Tunnel blasting over-break and under-break control method considering rock mass structural parameters

Abstract

The present application relates to the field of rock and soil blasting technology, and provides a method for controlling over-excavation and under-excavation of tunnel blasting taking into account rock mass structural parameters, using a three-dimensional laser scanner to scan the tunnel face and surrounding rock; obtaining surrounding rock structural parameters and blasting effect evaluation parameters; based on the current face image and scanning data, sorting out the linear relationship between surrounding rock structural parameters and blasting parameters and blasting effect evaluation parameters, and establishing an intelligent model based on a deep learning algorithm; inputting data with good on-site blasting effects into the constructed intelligent model for training and supplementing the data set; inputting blasting effect evaluation parameters and surrounding rock structural parameters after assignment into the intelligent model to obtain optimized blasting parameters; performing three-dimensional laser visualization point layout based on the optimized blasting parameters, implementing an optimized blasting plan, and realizing over-excavation and under-excavation control of tunnel blasting. Ensure the adaptability of the blasting plan to the surrounding rock and geological environment of super-large sections, effectively control over-excavation and under-excavation of blasting, reduce engineering costs and improve engineering construction efficiency.

Description

Technical Field

[0001] The invention belongs to the technical field of rock blasting, and in particular relates to a tunnel blasting over-break and under-break control method taking into account rock mass structural parameters. Background Art

[0002] The drilling and blasting method is widely used in tunnel construction due to its low cost and strong geological applicability. Due to the complex geological conditions, there are often a large number of stratification and joint fissures in the rock mass, and the drilling and blasting method is affected by the "instantaneous" characteristics of work, making it difficult to accurately control the over-excavation and under-excavation phenomenon during engineering application. Even landslides frequently occur due to blasting disturbances, affecting construction progress, quality and safety.

[0003] In view of the complexity of surrounding rock conditions and blasting process, rock mass structural parameters have a great influence on the smooth surface blasting effect of tunnels. Therefore, on the basis of satisfying normal blasting functions, it is necessary to reasonably control the blasting process, not only to reasonably and efficiently crush the rock mass within the excavation contour, but also to reduce the damage or disturbance to the rock mass outside the excavation area, so as to more effectively maintain the safety and stability of the surrounding rock and effectively control the occurrence of over-excavation and under-excavation. Therefore, it is necessary to optimize the smooth surface blasting parameters based on the blasting over-excavation and under-excavation effects on the basis of considering the rock mass structural parameters. However, at this stage, there is no systematic theory and method that can be directly applied to quickly determine the optimization scheme of smooth surface blasting parameters.

[0004] Therefore, it is necessary to design a set of control methods for blasting over-break and under-break in super-large cross-section hard rock tunnels that can take into account the rock mass structural parameters. Summary of the invention

[0005] In response to the above-mentioned problems in the prior art, the present application proposes a tunnel blasting over-break and under-break control method that takes into account the rock mass structural parameters. The blasting parameters are optimized based on the rock mass structural parameters and the over-break and under-break effect evaluation, and real-time feedback on the effect of the blasting construction is provided to ensure the quality and safety of the project.

[0006] The present invention proposes a tunnel blasting over-break and under-break control method taking into account rock mass structural parameters, comprising the following steps:

[0007] Carry out tunnel blasting and expose the tunnel face according to the on-site blasting plan;

[0008] Use a 3D laser scanner to scan the tunnel face and surrounding rock to obtain surrounding rock structure parameters and blasting effect evaluation parameters;

[0009] According to the current face image and scanning data, the linear relationship between the surrounding rock structure parameters and the blasting parameters and the blasting effect evaluation parameters is sorted out, and an intelligent model is established based on the deep learning algorithm; the data with good blasting effects on site are input into the constructed intelligent model for training and supplementing the data set;

[0010] Input the blasting effect evaluation parameters and the assigned surrounding rock structure parameters into the intelligent model to obtain the optimized blasting parameters;

[0011] Based on the optimized blasting parameters, three-dimensional laser visualization is used to arrange the points, implement the optimized blasting plan, and realize the control of over-excavation and under-excavation of tunnel blasting.

[0012] Furthermore, after the step of implementing the optimized blasting scheme, the method further includes:

[0013] The three-dimensional laser scanner is used again to scan the tunnel face and surrounding rock to obtain surrounding rock structure parameters and blasting effect evaluation parameters again, and the intelligent model is used to continue to optimize the blasting parameters.

[0014] Furthermore, it also includes: inputting blasting cost data into an intelligent model to obtain optimized blasting parameters, calculating blasting costs based on the optimized blasting parameters, and implementing an optimized blasting plan.

[0015] Furthermore, the step of establishing the intelligent model includes:

[0016] Creating a training sample database; the sample data in the sample database includes surrounding rock structure parameters, blasting parameters and blasting effect evaluation parameters;

[0017] The four predicted values ​​of the blasting parameters are set at the end of the existing neural network structure to obtain a neural network intelligent model capable of optimizing the tunnel blasting parameters;

[0018] The Bayesian regularization algorithm is selected as the training algorithm, and the acquired sample data is input into the established intelligent model for iterative training.

[0019] Furthermore, the step of establishing the intelligent model also includes:

[0020] The correlation coefficient between the true value and the predicted value is used to characterize the prediction accuracy of the intelligent model:

[0021]

[0022] Where: is the correlation coefficient; , Respectively represent the true value and predicted value of the blasting parameter, where is the input value, is the output value; Index for data; is the sample size;

[0023] When | |≤1, and if | The closer | is to 1, the greater the correlation between the true value and the predicted value; if | The closer | is to 0, the smaller the correlation between the true value and the predicted value.

[0024] Furthermore, the four predicted values ​​of the blasting parameters include slotting hole parameters, peripheral hole spacing, peripheral hole positions and charging parameters.

[0025] Further,

[0026] The blasting effect evaluation parameters include: blast hole trace rate, block size distribution, over-excavation and under-excavation value and blasting cost;

[0027] The surrounding rock fabric parameters include: rock mass dominant structural surface attitude parameters, rock mass integrity coefficient and joint spacing.

[0028] Furthermore, the step of scanning the tunnel face and surrounding rock with a three-dimensional laser scanner and obtaining surrounding rock structure parameters and blasting effect evaluation parameters comprises:

[0029] Use a 3D laser scanner to scan the surrounding rock exposed by tunnel blasting to obtain point cloud data;

[0030] Process and clean the collected point cloud data to remove noise and irrelevant data and retain valid internal information of the tunnel;

[0031] The improved Faster R-CNN method is used to intelligently process the visible blast holes on the tunnel side wall after blasting. Finally, by comparing the point cloud data at different time points, the blast hole traces and damage left after the blasting are identified, and then the blast hole trace rate is calculated.

[0032] Furthermore, the step of scanning the tunnel face and surrounding rock with a three-dimensional laser scanner and obtaining surrounding rock structure parameters and blasting effect evaluation parameters comprises:

[0033] The blasting slag image is obtained by scanning the tunnel face and surrounding rock using the three-dimensional laser scanner;

[0034] Build the Mask-R-CNN model, and perform model training and slag image segmentation;

[0035] After the training, HSV transformation and morphological optimization are performed, and HSV transformation is applied to enhance the color information in the slag explosion image;

[0036] The block size statistics are performed on the segmented blasting slag image, the size, shape and distribution of each block are calculated, the statistical results are analyzed, the block size distribution and quality of the blasting pile are evaluated, and the block size distribution statistics are obtained.

[0037] Furthermore, the step of obtaining the rock mass integrity coefficient of the surrounding rock fabric parameters includes:

[0038] The integrity coefficient of the surrounding rock mass is calculated by the ratio of the longitudinal wave velocity of the rock mass to the rock mass. :

[0039]

[0040] Where: It represents the elastic longitudinal wave velocity of the rock mass in km / s, which is obtained by conducting acoustic wave tests on the rock mass on site; It represents the elastic longitudinal wave velocity of rock in km / s, which is obtained by conducting acoustic wave tests on rock samples in the laboratory.

[0041] The beneficial effects of the present invention are as follows: after the blasting of the face, the rock mass structural parameters and blasting effect evaluation parameters are obtained by using three-dimensional laser scanning technology, and the scanned parameters are input into the intelligent model, so that the blasting optimization parameters can be quickly output, thereby ensuring the adaptability of the blasting scheme to the super-large section surrounding rock and geological environment, and effectively controlling the over-excavation and under-excavation of blasting, thereby reducing the project cost and improving the construction efficiency of the project; compared with the existing blasting parameter optimization method, the present invention takes into account the influence of surrounding rock structural parameters such as joint occurrence on the blasting effect, so that the blasting parameter optimization is faster and the scientific nature of the parameter optimization results is ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is the surrounding rock joint distribution and blasting effect diagram of the present invention;

[0043] Figure 2 It is a schematic diagram of the implementation steps of the method of the present invention;

[0044] Figure 3 It is a structural diagram of the neural network regression prediction algorithm of the present invention;

[0045] Figure 4 It is a schematic diagram of the formation of a single joint fracture surface of the present invention;

[0046] Figure 5 It is the stress state of the surrounding rock unit body of the present invention;

[0047] Figure 6 This is the relationship between the joint structural surface and the fracture control surface of the present invention. DETAILED DESCRIPTION

[0048] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0049] like Figure 1 In the surrounding rock joint distribution shown, J1 and J2 are two filling joints, and J3, J4, and J5 are three closed joints.

[0050] like Figure 2 The method of the present invention is shown as follows, comprising:

[0051] Conduct 3D laser scanning on the tunnel face after on-site blasting construction to obtain blasting effect evaluation parameters and surrounding rock structure parameters;

[0052] Input blasting effect evaluation parameters, assigned surrounding rock structure parameters and blasting cost data into the intelligent model to obtain optimized blasting parameters;

[0053] Calculate blasting cost based on optimized blasting parameters;

[0054] Based on the output optimized blasting parameters, three-dimensional laser visualization is used to arrange points and control over-excavation and under-excavation of blasting is achieved.

[0055] Among them, blasting parameters include slotting parameters, peripheral hole spacing, peripheral hole position and charging parameters. Blasting effect evaluation parameters include trace rate (R), fragmentation distribution (S) and over-excavation value (CW), blasting cost parameters. Surrounding rock fabric parameters include rock mass dominant structural surface occurrence parameters (joint strike, joint dip, joint inclination), rock mass integrity coefficient (Kv), joint spacing (s).

[0056] like Figure 3 The neural network regression prediction algorithm structure of the present invention is shown. The blasting effect evaluation parameters and the assigned surrounding rock structure parameters are input into the intelligent model, and the intelligent model automatically outputs the optimized blasting parameters: the surrounding rock structure parameters and blasting effect evaluation parameters are selected as the tunnel blasting effect prediction evaluation index system and combined with the engineering requirements of tunnel blasting parameter optimization, four regression prediction parameters are set at the end of the existing neural network algorithm structure, namely, the slotting eye parameters, the peripheral eye spacing, the peripheral eye position, and the charging parameters, to construct a neural network structure for tunnel blasting parameter optimization.

[0057] In terms of the rapid acquisition of surrounding rock structure parameters of the present invention, for the tunnel face exposed after on-site blasting, the scanner can obtain the three-dimensional point cloud data of the tunnel face. By introducing technologies such as deep learning and convolutional neural networks, it can intelligently identify and classify different lithologies, structural surfaces and joints in the point cloud data, and achieve a detailed analysis of the surrounding rock structure. This process greatly improves the automation and accuracy of data processing, and demonstrates the powerful ability of computer vision in the analysis of complex geological environments. At the same time, when interpreting point cloud data, innovative filtering algorithms and feature extraction technologies are used to effectively remove noise and retain key information, so that the geometric and texture features of the surrounding rock can be presented in detail, providing a solid foundation for the evaluation of blasting effects. By accurately matching the three-dimensional point cloud data with the designed intelligent model on site, and using an automated calculation process to quickly analyze the differences between the two, the effect of the blasting construction can be fed back in real time to ensure the quality and safety of the project.

[0058] In terms of quickly obtaining blasting effect evaluation parameters, laser scanner equipment is first used to scan the surrounding rock exposed by tunnel blasting to obtain point cloud data. Secondly, the collected point cloud data is processed and cleaned to remove noise and irrelevant data and retain effective internal information of the tunnel. The improved Faster R-CNN method is used to intelligently process the visible blast holes on the side wall of the tunnel after blasting. Finally, by comparing the point cloud data at different time points, the traces and damage left after the blasting are identified, and then the trace rate is calculated.

[0059] In order to improve the self-perception ability of slag image segmentation and flexibly count the blockiness information in the slag image, the present invention comprehensively considers the advantages of deep CNN and traditional algorithms, and proposes a new blockiness statistics method that combines Mask-R-CNN and HSV transformation with morphological optimization to perform blockiness distribution statistics. Specifically, the following steps are included: first, data collection and preprocessing are performed, and then model selection and training are performed to develop a Mask-R-CNN model suitable for slag image segmentation. After training, HSV transformation and morphological optimization are performed. Before or after slag image segmentation, HSV transformation is applied to enhance the color information in the image, which helps to better identify the boundaries and blockiness of the ore rock. Finally, blockiness statistics and analysis are performed, and blockiness statistics are performed on the segmented slag image, and the size, shape and distribution of each block are calculated. The statistical results are analyzed, the blockiness distribution and quality of the blast pile are evaluated, and blockiness distribution statistics are obtained.

[0060] A working condition data set with good blasting effect is constructed and selected for training data set to train the intelligent model of the present invention, so that the deep learning model can learn the complex mapping relationship between blasting effect and surrounding rock conditions. Compared with traditional rule-based or experience-based methods, this learning method based on actual successful cases can better capture subtle changes and uncertainties in the blasting process, and has more advantages in dealing with complex nonlinear problems than traditional methods. The surrounding rock structure parameters and over-excavation evaluation parameters (or blasting effect evaluation parameters) are used as input parameters of the intelligent model, and the blasting control parameters (or blasting parameters) are used as output parameters of the intelligent model. The difference between the input parameters and the output parameters makes the intelligent model more flexible and generalizable. The input parameters include surrounding rock structure parameters and over-excavation evaluation parameters, which reflect the diversity and complexity of the blasting environment; and the output parameters are blasting control parameters, which directly guide the blasting construction. The intelligent model can automatically extract useful information from the input parameters and generate accurate output parameters. Using the intelligent model to carry out model training, the fully trained deep learning model can automatically optimize the blasting parameters based on the input surrounding rock condition changes to achieve better blasting effects. This prediction is not only fast and accurate, but also can adapt to different geological conditions and engineering needs, providing strong technical support for blasting construction. Compared with traditional methods, the prediction results of deep learning models are more reliable and accurate, bringing significant benefits and safety improvements to engineering practice. After the model training is completed, the blasting parameters can be automatically optimized based on the changes in surrounding rock conditions. After obtaining the optimized blasting parameters, three-dimensional laser visualization points can be used to finally optimize the blasting plan to achieve the effect of controlling over-excavation and under-excavation.

[0061] The present invention increases the influence weight of the rock mass fabric parameters in the input parameters of the intelligent model and re-assigns the weight, so that accurate optimized blasting parameters can be obtained for different surrounding rocks.

[0062] The present invention adopts a deep learning method to construct a nonlinear mapping model between surrounding rock structure parameters and blasting effect evaluation parameters, and finally gives the blasting parameters under the control of target over-break and under-break indicators, which can be applied to over-break and under-break control under working conditions of surrounding rocks of different grades, lithologies and different section types.

[0063] The intelligent model of the present invention is established by the following steps:

[0064] Creation of training sample database: A sample database is obtained based on the construction of a super-large cross-section tunnel under construction, including surrounding rock structure parameters, blasting parameters, and blasting effect evaluation parameters;

[0065] Model training and accuracy analysis: The Bayesian regularization algorithm is selected as the training algorithm, and the acquired sample data is input into the established intelligent model for training. After n iterations, the training error tends to be stable, indicating that the parameters of the neural network model are gradually stable. In order to further evaluate the reliability of the established neural network prediction model, the correlation coefficient between the true value and the predicted value is used to characterize it. The closer the correlation coefficient value is to 1, the more linear the correlation between the true value and the predicted value of the blasting parameter is, indicating that the established tunnel blasting effect neural network evaluation model has sufficient reliability.

[0066] Engineering application and verification: To verify the reliability and practicality of the optimization results of the intelligent model, we can rely on the application and verification work of the subsequent construction blasting effect prediction and blasting parameter optimization of the project. First, the established intelligent model will be applied to the subsequent blasting effect prediction and blasting parameter optimization of the actual project, and verification work will be carried out; the verification results will be analyzed and evaluated to determine the reliability and practicality of the intelligent model; according to the verification results, the intelligent model will be optimized and improved as necessary to continuously improve its accuracy and reliability; the performance of the intelligent model will be continuously monitored and evaluated, and adjustments and improvements will be made according to actual conditions to ensure its effective application in engineering blasting, thereby realizing the monitoring, evaluation, control and optimization of the blasting process.

[0067] Among the parameters of surrounding rock structure, joint occurrence is one of the main factors affecting blasting effect. In order to ensure the rationality of the intelligent model, it is necessary to give higher weight to the joint occurrence factor in parameter analysis to ensure accurate description and evaluation of the joint characteristics of the surrounding rock, so as to more accurately optimize the blasting plan and formulate a more effective over-excavation and under-excavation control plan.

[0068] The establishment of an intelligent model provides specific indicator parameters for the evaluation of tunnel blasting effects. Four prediction values ​​are set at the end of the existing neural network structure, namely, the slotting eye parameters, the peripheral eye spacing, the peripheral eye position and the charging parameters, to obtain a neural network model that can achieve tunnel blasting parameter optimization.

[0069] The Bayesian regularization algorithm is selected as the training algorithm during model training, and the sample data is input into the neural network model for training. In order to further evaluate the reliability of the established model, the correlation coefficient between the true value and the predicted value is selected to characterize the prediction accuracy:

[0070] (1)

[0071] Where: is the correlation coefficient; x and y represent the true value (input value) and predicted value (output value) of the blasting parameter respectively; i is the data index; n is the sample size. |≤1, and if | The closer | is to 1, the greater the correlation; if | The closer | is to 0, the smaller the correlation is.

[0072] When evaluating the blasting effect:

[0073] The blasthole trace rate R requires that the traces of the surrounding blastholes be evenly distributed on the excavation contour surface. The surrounding holes do not include the surrounding holes of the bottom plate during calculation. When the blasthole trace preservation rate is greater than 70% of the hole length, it should be calculated as a visible effective trace blasthole. The calculation formula is shown in formula (2):

[0074] (2)

[0075] In the formula, R is the blasthole trace rate; n is the number of blastholes that can be counted on the contour line; N is the number of surrounding holes during drilling. The standard requirements for blasthole trace rate under different surrounding rock conditions are shown in Table 1:

[0076] Table 1 Blast hole trace rate standard

[0077]

[0078] If the blast hole trace rate in the actual site exceeds the standard value of the blast hole trace rate, it proves that the blasting quality is good, the construction technology level is high, the geological conditions are suitable, and the quality control standards are effectively met; if the blast hole trace rate in the actual site is lower than the standard value of the blast hole trace rate, it proves that the blasting effect is not ideal, there are problems with the construction technology, the geological conditions are complex, or the quality control is not strict. In order to improve the blast hole trace rate, it is necessary to take corresponding measures to improve and optimize according to the specific reasons.

[0079] The contour data of over-break and under-break control can be obtained through various methods such as on-site measurement, design drawings and calculations, automated monitoring and feedback, and other high-tech means. In actual projects, the appropriate acquisition method should be selected according to the specific situation to ensure the accuracy and reliability of the data. Over-break and under-break control requires that the excavation contour is smooth, without under-break or obvious over-break. The specification requirements for over-break values ​​allowed under different surrounding rock conditions are shown in Table 2.

[0080] Table 2 Allowable over-excavation value of tunnel (unit: cm)

[0081]

[0082] If the over-excavation value of the contour data in the actual project exceeds the allowable over-excavation value of the tunnel, it proves that the blasting effect is poor. The project should adjust the blasting parameters such as the amount of explosives, charging structure and detonation sequence according to the actual geological conditions and construction experience to improve the blasting control accuracy and keep the over-excavation value within the allowable range.

[0083] The blasting slag block size should be appropriate, avoid overly large blocks, avoid overly fine gravel, and have a reasonable block size distribution.

[0084] Blasting cost parameters are divided into direct cost parameters and indirect cost parameters, which can be calculated directly. Since the specific blasting cost parameters and standard values ​​may vary depending on factors such as region, project type, rock properties, etc., they need to be adjusted and analyzed according to specific circumstances in actual operations. The rationality of the current blasting cost can be evaluated by comparing the results with the historical data of the project. If the cost is too high or too low, if the blasting cost is higher than the historical data, it proves that the blasting optimization effect is poor, and it is necessary to analyze the reasons and take corresponding measures to adjust.

[0085] Among them, calculating the blasting cost includes the following steps: determining the scope and scale of the blasting, including the blasting area, depth, rock type, geological conditions, etc.; making estimates based on the actual blasting plan, including blasting design, blasting parameters, detonation sequence, etc.; statistically analyzing the actual use of blasting materials, including explosives, detonators, detonators, packaging materials, etc.; in addition to blasting materials, it is also necessary to consider the cost of labor and equipment and auxiliary costs. In actual operations, the cost estimation table should be updated in a timely manner, and adjusted and optimized according to actual conditions. When the estimated blasting cost is within a reasonable range, subsequent blasting construction operations should be carried out.

[0086] The surrounding rock mass integrity factor (usually expressed as K v ) is an important indicator for evaluating the integrity of rock mass. It mainly reflects the quality of rock mass from the perspective of mechanical integrity, and thus indirectly affects the blasting effect. It is calculated by the ratio of the longitudinal wave velocity of rock mass to that of rock, as shown in formula (3). K v Between 0 and 1, the closer the value is to 1, the better the integrity of the surrounding rock; and when the integrity coefficient is close to 0, it means that the integrity of the surrounding rock is very poor.

[0087] (3)

[0088] Where: It represents the elastic longitudinal wave velocity of the rock mass in km / s, which is obtained by conducting acoustic wave tests on the rock mass on site; It represents the elastic longitudinal wave velocity of rock in km / s, which is obtained by conducting acoustic wave tests on rock samples in the laboratory.

[0089] The parameters of the dominant structural surface of the rock mass (joint strike, joint inclination, joint inclination) have a significant impact on the blasting effect. These parameters not only determine the stress distribution and energy transfer mode inside the rock mass, but also directly affect the degree of rock mass fragmentation after blasting, over-excavation and under-excavation values, and other blasting parameters and blasting stability.

[0090] The influence of joint orientation on blasting effect is mainly reflected in the formation and extension direction of blasting fracture line. When the joint orientation is consistent with or close to the blasting direction, the blasting energy is easy to propagate along the joint surface, causing the fracture line to develop along the joint surface, forming longer cracks and larger broken blocks. This helps to increase the crushing effect of blasting, but it may also lead to problems such as slope instability; the influence of joint inclination on blasting effect is mainly reflected in the distribution of blasting energy and stress concentration. When the joint inclination is perpendicular or nearly perpendicular to the blasting direction, the blasting energy is easy to produce stress concentration near the joint surface, resulting in more serious local rock mass fragmentation. However, this may also increase the instability of the rock mass after blasting and the complexity of cracks; the influence of joint inclination on blasting effect is multifaceted. Large or small joint inclination may cause the blasting energy to disperse or lose, resulting in poor blasting effect.

[0091] Different joint spacings (s) have a significant impact on the blasting effect. The larger the joint spacing, the lower the degree of rock fragmentation. As the joint spacing increases, the stress wave generated by the explosive explosion encounters fewer joint surfaces during propagation, resulting in energy dispersion and attenuation, which reduces the rock fragmentation effect, resulting in large rock blocks and poor blasting effect; the joint spacing has a certain influence on the expansion direction and length of the cracks. When the joint spacing is small, the cracks generated by the blasting tend to expand along the joint surface to form a complex crack network. When the joint spacing is large, the expansion of the cracks may be more restricted, making it difficult to form effective through cracks, resulting in poor blasting effect; the joint spacing also affects the concentration and release of blasting energy. When the joint spacing is small, the blasting energy tends to be concentrated near the joint surface, resulting in severe local rock fragmentation. When the joint spacing is large, the blasting energy may be dispersed in a wider area, reducing the local rock fragmentation effect, which may lead to serious over-excavation and under-excavation.

[0092] The above input parameters directly or indirectly affect the engineering blasting effect, and the surrounding rock structure parameters have the most significant impact on the blasting effect in the above analysis, so they should be given a higher proportion in model construction. The present invention trains the intelligent model through the relationship data between the blasting effect and the blasting parameters, and then explores the most effective blasting parameter design scheme for different blasting effect requirements.

[0093] The effect of joint occurrence on blasting is as follows:

[0094] Combined with Figure 4 , Figure 5 , Figure 6 As shown in Figure 2, the conditions that must be met to achieve an ideal smooth blasting effect in jointed rock formations are:

[0095] (4)

[0096] Where: The angle between the fracture control surface, i.e. the design contour line and the normal direction of the structural surface, is called the failure angle; The angle between the fracture control surface and the structural surface is called the dip angle; , are the normal stresses in the x- and y-axis directions of the surrounding rock microelement, i.e., the radial stress component and the annular stress component generated by the explosion stress wave; is the dynamic tensile strength of rock; , is the cohesion and internal friction angle of the joint structure surface; is Poisson's ratio.

[0097] During the blasting process, the convexity of the fracture surface refers to the degree of unevenness on the fracture surface of the rock after blasting. Blasting will cause the rock to break and fracture, forming an irregular fracture surface. The magnitude of the convexity of the fracture surface can reflect the blasting effect, the degree of rock fragmentation, and the deformation of the rock. By observing and measuring the convexity of the fracture surface, the blasting effect and the stability of the rock can be evaluated, providing a reference for subsequent engineering design and construction. When there is only a single joint between two adjacent blastholes, the calculation formula for the convexity of the fracture surface is:

[0098] (5)

[0099] Where: E is the distance between blast holes, is the distance from the left blasthole to the joint, is the distance from the right blasthole to the joint, It is the convexity of the contour line around the rock blasting.

[0100] From formula (5), we can see that when β When the angle is less than 45°, the concave-convex degree of the fracture surface increases with the increase of the angle. β =45°, the concavity is the largest. β When the angle is greater than 45°, the concave-convex degree decreases with the increase of the angle. When the angle between the joint and the blasthole is greater than 70°, the joint has no effect on the blasting forming effect.

[0101] It can be seen that with the increase of joint length, the degree of over-break and under-break first increases and then decreases. When the angle between the joint and the blasthole is 45 degrees, the over-break and under-break of rock mass blasting is the most serious. At the same time, the smaller the hole spacing, the less obvious the over-break and under-break phenomenon. If you want to reasonably control the over-break and under-break, you should reasonably adjust the hole spacing according to the size of the failure angle; when the intersection of the joint and the excavation contour line is in the middle of the two blastholes, the internal stress concentration and over-break and under-break phenomenon generated on the joint are more prominent. At this time, the best way to deal with over-break and under-break is to reduce the charge of the adjacent blastholes at the intersection; when the position of the blasthole is on the joint, the existence of the joint will seriously affect the kinetic energy release of the blasthole. The blasthole orientation should be reasonably adjusted, and the hole spacing should be reasonably changed, and explosives should not be placed inside the joint as much as possible.

[0102] The present invention can quickly and accurately evaluate the blasting effect and provide a quantifiable fracture surface convexity index, so as to objectively evaluate the degree of rock fragmentation, deformation and stability, and provide a scientific basis for engineering design and construction. The blasting scheme can be quickly optimized according to different surrounding rock conditions to ensure that the blasting operation is carried out efficiently and safely under various complex surrounding rock conditions. By reducing over-excavation and under-excavation in a targeted manner, the blasting cost can be effectively reduced and the efficiency of the blasting operation can be improved.

[0103] Based on the optimized blasting parameters (or blasting optimization parameters) output by the intelligent model, the blasting plan is redesigned, and three-dimensional laser visualization is carried out on site to achieve real-time adjustment of blasting parameters, ensure the safety and effectiveness of the blasting process, and achieve optimization of blasting parameters and effective control of over-excavation and under-excavation.

[0104] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A tunnel blasting over-break and under-break control method considering rock mass structural parameters, characterized in that: The following steps are involved: Carry out tunnel blasting and expose the tunnel face according to the on-site blasting plan; Use a 3D laser scanner to scan the tunnel face and surrounding rock to obtain surrounding rock structure parameters and blasting effect evaluation parameters; According to the current face image and scanning data, the linear relationship between the surrounding rock structure parameters and the blasting parameters and the blasting effect evaluation parameters is sorted out, and an intelligent model is established based on the deep learning algorithm; the data with good blasting effects on site are input into the constructed intelligent model for training and supplementing the data set; Input the blasting effect evaluation parameters and the assigned surrounding rock structure parameters into the intelligent model to obtain the optimized blasting parameters; Based on the optimized blasting parameters, 3D laser visualization is used to arrange the points, and the optimized blasting plan is implemented to achieve the over-break and under-break control of tunnel blasting; The steps of establishing the intelligent model include: Creating a training sample database; the sample data in the sample database includes surrounding rock structure parameters, blasting parameters and blasting effect evaluation parameters; The four predicted values ​​of the blasting parameters are set at the end of the existing neural network structure to obtain a neural network intelligent model capable of optimizing the tunnel blasting parameters; The Bayesian regularization algorithm is selected as the training algorithm, and the acquired sample data is input into the established intelligent model for iterative training; The four predicted values ​​of the blasting parameters include the parameters of the cutout holes, the spacing between the peripheral holes, the positions of the peripheral holes and the charging parameters; The blasting effect evaluation parameters include: blast hole trace rate, block size distribution, over-excavation and under-excavation value and blasting cost; The surrounding rock fabric parameters include: rock mass dominant structural plane attitude parameters, rock mass integrity coefficient and joint spacing; The steps of scanning the tunnel face and surrounding rock with a three-dimensional laser scanner and obtaining surrounding rock structure parameters and blasting effect evaluation parameters include: Use a 3D laser scanner to scan the surrounding rock exposed by tunnel blasting to obtain point cloud data; Process and clean the collected point cloud data to remove noise and irrelevant data and retain valid internal information of the tunnel; The improved Faster R-CNN method is used to intelligently process the visible blastholes on the tunnel side wall after blasting. Finally, by comparing the point cloud data at different time points, the blasthole traces and damage left after blasting are identified, and then the blasthole trace rate is calculated. The blasting slag image is obtained by scanning the tunnel face and surrounding rock using the three-dimensional laser scanner; Build the Mask-R-CNN model, and perform model training and slag image segmentation; After the training, HSV transformation and morphological optimization are performed, and HSV transformation is applied to enhance the color information in the slag explosion image; Perform block size statistics on the segmented blast slag image, calculate the size, shape and distribution of each block, analyze the statistical results, evaluate the block size distribution and quality of the blast pile, and obtain block size distribution statistics; After the step of implementing the optimized blasting scheme, the method further includes: The three-dimensional laser scanner is used again to scan the tunnel face and surrounding rock to obtain surrounding rock structure parameters and blasting effect evaluation parameters again, and the intelligent model is used to continue to optimize the blasting parameters.

2. The tunnel blasting over-break and under-break control method considering rock mass structural parameters according to claim 1, characterized in that: Also includes: Input the blasting cost data into the intelligent model to obtain the optimized blasting parameters, calculate the blasting cost based on the optimized blasting parameters, and implement the optimized blasting plan.

3. The tunnel blasting over-break and under-break control method considering rock mass structural parameters according to claim 1, characterized in that: The step of establishing the intelligent model also includes: The correlation coefficient between the true value and the predicted value is used to characterize the prediction accuracy of the intelligent model: ； Where: is the correlation coefficient; , Respectively represent the true value and predicted value of the blasting parameter, where is the input value, is the output value; Index for data; is the sample size; When | |≤1, and if | The closer | is to 1, the greater the correlation between the true value and the predicted value; if | The closer | is to 0, the smaller the correlation between the true value and the predicted value.

4. The tunnel blasting over-break and under-break control method considering rock mass structural parameters according to claim 1, characterized in that: The step of obtaining the rock mass integrity coefficient of the surrounding rock fabric parameters comprises: The integrity coefficient of the surrounding rock mass is calculated by the ratio of the longitudinal wave velocity of the rock mass to the rock mass. : ； Where: It represents the elastic longitudinal wave velocity of the rock mass in km / s, which is obtained by conducting acoustic wave tests on the rock mass on site; It represents the elastic longitudinal wave velocity of rock in km / s, which is obtained by conducting acoustic wave tests on rock samples in the laboratory.

Images