Tunnel excavation face blasting effect evaluation method and system based on digital image technology

Abstract

The present application mainly relates to the technical field of tunnel engineering, in order to solve the problem of poor blasting effect caused by the lag of tunnel excavation face blasting effect evaluation and the single evaluation index, the present application provides a tunnel excavation face blasting effect evaluation method and system based on digital image technology, the core of which is: through establishing a rock mass identification model to identify the lithology distribution result of rock mass image, the lithology distribution and lithology area proportion of the excavation face are obtained, a blastability index analysis model is established based on the lithology distribution, lithology area proportion and flatness of rock mass, the influence of artificial experience on rock mass judgment is eliminated, the blastability characteristics of rock mass are quantified, data support is provided for the accurate matching of blasting parameters and blastability characteristics of rock mass in advance, and the blasting parameters are adjusted based on the blastability index combined with the flatness of rock mass, overbreak surface and underbreak surface, breaking through the limitation of single evaluation dimension of traditional method, and comprehensively improving the blasting effect.

Description

TECHNICAL FIELD

[0001] The present application mainly relates to the technical field of tunnel engineering, in particular to a tunnel excavation face blasting effect evaluation method and system based on digital image technology. BACKGROUND

[0002] In the field of tunnel excavation blasting effect evaluation, the existing technology has many limitations, which is difficult to meet the efficient and accurate construction requirements. First of all, the traditional manual measurement method not only needs to interrupt the construction process, but also takes more than 30 minutes to evaluate a single cycle, which seriously restricts the rhythm of rapid tunnel construction, and the measurement results are easily affected by human operation errors, and the reliability is difficult to guarantee. Secondly, although high-precision devices such as laser scanning can obtain three-dimensional data, the cost of purchasing the device is high, and the data post-processing process is complex, and the analysis time of a single group of data is long, which cannot provide real-time feedback for the next round of blasting parameter adjustment, especially not suitable for small and medium-sized tunnels or frequent monitoring scenes.

[0003] Furthermore, the judgment of rock mass blastability has long relied on the experience accumulation of engineers, and the evaluation of key indicators such as rock hardness and fracture development degree is highly subjective, which easily leads to mismatch between blasting parameters and actual rock properties, and further causes serious overbreak, poor excavation face flatness and other problems. At the same time, the evaluation dimension of the existing technology is relatively single, and is mostly focused on basic indicators such as overbreak, and lacks systematic analysis of the deep relationship between the correlation of excavation face flatness and support cost, the adaptability of rock mass distribution and blasting energy, etc., which is difficult to form a comprehensive blasting effect evaluation system.

[0004] In addition, the maintenance and labor cost of the traditional method is high, and the annual maintenance cost of the laser scanning device and the annual labor cost of the manual measurement team significantly increase the engineering cost in the long run. These defects together lead to parameter adjustment lag, serious resource waste in tunnel blasting construction, and restrict the improvement of construction efficiency and safety. SUMMARY

[0005] The technical problem to be solved by the present application is to provide a tunnel excavation face blasting effect evaluation method and system based on digital image technology, which aims to solve the problems of lagging blasting effect evaluation of the tunnel excavation face and poor blasting effect caused by single evaluation index.

[0006] The technical solution adopted by the present application to solve the above technical problems is:

[0007] On the one hand, the present application provides a tunnel excavation face blasting effect evaluation method based on digital image technology, which comprises:

[0008] S1: collecting the image of the rock mass after blasting, and pre-processing the image;

[0009] S2: Establish a rock mass identification model, and identify the pre-processed image based on the rock mass identification model to obtain the lithology distribution result of the excavation surface, including the lithology category and the area ratio of each lithology in the excavation surface;

[0010] S3: Establish a blastability index calculation model based on the flatness and the lithology distribution result;

[0011] S4: Calculate the blastability index of the target rock mass based on the blastability index calculation model;

[0012] S5: Adjust the target rock mass blasting parameters in combination with the blastability index of the target rock mass and the overbreak, underbreak and flatness.

[0013] Further, the rock mass image of the excavation surface after blasting collected in S1 comprises:

[0014] Image collection preparation: perform marker layout and image collection equipment installation;

[0015] Image collection: collect the excavation rock mass image after blasting within a set time according to a set image collection frequency.

[0016] Further, the pre-processing of the image in S1 comprises:

[0017] Image denoising: remove high-frequency noise in the image using a Gaussian filter method;

[0018] Image registration: take the reference marker as a feature point, calculate the conversion matrix M of the design coordinate system and the image coordinate system, and obtain the registered image:

[0019] Contrast enhancement: use adaptive histogram equalization to improve the lithology texture clarity in the image.

[0020] Further, S2 comprises:

[0021] Establish a rock mass identification model based on a U-Net network, use a cross-entropy function as a loss function, take the pre-processed excavation surface rock mass image as an input feature, and take the lithology category as a real label Train the U-Net network, and compare the predicted lithology category of the U-Net network with the real label during the training process to optimize the loss function; the loss function is specifically: ;

[0022] The area ratio calculation method of each lithology in the excavation surface is: , wherein represents the lithology category, , represents the area ratio of the th lithology in the excavation surface, is an indicator function, denotes the lithology class at the coordinate , satisfies , takes 1, otherwise takes 0, is the total number of pixel points in the excavation face image.

[0023] Further, the S3 said explosibility index calculation model is , wherein is the explosibility index, is the weight coefficient of the flatness , is the area ratio of different lithology in the excavation face, is the weight coefficient of the corresponding lithology.

[0024] Further, the value of the weight coefficient obtained by identifying the historical blasting data based on the neural network.

[0025] Further, the S4 said combining the explosibility index of the target rock mass and the overbreak volume, the underbreak volume and the flatness to adjust the rock mass blasting parameters comprises:

[0026] The peripheral hole charge quantity is corrected by the overbreak or underbreak volume;

[0027] The blast hole density is optimized according to the flatness;

[0028] The explosibility index is graded, and the slotting hole charge quantity, hole spacing and initiation sequence are adjusted according to the explosibility index grade.

[0029] Further, the overbreak volume and the underbreak volume calculation comprises:

[0030] The actual excavation contour line is obtained from the enhanced image based on the edge detection algorithm, The overbreak volume and the underbreak volume are calculated based on the actual excavation contour line and the design excavation contour line ; , denotes the pixel point inside the actual contour line but not inside the design contour line , denotes the pixel point inside the design contour line but not inside the actual contour line .

[0031] Further, the flatness calculation comprises: selecting the excavation face reference line , calculating the actual coordinate point in the excavation face rock mass image Distance to the baseline , the maximum distance The ratio of the average distance As flatness.

[0032] In another aspect, the present application also provides a tunnel excavation surface blasting effect evaluation system based on digital image technology, the system comprises:

[0033] An image acquisition module is configured to acquire images of the excavation rock surface in a set time according to a set image acquisition frequency.

[0034] An image preprocessing module is configured to preprocess the rock mass images acquired by the image acquisition module.

[0035] A rock mass identification module is configured to obtain the lithology distribution results of the preprocessed rock mass images, including the lithology categories and the area proportions of different lithologies in the excavation surface.

[0036] An explodability index calculation module is configured to calculate the rock mass explodability index according to the lithology distribution results.

[0037] A blasting parameter guidance module is configured to output the target rock mass blasting parameters according to the explodability index and the overbreak, underbreak and flatness.

[0038] The present application has the following advantages: the rock mass identification model is established to identify the lithology distribution results of the rock mass images, the lithology distribution and the lithology area proportion of the excavation surface are obtained, the explodability index analysis model is established based on the lithology distribution, the lithology area proportion and the flatness of the rock mass, the influence of artificial experience on the lithology judgment is eliminated, the rock mass explodability characteristics are quantified, the data support is provided for the accurate matching of the blasting parameters and the rock mass explodability characteristics in advance, and the blasting parameters are adjusted based on the explodability index, the rock mass flatness, the overbreak surface and the underbreak surface, the limitations of the single evaluation dimension of the traditional method are broken through, and the blasting effect is comprehensively improved. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 The present application provides a tunnel excavation surface blasting effect evaluation method based on digital image technology;

[0040] Figure 2 The overbreak surface and the underbreak surface are shown in the figure;

[0041] Figure 3 The flatness index is shown in the figure;

[0042] Figure 4 The optimization flowchart of the blasting parameters according to the explodability index is shown in the figure. DETAILED DESCRIPTION

[0043] The core of the tunnel excavation face blasting effect evaluation method based on digital image technology solving the above technical problems lies in: establishing a rock body identification model, identifying the rock body image of the excavation face after blasting based on the rock body identification model, obtaining the lithology distribution result of the excavation face, including the lithology category and the area ratio of each lithology in the excavation face; establishing a blastability index calculation model according to the lithology distribution result and the flatness of the excavation face, quantifying the rock body blastability characteristics of the rock body in data form, and combining the blastability index of the target rock body with the overbreak, underbreak and flatness to adjust the blasting parameters of the target rock body, thereby providing data support for the accurate matching of the blasting parameters and the lithology of the excavation face.

[0044] As shown in Figure 1 The tunnel excavation face blasting effect evaluation method based on digital image technology provided by the application comprises the following steps:

[0045] S1: Collecting the rock body image of the excavation face after blasting, and pre-processing the image.

[0046] Before collecting the rock body image of the excavation face, the markers are laid out and the image collection equipment is installed. The marker laying: fixing the reference markers, such as circular targets with a diameter of 5-10 cm, around the tunnel excavation face, and pre-measuring the center coordinates of the markers by a total station instrument as the registration reference of the image and the tunnel design coordinate system. The image collection equipment installation: fixing an industrial camera (resolution ≥ 12 million pixels) at a distance of 5-10 m behind the tunnel face, or using a drone with a high-definition camera to hover and shoot the rock body image within a safe distance, and setting the camera parameters as follows: focal length 50-100 mm, aperture f / 8-f / 16, exposure time 1 / 500-1 / 1000 s, to ensure the clarity of the lithological layering and blasting traces on the excavation face.

[0047] After the marker laying and image collection equipment installation are completed, the excavation rock body image is collected according to the set image collection frequency. In order to avoid the influence of weathering or secondary disturbance of the surrounding rock on the data accuracy, the rock body image is collected within 10 minutes after the blasting is completed and the smoke is discharged.

[0048] During the rock body image collection, if the surface texture of the excavation face rock body is blurred, the rock body is homogeneous, or the rock surface is smooth after blasting without obvious joints, or the natural features such as cracks and lithological boundaries are difficult to identify due to dust and light interference, the speckle spraying is needed to supplement the texture. If the natural texture features of the excavation face rock body, such as the layering of shale, the particle distribution of sandstone and the developed cracks, are clear, and these features can be stably identified in the image without being seriously disturbed by light and dust, the speckle spraying is not needed.

[0049] The collection frequency of the rock body image is set to 1 time per blasting cycle. If a lithology mutation occurs, such as from sandstone to granite, the collection frequency is increased to 2-3 times, and the rock body images are shot in different areas of the tunnel face.

[0050] After the image acquisition is completed, the acquired image and the excavation design contour line of the tunnel are imported into a backend server, and subsequent data processing is performed in the backend server.

[0051] The pre-processing of the acquired rock mass image includes image denoising, image registration, and contrast enhancement.

[0052] Image denoising: In this embodiment, a Gaussian filtering method is used to remove high-frequency noise in the image, and the formula is wherein is the original image, is a Gaussian function, is the image after noise removal.

[0053] Image registration: Taking the reference marker as the feature point, the conversion matrix of the design coordinate system and the image coordinate system is calculated .

[0054] , ;

[0055] wherein is the image pixel coordinate, is the design coordinate system coordinate after conversion, and the conversion matrix is solved by the least square method to minimize the marker coordinate error, wherein the elements of the conversion matrix are used to describe the coordinate conversion relationship between the image coordinate system and the tunnel design coordinate system, and the optimal matrix elements are determined by minimizing the sum of squares of the error between the converted coordinates and the actual design coordinates;

[0056] The acquisition of the elements of the conversion matrix : Define the error function , and calculate the partial derivatives of , respectively, and set the partial derivatives to 0 to obtain a linear equation group, and the element values of the conversion matrix are obtained through the linear equation group.

[0057] Contrast enhancement: adaptive histogram equalization (CLAHE) is used to improve the lithology texture clarity in the image, ;

[0058] The principle is to solve the problem of lithology texture blurring caused by uneven lighting and dust covering on the tunnel excavation surface, and to achieve accurate enhancement through three-step core processing:

[0059] Block-based adaptive processing: The image is divided into multiple local sub-blocks (e.g., 16×16 pixels), and the gray-level distribution of each sub-block is calculated independently, so that the lithological textures (e.g., rock layer joints, grain boundaries) in the shadow area and the bright area can be enhanced separately, avoiding the loss of local details caused by global processing;

[0060] Contrast Limit: Set clipLimit=2.0, which means that when the number of pixels in a single gray level exceeds 2% of the total pixels in the sub-block, the excess part is evenly distributed to other gray levels to prevent dust noise or local reflections from being over-amplified into "pseudo-texture" and to ensure that the enhancement focuses on the real lithological features.

[0061] Boundary smoothing: Bilinear interpolation is used to process overlapping areas of sub-blocks to avoid fractures in rock texture caused by block division, and to ensure the continuity of cross-regional features such as joints and fissures.

[0062] S2: Establish a rock mass identification model, and identify the preprocessed image based on the rock mass identification model to obtain the lithological distribution results of the excavation face, including lithological type and the area ratio of each type of lithology in the excavation face.

[0063] A rock mass identification model is established based on the U-Net network, using the cross-entropy function as the loss function, the preprocessed rock mass image of the excavation face as the input feature, and the lithology category as the true label. The U-Net network is trained, and the predicted lithology categories of the U-Net network are used during the training process. The loss function is optimized by comparing it with the true labels; the specific loss function is as follows: ;

[0064] The area ratio of various lithologies in the excavation face is calculated as follows: ,in Indicates lithological category, , Indicates the first The area percentage of similar lithologies in the excavation face. For indicator functions, Representing coordinates The lithology at that location meets the requirements. hour, Select 1 if the value is 1, otherwise select 0. This represents the total number of pixels in the excavation face image.

[0065] S3: Establish a calculation model for the explosiveness index based on the results of flatness and lithological distribution.

[0066] like Figure 3 As shown, the flatness calculation includes: selecting the baseline of the excavation surface (such as the tangent of the design outline). Calculate the actual coordinates of the rock mass in the excavation face image. Distance to the baseline The maximum distance The ratio of the average distance to the maximum distance is taken as the flatness.

[0067] The explosibility index calculation model is wherein is the explosibility index, is the flatness is the weight coefficient of the flatness, is the area proportion of different lithology in the excavation surface, is the weight coefficient of the corresponding lithology.

[0068] In the embodiment, the value of the weight coefficient is obtained based on the training of historical blasting data by the neural network.

[0069] S4: The image of the target rock mass is collected in the same manner as in step S1, and the explosibility index of the target rock mass is calculated based on the explosibility index calculation model.

[0070] S5: The blasting parameters of the target rock mass are adjusted in combination with the explosibility index of the target rock mass and the overbreak volume, the underbreak volume and the flatness.

[0071] Overbreak area and underbreak area calculation: the actual excavation contour line is obtained from the enhanced image based on the edge detection algorithm, and the overbreak volume and the underbreak volume are calculated based on the actual excavation contour line and the design excavation contour line according to the scale of the pixel area and the actual area, as shown in Figure 2 ; specifically, , , represents the pixel points inside the actual contour line but not inside the design contour line, represents the pixel points inside the design contour line but not inside the actual contour line. If the overbreak volume is greater than (the set value, and the design excavation area), it is suggested to reduce the peripheral hole charge amount, and the specific calculation method of the new charge amount is as follows: ,

[0072] is the initial design charge amount. If the flatness is greater than the set value 0.3, it is suggested to increase the blast hole spacing , . If the flatness is greater than the set value 0.3, it is suggested to increase the blast hole spacing ,

[0073] . If the flatness is greater than the set value 0.3, it is suggested to increase the blast hole spacing ,

[0074] .​Figure 4 The explosibility index is graded as shown, When the explosibility grade is low, it is suggested to increase the charge amount of the cut slot by 10%-15%, the hole spacing is reduced to 90%, and the initiation time difference is shortened by 5%. When the explosibility is medium, the cut slot charge amount, hole spacing and initiation time difference remain unchanged. When the explosibility is high, it is suggested to reduce the charge amount of the peripheral hole by 10%, increase the hole spacing by 10%, and extend the initiation time difference by 5%.

[0075] The tunnel excavation face blasting effect evaluation system based on digital image technology comprises:

[0076] An image acquisition module is configured to acquire excavation rock mass images according to a set image acquisition frequency within a set time;

[0077] An image preprocessing module is configured to preprocess the rock mass images acquired by the image acquisition module;

[0078] A rock mass identification module is configured to obtain a lithology distribution result of the preprocessed rock mass images, including a lithology category and an area proportion of different lithologies in the excavation face;

[0079] An explosibility index calculation module is configured to calculate the rock mass explosibility index according to the lithology distribution result;

[0080] A blasting parameter guidance module is configured to output target rock mass blasting parameters according to the explosibility index and overbreak, underbreak and flatness.

Claims

1. A method for evaluating the blasting effect of a tunnel excavation face based on digital image technology, characterized in that, The method comprises: S1: collecting the rock mass image of the excavation face after blasting, and pre-processing the image; S2: establishing a rock mass identification model, identifying the pre-processed image based on the rock mass identification model, and obtaining the lithology distribution result of the excavation face, including the lithology category and the area proportion of each lithology in the excavation face; S2 specifically comprises: A rock mass identification model is established based on the U-Net network, using the cross-entropy function as the loss function, the preprocessed rock mass image of the excavation face as the input feature, and the lithology category as the true label. The U-Net network is trained, and the predicted lithology categories of the U-Net network are used during the training process. The loss function is optimized by comparing it with the true labels; the specific loss function is as follows: ; The area proportion calculation method of each lithology in the excavation face is: wherein represents the lithology category, , represents the area proportion of the first lithology in the excavation face, is an indicator function, represents the lithology category at the coordinate , and satisfies , 1 is taken, otherwise 0 is taken, is the total number of pixel points in the excavation face image; S3: establishing a blastability index calculation model based on the flatness and lithology distribution results, the blastability index calculation model being wherein is a blastability index, is a flatness weight coefficient, is an area proportion of different lithology in the excavation face, is a weight coefficient of the corresponding lithology; S4: calculating the blastability index of the target rock mass based on the blastability index calculation model; S5: adjusting the target rock mass blasting parameters in combination with the blastability index of the target rock mass and the overbreak, underbreak and flatness.

2. The method for evaluating the blasting effect of a tunnel excavation face according to claim 1, characterized in that, The collection of the rock mass image of the excavation face after blasting in S1 comprises: Image collection preparation: marker layout and image collection equipment installation; Image collection: collecting the excavation rock mass image according to the set image collection frequency within the set time after blasting.

3. The method for evaluating the blasting effect of a tunnel excavation face according to claim 2, characterized in that, The pre-processing of the image in S1 comprises: Image denoising: removing high-frequency noise in the image by using Gaussian filtering method; Image registration: taking the reference marker as the feature point, calculating the conversion matrix M of the design coordinate system and the image coordinate system, and obtaining the registered image: Contrast enhancement: using adaptive histogram equalization to improve the lithology texture clarity in the image. 4.The method of claim 1, wherein, Weight coefficients obtained based on neural network identification of historical blasting data values.

5. The method for evaluating the blasting effect of a tunnel excavation face according to claim 1, characterized in that, The adjustment of the rock mass blasting parameters in combination with the blastability index of the target rock mass and the overbreak, underbreak and flatness in S4 comprises: Correcting the peripheral hole charge weight through the overbreak or underbreak; Optimizing the blast hole density according to the flatness; Classifying the blastability index, and adjusting the slotting hole charge weight, hole spacing and initiation time difference according to the blastability index grade.

6. The method for evaluating the blasting effect of a tunnel excavation face according to claim 5, characterized in that, The overbreak and underbreak calculation comprises: The actual excavation contour line is obtained from the preprocessed image based on an edge detection algorithm The overbreak and underbreak are calculated based on the actual excavation contour line and the design excavation contour line according to the scale of the pixel area and the actual area , , , , The pixel points inside the actual contour line but not inside the design contour line are represented by The pixel points inside the design contour line but not inside the actual contour line are represented by 7.The method of claim 5, wherein, Smoothness calculation includes: selecting the baseline of the excavation surface. Calculate the actual coordinates of the rock mass in the excavation face image. Distance to baseline , to the maximum distance Average distance The ratio of the two values ​​is used to measure the flatness.

8. A system for evaluating the blasting effect of a tunnel excavation face based on digital image technology, for implementing the method for evaluating the blasting effect of a tunnel excavation face based on digital image technology according to any one of claims 1-7, characterized in that, The system comprises: An image collection module for collecting the rock mass image of the excavation face according to the set image collection frequency within the set time; An image preprocessing module for pre-processing the rock mass image collected by the image collection module; A rock mass identification module for obtaining the lithology distribution result of the pre-processed rock mass image, including the lithology category and the area proportion of different lithologies in the excavation face; A blastability index calculation module for calculating the rock mass blastability index according to the lithology distribution result; A blasting parameter guidance module for outputting the target rock mass blasting parameters according to the blastability index and the overbreak, underbreak and flatness.

Images