A strain field evolution and crack prediction method for fully tailings cemented backfill

Abstract

The present invention belongs to the technical field of tailings strain field evolution prediction, and discloses a strain field evolution and crack prediction method for a fully tailings cemented backfill, comprising the following steps: preparing a CPB specimen of the fully tailings cemented backfill, and conducting an expansion test during the CPB specimen preparation process; evaluating the uniaxial compressive strength of the CPB specimen, and monitoring the surface changes of the specimen during the test process through digital image correlation acquisition technology; establishing a computer vision recognition model based on the HSV color mode, analyzing the area proportion and distribution of strain regions of different degrees, and realizing a comprehensive analysis of the image; and determining a warning signal for crack prediction based on the analysis results of the strain field changes. The present invention adopts the above-mentioned strain field evolution and crack prediction method for a fully tailings cemented backfill to realize quantitative analysis of the strain evolution of the backfill, provide a prediction and warning for the damage risk of the backfill, and is of great significance for ensuring the safety and efficiency of mining operations.

Description

Technical Field

[0001] The present invention relates to the technical field of tailings strain field evolution prediction, and in particular to a method for predicting the strain field evolution and cracks of a full tailings cemented filling body. Background Art

[0002] Tailings are an inevitable byproduct of mineral resource extraction, and their safe disposal is a critical issue facing the mining industry. In modern underground mines, tailings are often used as the primary component of cemented paste backfill (CPB), which consists of dewatered tailings (70-85% solid weight), a hydraulic binder (3-7% dry paste weight), and mixed water (fresh or mine-processed). CPB provides secondary ground support for mining operations and improves the underground working environment, offering significant economic, safety, and environmental advantages, thus attracting widespread attention in the mining industry. As a means of controlling mining subsidence, the effectiveness of backfill in mining is closely related to the integrity and stability of the backfill material. Localized failure of backfill often leads to overall material instability. Therefore, studying the strain evolution and failure prediction of CPB in structural backfill mining under overburden loads is of great practical significance for the stability of the stope.

[0003] In recent years, scholars have conducted extensive research on the mechanical properties of CPB, using various techniques to investigate the evolution of pores and microcracks within rocks and backfill materials. To assess deformation of backfill samples, digital image correlation (DIC) uses a high-speed camera to capture speckle field images of the sample surface, obtaining corresponding local deformation and instability data. This effectively reflects the damage and failure processes of the rock and backfill materials. These studies have advanced our understanding of the mechanical properties and deformation evolution of paste backfill materials.

[0004] However, no specific quantitative evolution pattern of the strain field of the full tailings cemented backfill and crack identification prediction research based on digital image correlation has been found. Since the material composition, pore characteristics and chemical composition of paste backfill materials are different from those of natural rock and soil, it is of great significance to study the deformation of paste backfill materials. The present invention uses DIC to observe the fracture evolution and deformation response of CPB specimens during the uniaxial compression test, and proposes a computer vision recognition of HSV color patterns (HSV-CVR) model based on the strain field data obtained by DIC to analyze the area share and distribution of strain areas of different degrees; then, by analyzing the changes in the strain field, the early warning signal for crack prediction is determined. Summary of the Invention

[0005] The purpose of the present invention is to provide a strain field evolution and crack prediction method for a fully tailings-cemented backfill. Based on a computer vision recognition model of the HSV color pattern, by dividing the principal strain field into eight strain regions, a quantitative analysis of the strain evolution of the backfill is achieved, providing high, medium, and low levels of early warning for the damage risk of the backfill. This not only enhances the understanding of the mechanical behavior of mine backfill materials, but also provides a scientific basis for the design, construction monitoring, and risk management of mine backfill materials, which is of great significance for ensuring the safety and efficiency of mine operations.

[0006] To achieve the above object, the present invention provides a method for strain field evolution and crack prediction of a fully tailings cemented backfill, comprising the following steps:

[0007] Step S1: preparing a full tailings cemented fill CPB sample and performing an expansion test during the CPB sample preparation process;

[0008] Step S2: using a fully automatic flexural and compressive testing machine to evaluate the uniaxial compressive strength of the CPB specimens, while simultaneously monitoring surface changes of the specimens during the test using digital image correlation (DIC) image acquisition technology;

[0009] Step S3: Based on the strain field data obtained by DIC, a computer vision recognition (HSV-CVR) model based on the HSV color model is established to analyze the area proportion and distribution of regions with different degrees of strain, thereby achieving a comprehensive analysis of the image;

[0010] Step S4: determining an early warning signal for crack prediction based on the analysis results of the strain field change.

[0011] Preferably, the whole tailings cemented filling body sample is prepared by mixing and pouring cement, whole tailings and tap water, and the preparation process is as follows:

[0012] First, use a cement slurry mixer to prepare the slurry. Pour mixing water into the mixing bucket, then add the evenly mixed solid materials, full tailings and cement, and stir at low speed for 120 seconds, then at high speed for 120 seconds.

[0013] Then, the slurry is injected into a mortar expansion tester to conduct expansion test;

[0014] Finally, the slurry was poured into a standard triple test mold, wrapped with plastic wrap, and placed in a constant temperature and humidity curing box for 28 days to prepare the CPB specimen.

[0015] Preferably, the specifications of the mortar expansion tester are 50mm*100mm*150mm; the specifications of the standard triple test mold are 70.7mm*70.7mm*70.7mm; the temperature of the constant temperature and humidity curing box is controlled at 20±2°C, and the humidity is greater than 90%.

[0016] Preferably, in step S2, the application of DIC technology involves a digital image acquisition system and digital image processing software, including a loading device, a light source, a high-resolution digital camera, and a computer;

[0017] The DIC image acquisition system consists of a CCD camera, a ring light source, and MV Viewer software. Before loading, the DIC system is calibrated to adjust the camera's angle and focus. The camera has a resolution of 4000 × 3000 pixels and a frame rate of 9 frames per second.

[0018] Preferably, in step S2, a uniaxial compression test is performed on the CPB sample, and the specific process is as follows:

[0019] Step S21: After the CPB sample reaches a specified age, a layer of white primer is first evenly applied on its surface, and then black spots are randomly sprinkled on it to form a spot pattern with high contrast;

[0020] Step S22: adjusting the light source of the DIC digital image acquisition system to obtain grayscale data during the loading process;

[0021] Step S23, setting the acquisition frequency of the camera to capture the sample image coated with artificial spots;

[0022] Step S24: The collected image data is stored in a computer, and the displacement change of the surface is obtained by analysis using VIC-2D software to obtain strain field data.

[0023] Preferably, in step S3, the strain field image obtained based on the DIC technology is segmented, and the specific process is as follows:

[0024] Step S31: First, use a camera to capture a high-definition speckle image of the sample during the uniaxial compression test;

[0025] Step S32: Subsequently, the strain field data of the study area is obtained using VIC-2D software;

[0026] Step S33: Then, the HSV-CVR program is used to process the strain field image data of the region of interest of the CPB sample;

[0027] Step S34: Finally, comprehensive analysis of the image is achieved through segmentation of the strain field and quantitative processing of visual recognition data.

[0028] Preferably, the strain field image obtained by VIC-2D software processing is a raster image; the image is composed of different pixels, each pixel containing color information; the color information is recorded using the RGB color mode, that is, the information of a pixel is superimposed with the red R, green G, and blue B color information;

[0029] The HSV-CVR program is used to process the strain field image data of the CPB specimen region of interest. The specific process is as follows:

[0030] Step S331: traverse all image files in the specified directory and use the OpenCV library to read the image files;

[0031] Step S332: Convert the color space of the image from RGB to HSV to process the color information, where H is hue, S is saturation, and V is brightness;

[0032] Step S333: define the number of tone groups and their corresponding ranges, and classify pixels of different tones;

[0033] Step S334: traverse each tone group, create a mask for each group, and determine the area ratio of the tone group in the entire image by counting the number of non-zero pixels in the mask. Pixels belonging to the interval are retained, and pixels not belonging to the interval are replaced with white pixels, thereby achieving image segmentation;

[0034] Step S335 : Integrate the above area percentage data into a table, calculate the ratio of the interval area color block in the frame image corresponding to different time to the entire region of interest, and obtain the curve of the change of strain percentage at the interval scale over time.

[0035] Preferably, in step S4, based on the analysis results of the strain field changes, three levels of early warning are provided for the damage risk of the filling body, namely high, medium and low, to determine an early warning signal for crack prediction.

[0036] Therefore, the present invention adopts the above-mentioned strain field evolution and crack prediction method of a fully tailings cemented filling body, and the beneficial effects are as follows:

[0037] (1) In this invention, the DIC and HSV-CVR models effectively reveal the deformation field and crack propagation law of CPB specimens under uniaxial compressive load;

[0038] (2) The HSV-CVR model proposed in this paper intuitively identifies and quantifies the area proportions of different strain intervals, thereby more accurately understanding the strain evolution characteristics of the filling body during loading, especially the strain concentration phenomenon during crack formation and propagation stages;

[0039] (3) The method proposed in the present invention can quantitatively identify the precursory characteristics of crack extension and penetration. Crack prediction is performed based on the values ​​and percentage changes of different strain intervals in the sample strain field. Three different levels of early warning signals, namely high strain, medium strain and low strain, are set. By analyzing the changes in the strain interval, the damage risk of the filling body is evaluated, providing a scientific basis for mine filling projects and helping engineers make more reasonable design decisions before construction.

[0040] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flow chart of the strain field evolution and crack prediction method of a fully tailings cemented backfill;

[0042] Figure 2 is a tailings particle size distribution diagram of the present invention;

[0043] Figure 3 This is the preparation process of the CPB sample of the present invention;

[0044] Figure 4 It is a DIC system device of the present invention;

[0045] Figure 5 This is a flow chart of the principal strain deformation field region segmentation of the present invention;

[0046] Figure 6 This is a flow chart of the present invention using the HSV-CVR program to process CPB specimen strain field image data;

[0047] Figure 7 It is the stress-time curve and deformation field distribution image during the loading process of the present invention; wherein A, B, C, and D are the marking points on the stress-time curve;

[0048] Figure 8 Figure 1 shows the strain ratio of a CPB specimen over time and the segmentation of the deformation field. (a) shows the variation of the ratio of eight strain intervals of the CPB specimen over time. (b) shows the eight regions of the strain field image at four observation points A, B, C, and D segmented based on the HSV-CVR model.

[0049] Figure 9 This is the first-order derivative distribution diagram of the principal strain field of the CPB specimen of the present invention in eight partitions;

[0050] Figure 10 It is the first-order derivative distribution diagram and early warning signal of the principal strain interval percentage curve of the present invention. DETAILED DESCRIPTION

[0051] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0052] like Figure 1 As shown, a strain field evolution and crack prediction method for a fully tailings cemented backfill body includes the following steps:

[0053] Step S1: preparing a full tailings cemented backfill (CPB) sample and performing an expansion test during the CPB sample preparation process;

[0054] Step S2: using a fully automatic flexural and compressive testing machine to evaluate the uniaxial compressive strength of the CPB specimens, while monitoring surface changes of the specimens during the test using digital image correlation (DIC) image acquisition technology;

[0055] Step S3: Based on the strain field data obtained by DIC, a computer vision recognition (HSV-CVR) model based on the HSV color model is established to analyze the area proportion and distribution of regions with different degrees of strain, thereby achieving a comprehensive analysis of the image;

[0056] Step S4: determining an early warning signal for crack prediction based on the analysis results of the strain field change.

[0057] Example 1

[0058] Step S1: Prepare a full tailings cemented backfill (CPB) sample and perform an expansion test during the CPB sample preparation process.

[0059] The whole tailings cemented filling body sample is made by mixing cement, whole tailings from a tungsten mine in Jiangxi Province and tap water. Among them, the particle size curve of the whole tailings is as follows Figure 2 As shown, the cement model is 42.5OPC.

[0060] The slurry was prepared using a cement slurry mixer. First, pour mixing water into the mixing bucket, then add the thoroughly mixed solid materials (whole tailings and cement). To ensure the slurry's uniformity, stir at a low speed for 120 seconds, then at a high speed for 120 seconds. The slurry was then injected into a mortar expansion tester (50mm*100mm*150mm) for expansion testing.

[0061] Then, the slurry was poured into a standard triple test mold (70.7 mm * 70.7 mm * 70.7 mm) prepared in advance, wrapped with plastic wrap, and placed in a constant temperature and humidity curing box (temperature controlled at (20 ± 2) ° C, humidity controlled at more than 90%) for curing for 28 days to prepare CPB specimens. Figure 3 shown.

[0062] Step S2: Use a fully automatic flexural and compressive testing machine to evaluate the uniaxial compressive strength of the CPB specimen. Simultaneously, digital image correlation (DIC) image acquisition technology is used to monitor the surface changes of the CPB specimen during the test to obtain strain field data.

[0063] The present invention uses a DYE-300-10S fully automatic flexural and compressive testing machine to evaluate the uniaxial compressive strength of CPB concrete specimens, and simultaneously monitors the surface changes of the specimens during the test using DIC image acquisition technology. The application of DIC technology involves a digital image acquisition system and digital image processing software, including a loading device, a light source, a high-resolution digital camera, and a computer. Figure 4 shown.

[0064] The DIC image acquisition system consists of a CCD camera, a ring light source, and MV Viewer software. Before loading, the DIC system is calibrated, and the camera's angle and focus are adjusted to ensure a clear and comprehensive capture of the entire specimen surface. The camera has a resolution of 4000 × 3000 pixels and a frame rate of 9 frames per second. Finally, the continuous sequence of spot images captured during the test is analyzed using VIC-2D software to determine the evolution of the strain field on the specimen surface.

[0065] The specific process of uniaxial compression test on CPB specimens is as follows:

[0066] Step S21: After the CPB sample reaches a specified age, a layer of white primer is first evenly applied on its surface, and then black spots are randomly sprinkled on it to form a spot pattern with high contrast.

[0067] Step S22: Adjust the light source of the DIC digital image acquisition system to ensure that the brightness is appropriate and stable, so as to obtain reliable grayscale data during the loading process.

[0068] Step S23: setting the acquisition frequency of the camera to capture the image of the sample coated with artificial spots.

[0069] Step S24: The collected image data is stored in a computer, and VIC-2D software is used to analyze the surface displacement change and obtain strain field data.

[0070] Step S3: Based on the strain field data obtained by DIC, a computer vision recognition (HSV-CVR) model based on the HSV color model is established to analyze the area proportion and distribution of regions with different degrees of strain, thereby achieving a comprehensive analysis of the image.

[0071] The basic principle of DIC is to compare two digital images recorded before and after deformation, and calculate the grayscale of the speckle image before and after deformation on the specimen surface to obtain an optical measurement method for parameters such as displacement and deformation of the research object. In order to simplify the visual recognition and image segmentation process of physical images, DIC technology is used to obtain strain field images. The strain field segmentation process is as follows: Figure 5 shown.

[0072] Step S31: First, use a camera to capture a high-definition speckle image of the sample during the uniaxial compression test.

[0073] Step S32: Subsequently, the maximum principal strain field data of the study area is obtained using VIC-2D software.

[0074] Here, the principal strain field ε1 data is exported in the form of high-definition video, and frame extraction is performed to obtain a high-definition image of the region of interest (ROI).

[0075] Step S33: Then, the HSV-CVR program is used to process the strain field image data of the CPB sample ROI.

[0076] The strain field images generated by VIC-2D software are raster images. These images are composed of distinct pixels, each of which contains color information. This color information is recorded using the RGB color model, which overlays the information for each pixel with the red (R), green (G), and blue (B) color information.

[0077] like Figure 6 As shown in Figure 2, the HSV-CVR program is used to process the strain field image data of the CPB specimen ROI. The specific process is as follows:

[0078] Step S331: traverse all image files in the specified directory and use the OpenCV library to read the image files;

[0079] Step S332: Convert the color space of the image from RGB to HSV (hue, saturation, brightness) to more accurately process the color information.

[0080] Step S333: defining the number of tone groups and their corresponding ranges, so as to facilitate classification of pixels of different tones;

[0081] Step S334: traverse each tone group, create a mask for each group, and determine the area ratio of the tone group in the entire image by counting the number of non-zero pixels in the mask. Pixels belonging to the interval are retained, and pixels that do not belong to the interval are replaced with white pixels, thereby achieving image segmentation.

[0082] Step S335: Integrate the above area ratio data into a table, calculate the ratio of the interval area color block in the frame image corresponding to different time to the entire ROI, and obtain the curve of the strain percentage change at the interval scale over time, laying the foundation for subsequent data analysis and visualization.

[0083] Step S34: Finally, comprehensive analysis of the image is achieved through segmentation of the strain field and quantitative processing of visual recognition data.

[0084] Here we mainly analyze the data of the maximum principal strain ε1, and set the strain scale range to 0 to 0.03. By uniform division, the strain scale is divided into 8 intervals to analyze the strain evolution pattern in more detail. These intervals are:

[0085] S1(26.25~30.00×10 ^-3 ); S2(22.50~26.25×10 ^-3 ); S3(18.75~22.50×10 ^-3 ); S4(15.00~18.75×10 ^-3 ); S5(11.25~15.00×10 ^-3 ); S6(7.50~11.25×10 ^-3 ); S7(3.75~7.50×10 ^-3 ); S8(0.00~3.75×10 ^-3 ).

[0086] This process not only improves the efficiency of color analysis, but also provides strong data support for a deeper understanding of the characteristics of color distribution in images.

[0087] Step S4: determining an early warning signal for crack prediction based on the analysis results of the strain field change.

[0088] Example 2

[0089] In this example, uniaxial compression and expansion tests were conducted on the full tailings cemented filling slurry, using slurry filling concentration and lime-sand ratio as the primary variables. The filling sample had a slurry mass concentration of 76%, a lime-sand ratio of 1:8, a 28-day compressive strength of 2.4 MPa, and an expansion of 300 mm.

[0090] The image captured at the beginning of the uniaxial compression test is used as the reference image, and the other images captured during the experiment are deformation images. Figure 7As shown, four points (AD) were marked on the stress-time curve, and the corresponding strain fields were determined. Based on the strength values ​​of the CPB specimens after 28 days of curing, point A, where the strength was 0.5 MPa, was selected as the initial loading stage for comparative analysis. The starting point of the persistent S7 deformation region (low strain region) was designated as point B (Low-Range Strain Point, LRSP); the starting point of the persistent S1 deformation region (high strain region) was designated as point C (High-Range Strain Point, HRSP); and the peak stress point was designated as point D. It should be noted that points B and C are rough positioning analysis points. Because the starting point of the S4 mid-range strain region (mid-range strain point, MRSP) cannot be accurately distinguished by the naked eye, no observation points were set for the mid-strain region. The starting time of the specific deformation value region will be quantified using the HSV-CVR algorithm and subsequent first-order derivatives.

[0091] Next, the strain evolution characteristics of the CPB specimens at different loading stages were quantitatively analyzed based on the HSV-CVR model. By monitoring the percentage of the maximum principal strain ε1 level (S1 to S8) over time, the strain evolution characteristics of the specimens throughout the experiment were revealed.

[0092] like Figure 8 Figure 2 shows the strain evolution of a control specimen with a fill concentration of 76% and a lime-sand ratio of 1:8. As stress increases, the percentage of the minimum strain interval S8 decreases, while the percentages of other horizontal strain intervals increase. These changes are not linear but rather fluctuate, reflecting the complex deformation behavior of the specimen during loading. For example, some regions of the specimen may return to their initial state after experiencing elastic strain, while other regions may undergo plastic strain. The percentages of the low-strain intervals S7 and S8 remain consistently high throughout the test, indicating that these intervals are relatively active throughout the test. Over time, the percentage of the higher-strain intervals increases, especially in the later stages of the test, when there is a sudden increase in the proportion of high-strain intervals due to irreversible plastic strain.

[0093] In particular, the curve for the S7 strain interval exhibits a distinct evolutionary pattern throughout the test: a rapid increase from a relatively stable state, followed by a slight slowdown, and finally a rapid increase again near crack coalescence. This pattern indicates that compressible deformation regions within the specimen rapidly respond to the stress increase in the initial stage. These regions then further increase in deformation and transition to other strain intervals. Finally, during the crack coalescence phase, strain values ​​rapidly expand, resulting in a sudden increase in the percentage of the S7 strain interval. For the CPB specimen, during the initial loading phase (point A), strain is primarily concentrated in the lowest strain interval, S8, accounting for as much as 99.95%, indicating that the specimen undergoes almost no significant deformation. As loading progresses, strain begins to shift toward higher intervals. By point B, the percentage of the S7 interval increases to 3.97%, while the S8 interval decreases to 96.03%. At point C, the strain distribution becomes more dispersed, with strain beginning to appear in intervals S1 to S6. The percentage of the S7 interval increases significantly to 14.23%, while the S8 interval decreases to 75.86%. Near the peak stress (point D), the strain distribution is more extensive, and the proportion of intervals S1 to S6 increases. The proportion of interval S7 reaches 30.26%, and the proportion of interval S8 decreases to 29.26%.

[0094] like Figure 9 As shown in Figure 1, the rate of change of the percentage curve (S1-S8) of the ε1 strain interval for the specimen can be used to predict crack development. Analysis of the ε1 strain interval reveals that with increasing stress, strain concentration shifts from the low-strain interval to the high-strain interval. In particular, the significant increase in the S7 interval indicates the impending onset of higher strains, and even the possibility of macroscopic coalescence cracks. The increase in the percentage of the S7 strain interval indicates that an increasing number of strain intervals below the S7 level are moving toward the S7 level, further confirming that rapid growth in the low-strain interval is a key characteristic of crack coalescence. For major mine filling projects with stringent crack control requirements, monitoring the changes in the percentage growth rate of the low-strain interval is crucial. Focusing on the S7 low-strain region is of great value in predicting crack coalescence in actual mine filling projects. With increasing stress, strain concentration shifts from the low-strain interval to the high-strain interval. Within the S7-S1 range, crack coalescence is accompanied by a rapid increase in strain.

[0095] In particular, the significant increase in the proportion of the S1 strain interval at point C indicates that the specimen has the most pronounced strain concentration in these intervals, which may be a direct result of crack merging and expansion. It can be determined that when the strain of S1 increases rapidly, it is caused by crack merging. Because the S4 strain interval can reflect the behavior of CPB specimens at intermediate strain levels, selecting the S4 strain interval as a key monitoring point can provide important information for early identification and prevention of cracks. In addition, significant changes in the S4 interval may indicate the intermediate stage of crack expansion, which is of great significance for evaluating material stability and taking timely engineering measures.

[0096] like Figure 10 As shown in the figure, by comprehensively monitoring the rate of change of the strain level curves in the S1, S4 and S7 intervals, the growth points with larger first-order derivative values ​​are found as key monitoring points. Here, the first-order derivative value of S1 is 0.01 as the HRSP, the first-order derivative value of S4 is 0.03 as the MRSP, and the first-order derivative value of S7 is 0.2 as the LRSP. This selection is based on a detailed analysis of the strain evolution data of the CPB specimens during the loading process to more accurately predict and evaluate the damage risk of the filling body.

[0097] Below the LRSP, the first-order derivative curve of the S7 strain interval percentage exhibits fluctuations, likely reflecting the inhomogeneity of the material's internal stress distribution during the initial loading phase or minor adjustments during the experiment. These fluctuations are insufficient to represent the material's stable strain response. However, above the LRSP, the strain in the S7 interval begins to increase significantly and continuously, marking the onset of strain concentration. Furthermore, the strain increase after the LRSP is more pronounced and sustained, indicating that the material begins to experience more significant strain concentration. The significance of this change makes the LRSP a suitable reference point for evaluating material strain evolution and crack development, similar to the selection of the MRSP and HRSP.

[0098] Since the HRSP indicates that the filling structure is about to reach its maximum bearing capacity, the value of the first-order derivative in the S7 region is set to be small. The MRSP indicates that cracks have begun to appear or cracks have propagated inside the filling body. The LRSP indicates that the filling body is under external pressure and the strain concentration area has begun to increase continuously.

[0099] In practical applications, these mutation points can serve as early warning signals, helping engineers and researchers predict material damage or failure, and can also be further adjusted based on specific mine filling requirements. Selecting the HRSP, MRSP, and LRSP key points helps avoid experimental noise interference and focus on analyzing the engineering-relevant strain growth phases, providing key indicators for crack identification and prediction in practical engineering applications.

[0100] The proposed method provides new insights into crack prediction, helping to improve project stability and safety. When data from key areas of the filling structure exceed the MRSP, remedial measures are implemented to prevent potential damage. When data exceed the HRSP, macrocracks in the mine filling must be promptly addressed to ensure mine safety and stability. By monitoring the derivative values ​​within these strain intervals, the risk of filling failure can be more accurately predicted and assessed, providing valuable insights for mine filling projects.

[0101] Therefore, the present invention adopts the above-mentioned strain field evolution and crack prediction method of a fully tailings cemented backfill body, and based on the computer vision recognition model of the HSV color mode, realizes the quantitative analysis of the strain evolution of the backfill body by dividing the principal strain field into 8 strain regions, and provides high, medium and low levels of early warning for the damage risk of the backfill body; it not only enhances the understanding of the mechanical behavior of mine backfill materials, but also provides a scientific basis for the design, construction monitoring and risk management of mine backfill materials, which is of great significance for ensuring the safety and efficiency of mine operations.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A strain field evolution and crack prediction method for a fully tailings cemented backfill, characterized in that: The following steps are involved: Step S1: preparing a full tailings cemented fill CPB sample and performing an expansion test during the CPB sample preparation process; Step S2: using a fully automatic flexural and compressive testing machine to evaluate the uniaxial compressive strength of the CPB specimens, while simultaneously monitoring surface changes of the specimens during the test using digital image correlation (DIC) image acquisition technology; Among them, the uniaxial compression test was carried out on the CPB sample, and the specific process is as follows: Step S21: After the CPB sample reaches a specified age, a layer of white primer is first evenly applied on its surface, and then black spots are randomly sprinkled on it to form a spot pattern with high contrast; Step S22: adjusting the light source of the DIC digital image acquisition system to obtain grayscale data during the loading process; Step S23, setting the acquisition frequency of the camera to capture the sample image coated with artificial spots; Step S24: storing the collected image data in a computer, and analyzing the surface displacement changes using VIC-2D software to obtain strain field data; Step S3: Based on the strain field data obtained by DIC, a computer vision recognition (HSV-CVR) model based on the HSV color model is established to analyze the area proportion and distribution of regions with different degrees of strain, thereby achieving a comprehensive analysis of the image; Among them, based on the strain field image obtained by DIC technology, the strain field is segmented. The specific process is as follows: Step S31: First, use a camera to capture a high-definition speckle image of the sample during the uniaxial compression test; Step S32: Subsequently, the strain field data of the study area is obtained using VIC-2D software; Step S33: Then, the HSV-CVR program is used to process the strain field image data of the region of interest of the CPB sample; The strain field image processed by the VIC-2D software is a raster image. The image is composed of different pixels, each of which contains color information. This color information is recorded using the RGB color mode, which superimposes the information of a pixel with the red (R), green (G), and blue (B) color information. The HSV-CVR program is used to process the strain field image data of the CPB specimen region of interest. The specific process is as follows: Step S331: traverse all image files in the specified directory and use the OpenCV library to read the image files; Step S332: Convert the color space of the image from RGB to HSV to process the color information; where H is hue, S is saturation, and V is brightness; Step S333: define the number of tone groups and their corresponding ranges, and classify pixels of different tones; Step S334: traverse each tone group, create a mask for each group, and determine the area ratio of the tone group in the entire image by counting the number of non-zero pixels in the mask. Pixels belonging to the interval are retained, and pixels not belonging to the interval are replaced with white pixels, thereby achieving image segmentation; Step S335: Integrate the above area ratio data into a table, calculate the ratio of the interval area color block in the frame image corresponding to different time to the entire region of interest, and obtain a curve of the change of the strain percentage of the interval area color block over time; Step S34: Finally, a comprehensive analysis of the image is achieved through the segmentation of the strain field and the quantitative processing of the visual recognition data; Step S4: determining an early warning signal for crack prediction based on the analysis results of the strain field change.

2. The method for strain field evolution and crack prediction of a fully tailings cemented filling body according to claim 1, characterized in that: The full tailings cemented filling sample is made by mixing and pouring cement, full tailings and tap water. The preparation process is as follows: First, use a cement slurry mixer to prepare the slurry. Pour mixing water into the mixing bucket, then add the evenly mixed solid materials, full tailings and cement, and stir at low speed for 120 seconds, then at high speed for 120 seconds. Then, the slurry is injected into a mortar expansion tester to conduct expansion test; Finally, the slurry was poured into a standard triple test mold, wrapped with plastic wrap, and placed in a constant temperature and humidity curing box for 28 days to prepare the CPB specimen.

3. The method for strain field evolution and crack prediction of a fully tailings cemented backfill according to claim 2, characterized in that: The specifications of the mortar expansion meter are 50mm*100mm*150mm; the specifications of the standard triple test mold are 70.7mm*70.7mm*70.7mm; the temperature of the constant temperature and humidity curing box is controlled at 20±2℃, and the humidity is greater than 90%.

4. The method for strain field evolution and crack prediction of a fully tailings cemented backfill according to claim 1, characterized in that: In step S2, the application of DIC technology involves a digital image acquisition system and digital image processing software, including a loading device, a light source, a high-resolution digital camera, and a computer; The DIC image acquisition system consists of a CCD camera, a ring light source, and MV Viewer software. Before loading, the DIC system is calibrated to adjust the camera's angle and focus. The camera has a resolution of 4000 × 3000 pixels and a frame rate of 9 frames per second.

5. The method for strain field evolution and crack prediction of a fully tailings cemented backfill according to claim 1, characterized in that: In step S4, based on the analysis results of the strain field changes, three levels of early warning are provided for the damage risk of the filling body, namely high, medium and low, to determine the early warning signal for crack prediction.

Images