Methods for Identifying and Classifying Reaction Products in Alkali-Activated Fly Ash and Slag Cementitious Systems

Abstract

This invention provides a method for identifying and classifying reaction products in an alkali-activated fly ash-slag cementing system, comprising: acquiring BSE and EDS images of an alkali-activated fly ash-slag cementing system sample; performing clustering based on pixel color features of the EDS image to obtain phase masks; fusing phase features of the BSE image into the EDS image, and extracting pore structures based on grayscale features of the BSE image in the same field of view as the EDS image; obtaining unreacted precursor phase masks (with pore masks removed) and reaction product phase masks (with pore masks removed) based on the phase masks and pore structures; performing superpixel segmentation on the reaction product phase masks to obtain the Mg / Al ratio of each superpixel; and performing cluster analysis on the Mg / Al ratio of each superpixel to achieve the identification and classification of slag reaction products, mixed reaction products, and fly ash reaction products. This invention can achieve refined classification of reaction products in an alkali-activated fly ash-slag cementing system, improving the accuracy of reaction product identification and classification.

Description

Technical Field

[0001] This invention relates to the field of phase characterization technology for alkali-activated fly ash and slag cementing systems, specifically, to a method for identifying and classifying reaction products of alkali-activated fly ash and slag cementing systems. Background Technology

[0002] Accurate characterization of the content and distribution of each phase and pore is a key step in studying the evolution of the microstructure of cementitious material systems and analyzing the development process of their mechanical and durability properties. This is of great significance for revealing the reaction mechanism of cementitious material systems and exploring their performance regulation mechanism.

[0003] Existing quantitative analysis methods based on BSE images of alkali-activated cementitious material systems identify and distinguish pores, reaction products, and unreacted precursors by dividing pixel grayscale thresholds, thereby characterizing pore size and phase content and distribution. However, BSE image quantitative analysis methods only classify phases based on grayscale values, making it difficult to distinguish phases with overlapping grayscale ranges and unable to precisely differentiate different types of reaction products. Quantitative analysis methods based on EDS images of alkali-activated cementitious material systems can achieve accurate phase identification based on elemental characteristics. However, due to the large interaction volume between characteristic X-rays and the sample, EDS images are severely affected by noise, resulting in lower spatial resolution than BSE images. Furthermore, acquiring high-quality EDS images is time-consuming, requires sophisticated equipment, and presents significant challenges in image analysis.

[0004] A search revealed Chinese invention patent application CN105241904A, which discloses a method for phase analysis of fly ash based on energy dispersive X-ray spectroscopy. The specific steps are: (1) analyzing the chemical composition of fly ash through chemical analysis or X-ray fluorescence spectroscopy; (2) analyzing the mineral composition of fly ash through X-ray diffraction; (3) determining the types of elements for which energy dispersive X-ray spectral distribution images need to be acquired; (4) acquiring backscattered electron images and elemental energy dispersive X-ray spectral distribution images of fly ash, and performing energy dispersive X-ray point analysis on the main phases; (5) designing a phase analysis method, comprehensively processing the energy dispersive X-ray distribution images of each element in the same area, and then removing noise from the analysis results to obtain a phase-separated pseudo-color image of fly ash and phase analysis results. This patent overlays multiple elemental energy dispersive X-ray distribution images of the same test area to determine the phase type of each pixel in the test area. The accuracy depends on the effectiveness of the image processing algorithm, and the quantitative analysis capability is limited, especially for complex phases.

[0005] Therefore, in order to address the bottlenecks faced by BSE and EDS images in the analysis and application of alkali-activated cementitious material systems, it is urgent to establish a BSE-EDS image collaborative analysis method to achieve accurate identification and classification of pores, different types of reaction products, and unreacted precursors in alkali-activated cementitious systems. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method for identifying and classifying reaction products in an alkali-activated fly ash slag cementing system.

[0007] This invention is achieved through the following technical solution:

[0008] This invention provides a method for identifying and classifying reaction products of an alkali-activated fly ash-slag cementing system, comprising:

[0009] Obtain BSE and EDS images of samples from alkali-activated fly ash slag cementing systems;

[0010] Clustering is performed based on the pixel color features of the EDS image to obtain phase masks; the phase features of the BSE image are fused into the EDS image, and the aperture structure is extracted based on the grayscale features of the BSE image in the same field of view as the EDS image.

[0011] Based on the phase masks and the pore structure, the unreacted precursor phase mask with pores removed and the reaction product phase mask with pores removed are obtained.

[0012] The reaction product phase mask with the pores removed is segmented into superpixels to determine the Mg / Al ratio of each superpixel;

[0013] Cluster analysis is performed on the Mg / Al ratio of each superpixel to identify and classify slag reaction products, mixed reaction products, and fly ash reaction products.

[0014] Further, clustering is performed based on the pixel color features of the EDS image, including:

[0015] A guided filter is used, with the BSE image as the guide image, to perform guided filtering on the qualitative element surface spectrum of the same field of view.

[0016] A color EDS image is synthesized from grayscale element surface spectra, and the pixel values ​​of the color EDS image are clustered using a Gaussian mixture clustering algorithm.

[0017] Furthermore, the pixel values ​​of the color EDS image are clustered using the Gaussian mixture clustering algorithm, wherein the clustering analysis is performed with a cluster number of 5.

[0018] Furthermore, the extraction of the pore structure based on the grayscale features of the BSE image in the same field of view as the EDS image includes: extracting the pore distribution by applying the tangent slope method on the BSE image.

[0019] Furthermore, the step of extracting pore distribution from the BSE image using the tangent slope method includes: using the threshold corresponding to the first inflection point of the frequency distribution curve of the BSE image as the upper threshold of the pore grayscale, and identifying and extracting the pore distribution in the BSE image.

[0020] Further, based on the phase masks and the pore structure, an unreacted precursor phase mask (without the pore mask) and a reaction product phase mask (without the pore mask) are obtained, including:

[0021] Based on the phase type of each phase mask, phase masks of the same phase type are merged to obtain unreacted precursor masks and reaction product masks in the alkali-activated fly ash slag cementing system.

[0022] The pore mask extracted by BSE image is subtracted from the unreacted precursor phase mask and the reaction product phase mask to achieve a visualized and quantitative characterization of the pore content and distribution of reaction products and unreacted precursors in the alkali-activated fly ash slag cementing system.

[0023] Further, the reaction product phase mask with the pores removed is segmented into superpixels to determine the Mg / Al ratio of each superpixel. This includes: converting the reaction product phase mask of the alkali-activated fly ash slag cementing system into a superpixel representation using a simple linear iterative clustering method, and obtaining the Mg / Al ratio of each superpixel based on the average Mg element signal intensity and Al signal intensity within each superpixel range.

[0024] Furthermore, cluster analysis is performed on the Mg / Al ratio of each superpixel, including: using the GMM clustering algorithm to perform cluster analysis on the Mg / Al ratio of each superpixel, and using the intersection of adjacent Gaussian functions obtained by fitting as the threshold for dividing the reaction products, so as to characterize the content and distribution of the reaction products.

[0025] Furthermore, the GMM clustering algorithm is used to perform cluster analysis on the Mg / Al ratio of each superpixel, wherein the cluster analysis is performed with a cluster number of 3.

[0026] Furthermore, the step of using the intersection of adjacent fitted Gaussian functions as the threshold for dividing the reaction products, thereby characterizing the content and distribution of the reaction products, includes:

[0027] Reaction products with a Mg / Al ratio less than the first threshold are defined as fly ash reaction products.

[0028] Reaction products with a Mg / Al ratio between the first and second thresholds are defined as mixed reaction products.

[0029] Reaction products with a Mg / Al ratio greater than the second threshold are defined as slag reaction products.

[0030] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0031] 1. This invention utilizes a BSE-EDS image joint analysis method to achieve elemental color clustering analysis from EDS images and extract pore structure from BSE images in the same field of view as EDS images, thereby comprehensively realizing the quantitative characterization of the cementitious material system and solving the problem of difficulty in pore identification from low-resolution EDS images; and through a phase composition research method based on superpixel segmentation, it achieves refined classification of reaction products in the alkali-activated cementitious system of fly ash slag.

[0032] 2. This invention uses pixel color features based on EDS images for clustering, which can significantly optimize the clustering effect and the accuracy of phase identification, and realize the automated identification and classification of phases. By fusing the phase features of BSE images into the corresponding EDS images, detection noise interference can be reduced, and more distinguishable and information-rich elemental surface spectra can be provided, which is conducive to improving the accuracy of identification and classification of reaction products in alkali-activated fly ash slag cementing systems. Attached Figure Description

[0033] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0034] Figure 1 This is a flowchart illustrating a method for identifying and classifying reaction products in an alkali-activated fly ash slag cementing system according to an embodiment of the present invention.

[0035] Figure 2 The image shows the BES image of the sample in Example 1 of this invention.

[0036] Figure 3 The above are the EDS images of the samples in Example 1 of this invention;

[0037] Figure 4 This is a schematic diagram illustrating the visualization and quantitative characterization of pore content and distribution in Embodiment 1 of the present invention;

[0038] Figure 5 This is a schematic diagram illustrating the visualization and quantitative characterization of the content and distribution of unreacted precursors in Example 1 of the present invention;

[0039] Figure 6 This is a schematic diagram illustrating the visualization and quantitative characterization of the content and distribution of reaction products in Example 1 of the present invention;

[0040] Figure 7 The image shows the BES image of the sample in Example 2 of this invention.

[0041] Figure 8 The above are the EDS images of the samples in Example 2 of this invention;

[0042] Figure 9 This is a schematic diagram illustrating the visualization and quantitative characterization of pore content and distribution in Embodiment 2 of the present invention;

[0043] Figure 10 This is a schematic diagram illustrating the visualization and quantitative characterization of the content and distribution of unreacted precursors in Example 2 of the present invention.

[0044] Figure 11 This is a schematic diagram illustrating the visualization and quantitative characterization of the content and distribution of reaction products in Example 2 of the present invention;

[0045] Figure 12 The image shows the BES image of the sample in Example 3 of this invention.

[0046] Figure 13 The above are the EDS images of the samples in Example 3 of this invention;

[0047] Figure 14 This is a schematic diagram illustrating the visualization and quantitative characterization of pore content and distribution in Embodiment 3 of the present invention;

[0048] Figure 15 This is a schematic diagram illustrating the visualization and quantitative characterization of the content and distribution of unreacted precursors in Example 3 of the present invention;

[0049] Figure 16 This is a schematic diagram illustrating the visualization and quantitative characterization of the content and distribution of reaction products in Example 3 of the present invention. Detailed Implementation

[0050] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention. These all fall within the scope of protection of the present invention.

[0051] The pixel values ​​of a qualitative elemental spectrum reflect the elemental signal intensity at that pixel, while the pixel values ​​of a quantitative elemental spectrum reflect the relative abundance of the element at that pixel. Qualitative elemental spectra have a relatively short acquisition time and are generally only used for simple determination of elemental distribution, with limited characterization of phases. Quantitative elemental spectra have a longer acquisition time, better image quality, and superior characterization of phases compared to qualitative elemental spectra, but data analysis is more difficult and the application threshold is higher. To improve the flexibility and simplicity of image analysis, solve the problem of noise interference caused by insufficient scanning time, improve the resolvability and integrity of phases in low-quality EDS images, enhance the accuracy of elemental signal characterization, effectively solve the problem of sparse and discrete elemental signal distribution, and achieve visualized, quantitative, and global characterization of the phase composition and microstructure of cementitious material systems, this invention provides a method for identifying and classifying reaction products of an alkali-activated fly ash slag cementing system based on qualitative EDS elemental spectra.

[0052] Reference Figure 1 The flowchart shown illustrates an embodiment of the present invention that provides a method for identifying and classifying reaction products in an alkali-activated fly ash slag cementing system. This method utilizes backscattered electron (BSE) images and energy dispersive X-ray spectroscopy (EDS) images to identify and extract reaction products from the alkali-activated fly ash slag cementing system. The method includes:

[0053] S1. Obtain BSE and EDS images of samples from alkali-activated fly ash slag cementing systems;

[0054] S2. Clustering is performed based on the pixel color features of the EDS image to obtain the phase mask for each phase; the phase features of the BSE image are fused into the EDS image, and the aperture structure is extracted based on the grayscale features of the BSE image in the same field of view as the EDS image.

[0055] S3. Based on each phase mask and pore structure, obtain the unreacted precursor phase mask after removing the pore mask and the reaction product phase mask after removing the pore mask.

[0056] S4. Perform superpixel segmentation on the reaction product phase mask after removing the pore mask, and determine the Mg / Al ratio of each superpixel;

[0057] S5. Perform cluster analysis on the Mg / Al ratio of each superpixel to identify and classify slag reaction products, mixed reaction products, and fly ash reaction products.

[0058] In some embodiments, to obtain an EDS image of an alkali-activated fly ash slag cementing system sample, step S1 includes the following steps:

[0059] Step S11: Fix the prepared alkali-activated fly ash slag cementing system sample onto the sample holder. During the placement into the sample chamber, take care not to touch the objective lens pole piece or the backscatter signal detector. Ensure the sample chamber reaches a high vacuum, typically 5 × 10⁻⁶. -5 After Pa, switch to low vacuum mode to maintain the air pressure inside the chamber at around 80 Pa;

[0060] Step S12: Turn on the electron gun, adjust the accelerating voltage to 15kV, and the electron beam spot size to 5. Observe the sample position in the CCD camera window, raise the sample height to the optimal working distance, which is generally 7mm; ensure that the X-ray count rate is 15,000-20,000 per second, the processing time is set to 4 microseconds (μs), and the dead time during the energy spectrum acquisition process does not exceed 20%.

[0061] Step S13: Export the qualitative elemental surface spectrum (EDS) image of the alkali-activated fly ash slag cementing system in JPG format.

[0062] In this embodiment of the invention, the phase features of the BSE image are fused into the EDS image to obtain the phase masks for each phase through cluster analysis, and the pore structure is extracted.

[0063] In some implementations, in step S2, clustering is performed based on the pixel color features of the EDS image, including:

[0064] S21. Using the guided filter in the classic image fusion filter, with the BSE image as the guided image, guided filtering is performed on the qualitative element spectrum (EDS image) of the same field of view region. For example, the pixel neighborhood diameter of the filter kernel is set to 3 and the normalization coefficient is set to 0.1.

[0065] S22. Using the "Color Channel Merge" tool in ImageJ (calcium-red, silicon-blue, aluminum-green, magnesium-white, iron-yellow), the grayscale elemental spectra, i.e., the grayscale distribution maps of each element after phase mask processing, are synthesized into a color EDS image. Compared with the EDS image obtained in step S1, the synthesized color EDS image significantly improves resolution and element boundary clarity. Gaussian Mixture Model (GMM) clustering algorithm is used to perform cluster analysis on the pixel values ​​(RGB values) of the color EDS image. The number of clusters is set according to the known phase types, complexity, and the interpretability requirements of the algorithm. Preferably, to accurately classify different phases and ensure that each phase has a clear classification, effectively distinguishing the main phase from noise or porosity, a cluster number of 5 is used for cluster analysis. In other embodiments, other cluster numbers can be used depending on the actual situation of the phases being studied. Thus, phase masks can be obtained.

[0066] By using a guided filter, the phase features of the BSE image are fused into the corresponding EDS image. Compared to the unprocessed image, the element boundary clarity is significantly improved, reducing detection noise interference and providing a distinguishable and information-rich elemental spectrum. Cluster analysis, used to compare elements before and after the reaction, reveals the distribution of unreacted precursors and reaction products. The image output uses different colors to correspond to different elements, clearly showing the phase outline distribution. Cluster analysis assigns phase labels to each pixel, and a phase mask creates a visualization tool based on this, displaying the location and extent of different phases through color, shape, or texture, which helps improve the accuracy of reaction product identification and classification.

[0067] Due to the limited resolution of EDS images, although guided filters can improve the resolution of EDS images, directly identifying pore distribution from the denoised EDS image may still be challenging. In some implementations, in step S2, pore structures are extracted based on the grayscale features of a BSE image in the same field of view as the EDS image. This includes: applying the tangent slope method to extract pore distribution on the BSE image acquired in step S1, and using the tangent-slope method to quantize the pore distribution on the BSE image. Specifically, this includes: using the threshold corresponding to the first inflection point of the frequency distribution curve of the BSE image as the upper threshold for pore grayscale; grayscale values ​​below this threshold are considered pore structures, thereby identifying and extracting pore distribution in the BSE image.

[0068] To address the problem that the large electron-sample interaction leads to severe degradation of the element signals in the acquired mapping by detection noise, this embodiment of the invention employs an image fusion guided filter for data denoising. Using a high-resolution BSE image as a guide, texture and edge features are fused into the element mapping, thereby obtaining a more obvious and distinguishable element distribution.

[0069] This invention employs BSE-EDS image analysis to extract phases from the alkali-activated fly ash-slag cementing system. Specifically, BSE images determine the pore distribution, while EDS images differentiate between reaction products and unreacted precursors in the alkali-activated cementing system. In the EDS images, unreacted fly ash rich in silicon and aluminum appears blue or fluorescent green. Some fly ash rich in iron may appear yellow. Due to the relatively high calcium and magnesium content in the slag, the EDS images of the slag appear pale red.

[0070] In some embodiments, in step S3, phase masks of the same phase type are merged according to the phase type to obtain unreacted precursor masks and reaction product masks in the alkali-activated fly ash slag cementing system. For example, the "Image Calculator" in ImageJ is used to verify the phase type to which the phase masks obtained by automatic clustering belong. Based on this, the "Image Calculator" in ImageJ is used to merge phase masks of the same phase type to realize the identification and differentiation of unreacted precursors and reaction products in the alkali-activated fly ash slag cementing system. The pore mask extracted by BSE image (i.e., the pore structure extracted in step S2) is subtracted from the unreacted precursor phase mask and the reaction product phase mask to realize the visualized and quantitative characterization of the content and distribution of pores, reaction products, and unreacted precursors in the alkali-activated fly ash slag cementing system.

[0071] In some implementations, in step S4, the phase mask of the reaction products of the alkali-activated fly ash slag cementing system is converted into a superpixel representation using the Simple Linear Iterative Clustering (SLIC) method, and superpixel segmentation is performed. For example, the phase mask of the reaction products of the alkali-activated fly ash slag cementing system is converted into a representation of about 500 superpixels using the superpixel segmentation algorithm SLIC. Based on the average Mg element signal intensity and Al element signal intensity of all pixels within each superpixel range, that is, the pixel grayscale values ​​of the qualitative Mg element spectrum and the qualitative Al element spectrum, the Mg / Al ratio of each superpixel is obtained.

[0072] In some implementations, in step S5, the GMM clustering algorithm is used to perform cluster analysis on the Mg / Al ratio of each superpixel. For example, to improve the clustering effect and balance the fineness of classification with interpretability, cluster analysis is performed with a cluster number of 3. The intersection of adjacent Gaussian functions obtained from the fitting is used as the threshold for classifying reaction products. Slag reaction products contain a relatively high amount of Mg, fly ash reaction products contain a relatively high amount of Al, and if it is a mixed reaction product, the content of Mg and Al is moderate. Specifically, reaction products with a Mg / Al ratio less than the first threshold are defined as fly ash reaction products; reaction products with a Mg / Al ratio between the first and second thresholds are defined as mixed reaction products; and reaction products with a Mg / Al ratio greater than the second threshold are defined as slag reaction products, thus characterizing the content and distribution of the three types of reaction products. In other implementations, other cluster numbers can be used, as long as the same function can be achieved.

[0073] In the above embodiments of the present invention, an unsupervised clustering algorithm, GMM, is used to perform cluster analysis on the RGB values ​​of the EDS image synthesized from the denoised qualitative elemental surface spectra based on different cluster numbers (such as 5 clusters in step S2 and 3 clusters in step S3). This significantly optimizes the clustering effect and the accuracy of phase identification, realizing automated phase identification and classification. Through the BSE-EDS image joint analysis method, pore structure is extracted from the BSE image in the same field of view as the EDS image, and elemental color cluster analysis is performed from the EDS image, comprehensively achieving quantitative characterization of the cementitious material system and solving the problem of difficulty in pore identification from low-resolution EDS images. Furthermore, through a phase composition research method based on superpixel segmentation, refined classification of reaction products in the alkali-activated cementitious system of fly ash and slag is achieved.

[0074] The above embodiments of the present invention combine the advantages of BSE and EDS images. By clustering based on the pixel color features of EDS images, the clustering effect and the accuracy of phase identification can be significantly optimized. By fusing the phase features of BSE images into the corresponding EDS images, detection noise interference can be reduced, and more distinguishable and information-rich elemental surface spectra can be provided.

[0075] The present application's solution will be explained below with reference to specific embodiments. Those skilled in the art will understand that the following embodiments are for illustrative purposes only and should not be construed as limiting the scope of the application. Where specific techniques or conditions are not specified in the embodiments, they are performed according to the techniques or conditions described in the literature in the field or according to the product instructions. Reagents or instruments used without specified manufacturers are all conventional products that can be obtained through commercial channels.

[0076] Example 1

[0077] Continue to refer to Figure 1 The method for identifying and classifying reaction products of the alkali-activated fly ash slag cementing system in this embodiment includes the following steps:

[0078] Step S1: Preparation and Image Acquisition of Alkali-Activated Fly Ash Slag Cementitious Material

[0079] Low-calcium fly ash and granulated blast furnace slag were used as solid precursors for preparing alkaline activated materials. Sodium silicate was used as the alkaline activator, with a modulus (silica / sodium oxide ratio) of 2. The alkaline-activated fly ash-slag gel was prepared as follows: the contents of the alkaline activator, fly ash, blast furnace slag, and deionized water were 6 g, 24.5 g, 10.5 g, and 13 g, respectively. The fly ash, blast furnace slag, and alkaline activator were first dry-mixed for 2 minutes, then deionized water was added and stirred for 3 minutes. The suspension was then cast into a rubber mold. The sample was demolded after 24 hours and sealed for 6 days.

[0080] During the curing period, 10mm × 10mm × 10mm fragments were cut from the core of the sample to ensure efficient termination of hydration. The sample was immersed in isopropanol for 7 days to prevent further hydration reaction, ensuring that the BSE and EDS images obtained by electron microscopy fully reflect the phase composition and microstructure of the cementitious material system at the specific curing age, while also removing pore solutions from the cementitious material system to avoid interference with image acquisition. Subsequently, the sample was vacuum dried in an oven at 40°C for 3 days to remove residual solvents.

[0081] Alkali-excited fly ash slag samples were imaged using a field emission scanning electron microscope equipped with a silicon-drifted EDS detector and EDS software. Images were acquired in low-vacuum imaging mode (80 Pa), with an accelerating voltage of 15 kV, a working distance of 7 mm, and a spot size of 5. The images were digitized to 1024 × 896 pixels, with a pixel area of ​​0.084 μm. 2 The image field of view was 301×264μm. The processing time was set to 4, with the dead time limited to within 20%. The scan time per frame was 100μs / pixel, with 5 frames scanned, for a total scan time of approximately 10 minutes. Based on the automatic scanning results and material composition knowledge, aluminum (Al), magnesium (Mg), calcium (Ca), iron (Fe), sodium (Na), and silicon (Si) were selected for analysis. The obtained BSE and EDS images are shown below. Figure 2 and Figure 3 As shown.

[0082] Step S2: Image clustering analysis

[0083] Specifically, the "Color Channel Merge" tool in ImageJ (calcium-red, silicon-blue, aluminum-green, magnesium-white, iron-yellow) was used to synthesize the elemental surface spectra into a color EDS image. Since the signal intensity of sodium is low, its influence is not considered during the channel merging process. Gaussian mixture clustering was used to perform cluster analysis on the RGB values ​​of the color EDS image with a cluster size of 5.

[0084] Step S3: Phase Extraction

[0085] Phase masks of the same phase type are merged to identify and distinguish unreacted precursors and reaction products in the alkali-activated fly ash slag cementing system. The tangent-slope method is used to quantify pore distribution on the BSE image. A schematic diagram of the visualized quantitative characterization of the content and distribution of pores, unreacted precursors, and reaction products is shown below. Figures 4-6 As shown.

[0086] Step S4: Superpixel segmentation

[0087] The SLIC algorithm was used to perform superpixel segmentation on the phase mask of the identified reaction products, and the average Mg element signal intensity and Al signal intensity of each pixel within the superpixel range of the reaction products were calculated.

[0088] Step S5: Classification of reaction products

[0089] Clustering algorithms were used to perform cluster analysis on the Mg / Al ratio of the superpixels of the reaction products in the alkali-activated fly ash-slag cementation system, enabling the identification and differentiation of slag reaction products, mixed reaction products, and fly ash reaction products. Reaction products with a Mg / Al ratio less than a first threshold were defined as fly ash reaction products; those with a Mg / Al ratio between the first and second thresholds were defined as mixed reaction products; and those with a Mg / Al ratio greater than the second threshold were defined as fly ash reaction products. In this embodiment, the first threshold for the Mg / Al ratio used to differentiate reaction products was 0.2578, and the second threshold was 0.3853.

[0090] Example 2

[0091] The difference between Example 2 and Example 1 is that: (1) the analytical field of view determined in step 1 is different; (2)

[0092] The first threshold for the Mg / Al ratio that distinguishes the reaction products determined in Example 2 is 0.1838, and the second threshold is 0.2491.

[0093] The BSE and EDS images obtained in this embodiment are as follows: Figure 7 and Figure 8 As shown in the diagram, a visual quantitative characterization of the content and distribution of pores, unreacted precursors, and reaction products is presented. Figures 9-11 As shown.

[0094] Example 3

[0095] The difference between Example 3 and Example 1 is that: (1) the analytical field of view determined in step 1 is different; (2) the first threshold for the Mg / Al ratio that distinguishes the reaction products determined in Example 3 is 0.2525 and the second threshold is 0.4223.

[0096] The BSE and EDS images obtained in this embodiment are as follows: Figure 12 and Figure 13 As shown in the diagram, a visual quantitative characterization of the content and distribution of pores, unreacted precursors, and reaction products is presented. Figures 14-16 As shown.

[0097] Based on the results of Examples 1-3, accurate identification results can be obtained using the methods in the embodiments of the present invention under different analysis fields. The methods in the above embodiments of the present invention have wide applicability and accuracy.

[0098] The embodiments of the present invention, through the joint analysis method of BSE-EDS images, achieve elemental color clustering analysis from EDS images and extract pore structure from BSE images in the same field of view as EDS images, thereby comprehensively realizing the quantitative characterization of the cementitious material system and solving the problem of difficulty in pore identification from low-resolution EDS images; through the phase composition research method based on superpixel segmentation, the reaction products in the alkali-activated cementitious system of fly ash slag are refined and classified.

[0099] The above embodiments of the present invention can significantly optimize the clustering effect and the accuracy of phase identification by clustering based on the pixel color features of EDS images, and realize the automated identification and classification of phases. By fusing the phase features of BSE images into the corresponding EDS images, detection noise interference can be reduced, and distinguishable and information-rich elemental surface spectra can be provided, which is conducive to improving the accuracy of identification and classification of reaction products of alkali-activated fly ash slag cementing system.

[0100] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various modifications or variations within the scope of the claims, which do not affect the essence of the present invention. The above preferred features can be used in any combination without conflict.

Claims

1. A method for identifying and classifying reaction products of an alkali-activated fly ash-slag cementing system, characterized in that, include: Obtain BSE and EDS images of samples from alkali-activated fly ash slag cementing systems; Clustering is performed based on the pixel color features of the EDS image to obtain phase masks; the phase features of the BSE image are fused into the EDS image, and the aperture structure is extracted based on the grayscale features of the BSE image in the same field of view as the EDS image. Based on the phase masks and the pore structure, the unreacted precursor phase mask with pores removed and the reaction product phase mask with pores removed are obtained. The reaction product phase mask with the pores removed is segmented into superpixels to determine the Mg / Al ratio of each superpixel; Cluster analysis is performed on the Mg / Al ratio of each superpixel to identify and classify slag reaction products, mixed reaction products, and fly ash reaction products; wherein: Clustering based on the pixel color features of the EDS image includes: A guided filter is used, with the BSE image as the guide image, to perform guided filtering on the qualitative element surface spectrum of the same field of view. A color EDS image is synthesized from grayscale element surface spectra, and the pixel values ​​of the color EDS image are clustered using a Gaussian mixture clustering algorithm. The reaction product phase mask with the pores removed is segmented into superpixels to determine the Mg / Al ratio of each superpixel. This includes: converting the reaction product phase mask of the alkali-activated fly ash slag cementing system into a superpixel representation using a simple linear iterative clustering method; and obtaining the Mg / Al ratio of each superpixel based on the average Mg element signal intensity and Al signal intensity within each superpixel range.

2. The method for identifying and classifying reaction products of the alkali-activated fly ash slag cementing system according to claim 1, characterized in that, The Gaussian mixture clustering algorithm is used to perform cluster analysis on the pixel values ​​of the color EDS image, wherein the cluster analysis is performed with a cluster number of 5.

3. The method for identifying and classifying reaction products of the alkali-activated fly ash slag cementing system according to claim 1, characterized in that, The method for extracting the pore structure based on the grayscale features of the BSE image in the same field of view as the EDS image includes: extracting the pore distribution on the BSE image using the tangent slope method.

4. The method for identifying and classifying reaction products of the alkali-activated fly ash slag cementing system according to claim 3, characterized in that, The method of extracting pore distribution from BSE images using the tangent slope method includes: using the threshold corresponding to the first inflection point of the frequency distribution curve of the BSE image as the upper threshold of pore grayscale, and identifying and extracting the pore distribution in the BSE image.

5. The method for identifying and classifying reaction products of the alkali-activated fly ash slag cementing system according to claim 1, characterized in that, Based on the phase masks and the pore structure, an unreacted precursor phase mask (without the pore mask) and a reaction product phase mask (without the pore mask) are obtained, including: Based on the phase type of each phase mask, phase masks of the same phase type are merged to obtain unreacted precursor masks and reaction product masks in the alkali-activated fly ash slag cementing system. The pore mask extracted by BSE image is subtracted from the unreacted precursor phase mask and the reaction product phase mask to achieve a visualized and quantitative characterization of the pore content and distribution of reaction products and unreacted precursors in the alkali-activated fly ash slag cementing system.

6. The method for identifying and classifying reaction products of the alkali-activated fly ash slag cementing system according to claim 1, characterized in that, Cluster analysis is performed on the Mg / Al ratio of each superpixel, including: using the GMM clustering algorithm to perform cluster analysis on the Mg / Al ratio of each superpixel, and using the intersection of adjacent Gaussian functions obtained by fitting as the threshold for dividing the reaction products, so as to characterize the content and distribution of the reaction products.

7. The method for identifying and classifying reaction products of the alkali-activated fly ash slag cementing system according to claim 6, characterized in that, The GMM clustering algorithm was used to perform cluster analysis on the Mg / Al ratio of each superpixel, with a clustering number of 3.

8. The method for identifying and classifying reaction products of the alkali-activated fly ash slag cementing system according to claim 6, characterized in that, The step of using the intersection of adjacent Gaussian functions obtained from the fitting as the threshold for dividing the reaction products, thereby characterizing the content and distribution of the reaction products, includes: Reaction products with a Mg / Al ratio less than the first threshold are defined as fly ash reaction products. Reaction products with a Mg / Al ratio between the first and second thresholds are defined as mixed reaction products. Reaction products with a Mg / Al ratio greater than the second threshold are defined as slag reaction products.

Images