one of claims 1-5, characterized in that, In S2, the median filtering method is used to denoise the two-dimensional CT images of the concrete, and the denoised images are binarized.

7. The method for segmenting the microstructure of frost-thaw damage of hydraulic concrete according to any one of claims 1-5, characterized in that, In S3, the Adam algorithm is used to optimize the concrete microstructure segmentation model; k-Fold cross-validation is used to evaluate the concrete microstructure segmentation model.

8. An electronic device, characterized in that, It includes: One or more processors; A memory on which one or more programs are stored. When the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the steps of the method according to any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, It stores a computer program which, when executed by a processor, implements the steps of the method according to any one of claims 1-7.

Images