A tunnel face surrounding rock zoning blasting borehole parameter optimization method and system

Abstract

The application belongs to the technical field of tunnel engineering, and particularly relates to a tunnel face surrounding rock zoning blasting borehole parameter optimization method and system. The scheme integrates the surface form characteristics, joint density characteristics, resistivity characteristics and rebound strength characteristics of the tunnel face, realizes all-around perception of the surface and internal characteristics of the tunnel face, eliminates the problem of inaccurate zoning caused by the dependence of the traditional scheme on experience or single-dimensional data, improves the accuracy of parameter optimization while improving the zoning accuracy, introduces historical blasting data in the zoning process and parameter optimization process, uses the historical blasting data to effectively guide the model, further improves the accuracy of parameter optimization, simultaneously feeds back the blasting effect to the parameter optimization model, dynamically adjusts the loss function of the parameter optimization model, combines the incremental training technology, forms a self-adaptive optimization closed loop of the blasting parameter optimization, and improves the adaptability of the blasting parameter optimization model.

Description

Technical Field

[0001] This invention belongs to the field of tunnel engineering technology, specifically relating to a method and system for optimizing the parameters of blasting boreholes in the surrounding rock zone of a tunnel face. Background Technology

[0002] Although some progress has been made in the research on optimizing blasting parameters at tunnel faces, many technical shortcomings still exist in practical applications, specifically in the following aspects: (1) Most existing blasting effect assessments use three-dimensional laser scanning or close-range photogrammetry to obtain the surface morphology of the tunnel face. This method can only obtain surface information, resulting in the effect assessment being heavily reliant on a single data source and lacking effective fusion of multiple data sources.

[0003] (2) The accuracy of blasting parameter optimization depends on the precision of the partitioning algorithm. Most current partitioning algorithms are based on static rules, empirical thresholds, or single-dimensional data, which are highly subjective. At the same time, most current research on blasting parameter optimization only considers parameter optimization and lacks collaborative optimization for partitioning and parameter optimization.

[0004] (3) Current optimization methods for blasting parameters are somewhat crude and lack a rigorous feedback loop mechanism. At the same time, current optimization methods for blasting parameters are often only applicable to a certain region and lack universality. Summary of the Invention

[0005] This invention addresses the technical problems existing in the prior art by providing a method and system for optimizing the parameters of blasting boreholes in the surrounding rock of a tunnel face, effectively solving the problems present in the prior art.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a method for optimizing the parameters of blasting boreholes in the surrounding rock of a tunnel face, comprising: The tunnel face is subjected to multi-source features, including surface morphology features, joint density features, resistivity features, rebound strength features, and historical blasting features. Based on the obtained multi-source features, the current working face is partitioned using a pre-built deep learning-based partitioning model to obtain the partitioning results; Based on the obtained partitioning results, historical blasting parameters, effect evaluation data, and engineering environment parameters as input, the optimized blasting borehole parameters are obtained through a pre-built deep learning-based adaptive optimization model for borehole parameters. Specifically, after blasting using the blasting borehole parameters, on-site data is acquired and a comprehensive effect evaluation result is obtained through a preset effect evaluation strategy. Based on the obtained comprehensive effect evaluation result, the loss function of the borehole parameter adaptive optimization model is dynamically adjusted in conjunction with a preset triggering mechanism, and incremental training of the borehole parameter adaptive optimization model is performed. The deep learning-based adaptive optimization model for borehole parameters specifically performs the following processing steps: The zoning results, historical blasting parameters, effect evaluation data, and engineering environmental parameters are used as inputs. Based on the feature tensor and partition adjacency matrix in the partitioning results, the spatial topological relationship of each partition is obtained through the partition feature extraction branch. Based on historical blasting parameters and effect evaluation data, the long-term dependencies of blasting parameters are obtained through historical data coding branches; Based on engineering environmental parameters, the corresponding engineering constraint codes are obtained through environmental parameter coding branches; Based on the obtained spatial topological relationships, long-term dependencies of blasting parameters, and engineering constraint encoding, combined with an attention mechanism, cross-modal feature fusion is achieved to obtain fused features; Based on the obtained fusion features, the optimized blasting parameters for different partitions are obtained through a preset prediction head.

[0007] Furthermore, the preset effect evaluation strategy includes the following processing steps: The following data were acquired: three-dimensional point cloud data of the working face before and after blasting, multi-view image data of the blast pile, sound wave velocity test data before and after blasting, vibration monitoring time history data, and high-speed photography recording data. Based on the obtained data, three-dimensional forming quality assessment, blast block distribution assessment, blast damage assessment, and blast dynamic response assessment of the working face after blasting were performed respectively, and the assessment results were obtained. Based on the obtained evaluation results, a comprehensive effect evaluation result is obtained through fusion processing.

[0008] Preferably, a geological difficulty coefficient is introduced into the effect evaluation, and the expected benchmark of the effect evaluation results is dynamically adjusted based on the geological difficulty coefficient, wherein the geological difficulty coefficient is expressed as follows: in, For actual rock strength, As a reference rock strength, This represents the actual joint density. As the reference joint density, The rock mass integrity coefficient. The baseline rock mass integrity coefficient, These are the weighting coefficients for each dimension.

[0009] Preferably, the calculation of the comprehensive effect evaluation result adopts the following processing procedure: in, For the comprehensive effect evaluation results, The profile quality index. Geological difficulty coefficient, The quality index for over-excavation control. This is the block size quality index. The quality index of the half-hole mark. The average damage depth, The vibration control index, These are the weighting coefficients for different evaluation dimensions.

[0010] Preferably, the loss function of the borehole parameter adaptive optimization model is specifically expressed as follows: in, Predict the loss for the parameters. Due to the loss of geological consistency, For engineering constraint losses, To provide feedback on the losses, To adjust the parameters.

[0011] Preferably, the adjustment parameter of the feedback loss in the loss function is dynamically adjusted in the following manner: in, These are the initial values ​​for the feedback loss adjustment parameters in the current model's loss function. For the comprehensive effect evaluation results, To adjust the parameters.

[0012] A system for optimizing blast hole parameters in zoned blasting of surrounding rock at a tunnel face includes: The data acquisition module is used to acquire multi-source features of the tunnel face, including surface morphology features, joint density features, resistivity features, rebound strength features, and historical blasting features. The partitioning module is used to partition the current face of the tunnel based on the obtained multi-source features using a pre-built deep learning-based partitioning model, and obtain the partitioning results. The parameter optimization unit is used to obtain optimized blasting borehole parameters based on the obtained partitioning results, historical blasting parameters, effect evaluation data, and engineering environment parameters, through a pre-built deep learning-based adaptive optimization model for borehole parameters. Specifically, after blasting using the blasting borehole parameters, on-site data is acquired and a comprehensive effect evaluation result is obtained through a preset effect evaluation strategy. Based on the obtained comprehensive effect evaluation result, the loss function of the borehole parameter adaptive optimization model is dynamically adjusted in conjunction with a preset triggering mechanism, and incremental training of the borehole parameter adaptive optimization model is performed. The deep learning-based adaptive optimization model for borehole parameters specifically performs the following processing steps: The zoning results, historical blasting parameters, effect evaluation data, and engineering environmental parameters are used as inputs. Based on the feature tensor and partition adjacency matrix in the partitioning results, the spatial topological relationship of each partition is obtained through the partition feature extraction branch. Based on historical blasting parameters and effect evaluation data, the long-term dependencies of blasting parameters are obtained through historical data coding branches; Based on engineering environmental parameters, the corresponding engineering constraint codes are obtained through environmental parameter coding branches; Based on the obtained spatial topological relationships, long-term dependencies of blasting parameters, and engineering constraint encoding, combined with an attention mechanism, cross-modal feature fusion is achieved to obtain fused features; Based on the obtained fusion features, the optimized blasting parameters for different partitions are obtained through a preset prediction head.

[0013] An electronic device includes a memory, a processor, and a computer program stored in the memory and running on the memory. When the processor executes the program, it implements the method for optimizing the parameters of blast holes in the surrounding rock of a tunnel face.

[0014] A non-transitory computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for optimizing blast hole parameters in zoned blasting of surrounding rock at a tunnel face.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: (1) This application provides a method and system for optimizing the parameters of blasting boreholes in the surrounding rock of a tunnel face. The scheme integrates the surface morphology features, joint density features, resistivity features and rebound strength features of the tunnel face to achieve comprehensive perception of the surface and internal features of the tunnel face. This eliminates the problem of inaccurate partitioning caused by relying on experience or single-dimensional data in traditional schemes. While improving the accuracy of partitioning, it also effectively improves the accuracy of parameter optimization. Furthermore, the scheme of this application introduces historical blasting data in both the partitioning process and the parameter optimization process. The historical blasting data effectively guides the model and further improves the accuracy of parameter optimization. At the same time, by feeding back the blasting effect to the parameter optimization model, the loss function of the parameter optimization model is dynamically adjusted. Combined with incremental training technology, an adaptive optimization closed loop for blasting parameter optimization is formed, which further improves the adaptability of the blasting parameter optimization model.

[0016] (2) This application provides a comprehensive blasting effect evaluation scheme based on multidimensional data. While introducing multidimensional data to ensure the accuracy of blasting effect evaluation, it introduces geological difficulty coefficient to dynamically adjust the expected benchmark of effect evaluation results, which effectively improves the adaptability of the blasting borehole parameter optimization method to different regions.

[0017] (3) This application sets a triggering mechanism for the dynamic adjustment of the loss function of the adaptive optimization model for blasting borehole parameters. Through this triggering mechanism, the model only performs incremental training under necessary conditions, rather than performing incremental training frequently, which saves time costs to a certain extent. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below: Figure 1 This is a basic flowchart of a method for optimizing the parameters of blasting boreholes in the surrounding rock of a tunnel face, as described in Example 1. Figure 2 This is a schematic diagram of the overall data flow framework for the method of optimizing blast hole parameters in zoned blasting of surrounding rock at the tunnel face as described in Example 1. Figure 3 This is a schematic diagram of the preset effect evaluation strategy described in Example 1; Figure 4 This is a schematic diagram of the data processing flow of the deep learning-based adaptive optimization model for borehole parameters described in Example 1. Figure 5 This is a schematic diagram of the network structure of the adaptive optimization model for borehole parameters described in Example 1; Figure 6 This is a schematic diagram of the tunnel face surrounding rock zone blasting borehole parameter optimization system described in Example 1. Detailed Implementation

[0019] To better understand the above-mentioned objectives, features and advantages of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0020] Numerous specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways than those described herein, and therefore the invention is not limited to the specific embodiments disclosed in the following specification.

[0021] Example 1: The following describes in detail, with reference to the accompanying drawings, a method for optimizing the parameters of blast holes in the surrounding rock of a tunnel face in a zoned blasting configuration, as described in Embodiment 1.

[0022] like Figure 1 As shown, a method for optimizing the parameters of blasting boreholes in zoned blasting of surrounding rock at a tunnel face includes: The tunnel face is subjected to multi-source features, including surface morphology features, joint density features, resistivity features, rebound strength features, and historical blasting features. Based on the obtained multi-source features, the current working face is partitioned using a pre-built deep learning-based partitioning model to obtain the partitioning results; Based on the obtained partitioning results, historical blasting parameters, effect evaluation data, and engineering environment parameters as input, the optimized blasting borehole parameters are obtained through a pre-built deep learning-based adaptive optimization model for borehole parameters. Specifically, after blasting using the blasting borehole parameters, on-site data is acquired and a comprehensive effect evaluation result is obtained through a preset effect evaluation strategy. Based on the obtained comprehensive effect evaluation result, the loss function of the borehole parameter adaptive optimization model is dynamically adjusted in conjunction with a preset triggering mechanism, and incremental training of the borehole parameter adaptive optimization model is performed.

[0023] In specific implementation, such as Figure 2 As shown, the surface morphology features are obtained based on three-dimensional laser scanning; the joint density features are obtained based on digital photogrammetry combined with point cloud analysis; the resistivity features are obtained through high-density electrical resistivity tomography; the rebound strength features are obtained through digital rebound hammer testing; and the historical blasting features are obtained by using a pre-established historical rock mass blastability database to obtain blasting effect evaluation data. This database integrates previous blasting parameters and blasting effect evaluation data, and is updated in real time after each blasting effect data is obtained.

[0024] In specific implementation, such as Figure 3 As shown, the effect evaluation includes the following process: The following data were acquired: three-dimensional point cloud data of the working face before and after blasting, multi-view image data of the blast pile, sound wave velocity test data before and after blasting, vibration monitoring time history data, and high-speed photography recording data. Based on the obtained data, three-dimensional forming quality assessment, blast block distribution assessment, blast damage assessment, and blast dynamic response assessment of the working face after blasting were performed respectively, and the assessment results were obtained. Based on the obtained evaluation results, a comprehensive effect evaluation result is obtained through fusion processing.

[0025] The following is a detailed description of the effect evaluation process: First, define the input dataset. ,in: : 3D point cloud data of the tunnel face before blasting, containing a set of coordinates (x, y, z); : 3D point cloud data of the tunnel face after blasting, containing a set of coordinates (x, y, z); Burst-heap multi-view image data, consisting of a sequence of several images; : Test data of sound wave velocity before blasting (m / s); : Test data of sound wave velocity after blasting (m / s); Vibration monitoring time history data, including velocity records; it is mainly used for vibration energy calculation and vibration control effectiveness analysis. High-speed photography records data, which is mainly used for the analysis of the blasting process.

[0026] Secondly, based on the obtained data, we performed three-dimensional forming quality assessments of the tunnel face after blasting, assessments of blast block distribution, blasting damage assessments, and blasting dynamic response assessments, obtaining the assessment results. It should be noted that, to address tunnel construction in different regions, we introduced a geological difficulty coefficient. : in, For actual rock strength, As a reference rock strength, This represents the actual joint density. As the reference joint density, The rock mass integrity coefficient. The baseline rock mass integrity coefficient, These are the weighting coefficients for each dimension.

[0027] In one or more implementations, the weight coefficients for each dimension can be set as follows: It is understandable that the weighting coefficients here can be adjusted adaptively according to actual needs.

[0028] Simultaneously, based on the geological difficulty coefficient, dynamic adjustments are made to the over-excavation benchmark and the half-hole ratio benchmark to obtain the dynamic over-excavation benchmark and the dynamic half-hole ratio benchmark, wherein: Dynamic over-excavation benchmark: ; Dynamic semi-porosity benchmark: .

[0029] In practice, the evaluation methods for each dimension are as follows: (1) Three-dimensional forming quality assessment, including over-excavation rate calculation based on dynamic benchmark and profile quality spectrum analysis, wherein the over-excavation rate calculation based on dynamic benchmark specifically includes: ; ; in, To standardize the over-excavation rate, This represents the actual over-excavation volume. As a dynamic over-excavation benchmark, The over-excavation control quality index has a value range of [0, 1]. It is a function for maximizing the value.

[0030] The profile mass spectrum analysis specifically includes: in, The profile quality index. The root mean square roughness of the profile surface. To allow for roughness.

[0031] The above data can be obtained from the point cloud data before and after the blasting. and The results were obtained through standard formula calculations and by consulting relevant standard reference books. Since this is not the focus of this solution and can be obtained using existing formulas, it will not be described in detail here.

[0032] (2) Evaluation of the distribution of blasted blocks, which mainly considers the block size and quality, specifically: in, This is the block size quality index, with values ​​ranging from [0, 1]. For block size variables, For the maximum block size, Minimum block size, This is the actual block size distribution density function. It is the ideal block size distribution density function.

[0033] The physical meaning of the above formula is explained below: when When the actual block size distribution is exactly the same as the ideal distribution, the blasting effect is optimal. when When the actual block size distribution is completely orthogonal to the ideal distribution, the blasting effect is the worst. when The closer the value is to 1, the more the block size distribution produced by the blasting meets the engineering requirements.

[0034] The aforementioned block size distribution function and other data can be based on the multi-view image data of the exploded pile obtained earlier. This was obtained through existing image analysis, and will not be elaborated further here.

[0035] (3) Blasting damage assessment, mainly including assessment of the preservation quality of half-hole marks based on dynamic benchmarks and assessment of the depth of damage to the surrounding rock; wherein, the assessment of the preservation quality of half-hole marks based on dynamic benchmarks specifically includes: in, The quality index of the half-hole mark. This represents the actual half-well retention rate. As a dynamic half-porosity benchmark, This represents the average borehole angle deviation (in degrees). The standard deviation of the half-hole length, This represents the average length of the half-hole.

[0036] The assessment of surrounding rock damage depth is mainly achieved by averaging the damage depths from multiple measuring points to obtain the average damage depth. .

[0037] The data required above is mainly based on the acoustic wave velocity test data obtained before and after the blasting, and is estimated based on the traditional damage depth estimation algorithm, which will not be elaborated here.

[0038] (4) Evaluation of blasting dynamic response, specifically: Based on vibration monitoring time history data, vibration energy-related data are obtained; wherein, the vibration energy-related data includes, but is not limited to, rock mass density, rock mass wave velocity, equivalent area of ​​each monitoring point, and vibration velocity time history of the monitoring point; Based on the obtained vibration energy data, a vibration control index is determined, wherein the vibration control index is determined using the following calculation method: in, The vibration control index, Let be the peak vibration velocity (unit: m / s) at the j-th monitoring point. To allow peak vibration velocity (m / s). Let J be the vibration energy at the j-th monitoring point (unit: J). The allowable vibration energy (unit: J) is given, and m represents the number of monitoring points.

[0039] Finally, based on the obtained evaluation results, a comprehensive effect evaluation result is obtained through fusion processing. The calculation of the comprehensive effect evaluation result specifically adopts the following processing procedure: in, For the comprehensive effect evaluation results, The profile quality index. Geological difficulty coefficient, The quality index for over-excavation control. This is the block size quality index. The quality index of the half-hole mark. The average damage depth, The vibration control index, These are the weighting coefficients for different evaluation dimensions.

[0040] Furthermore, based on the comprehensive effect evaluation results, the blasting effect was classified into different levels, as shown in Table 1: Table 1 Classification of Explosive Effect Levels Based on the grading in Table 1, when the grading result indicates that improvement is needed, feedback is performed on the adaptive optimization model for borehole parameters. The feedback data includes seven indicators: contour quality index, geological difficulty coefficient, over-excavation control quality index, block size quality index, half-hole mark quality index, average damage depth, and vibration control index. At the same time, incremental training of the model is triggered. It can be understood that the scheme described in this embodiment does not strictly limit the specific threshold range of the feedback, and the grading threshold range of each level can be controlled according to actual needs.

[0041] In specific implementation, the deep learning-based partitioning model adopts the U-Net network, which is widely used in face partitioning. During the face partitioning process, the U-Net network has five input channels, of which surface morphology features, joint density features, resistivity features, rebound strength features, and historical blasting features each correspond to one input channel. It should be noted that the input features of the first four channels characterize the static geological properties of the current face (including surface and internal features), while the fifth (historical blasting features) introduces the subsequent blasting results into the partitioning model, realizing the dual driving force of geological conditions and engineering feedback, and realizing the dynamic guidance of engineering results on the partitioning model.

[0042] Secondly, the U-Net network obtains an output tensor by encoding and decoding the input data of each channel. Based on the obtained output tensor, a high-quality partitioning result can be obtained by refining the partition boundaries using a conditional random field. The partitioning result includes a feature tensor and a partition adjacency matrix.

[0043] In specific implementation, such as Figure 4 and Figure 5 As shown, the deep learning-based adaptive optimization model for borehole parameters specifically performs the following processing steps: The zoning results, historical blasting parameters, effect evaluation data, and engineering environmental parameters are used as inputs; wherein, the engineering environmental parameters include, but are not limited to, tunnel span, burial depth, support type, excavation direction, and hydrological conditions. Based on the feature tensor and partition adjacency matrix in the partitioning results, the spatial topological relationship of each partition is obtained through the partitioning feature extraction branch (using graph convolutional neural network); Based on historical blasting parameters and effect evaluation data, the long-term dependencies of blasting parameters are obtained through historical data encoding branches (using a bidirectional long short-term memory (LSTM) network). Based on engineering environmental parameters, the corresponding engineering constraint codes are obtained through an environmental parameter coding branch (using a multilayer perceptron (MLP)). Based on the obtained spatial topological relationships, long-term dependencies of blasting parameters, and engineering constraint encoding, combined with an attention mechanism, cross-modal feature fusion is achieved to obtain fused features; Based on the obtained fusion features, the optimized blasting parameters for different partitions are obtained through a preset prediction head.

[0044] In one or more embodiments, the scheme autonomously designs the loss function of the above-mentioned adaptive optimization model for borehole parameters, as specifically expressed below: in, Predict the loss for the parameters. Due to the loss of geological consistency, For engineering constraint losses, To provide feedback on the losses, The parameters can be adjusted according to the actual situation.

[0045] Specifically, the parameter predicts the loss. The determination is mainly based on the mean square error between the predicted and target values ​​of the blasting parameters; The loss of geological consistency Based on the rationality of the predicted values ​​of blasting parameters and geological features, correlation loss (such as negative Pearson correlation coefficient) and gradient-based loss (such as Frobenius norm of Jacobi matrix) are usually used. In this embodiment, negative Pearson correlation coefficient is mainly used to construct geological consistency loss. The engineering constraint loss It is mainly used to ensure that the predicted values ​​of blasting parameters meet basic engineering constraints, and the mean square error is used for determination here; The feedback loss It is mainly used to adjust the model based on blasting effect data, so that the predicted parameters can effectively improve the blasting effect. Since the first three losses are commonly used loss functions in this field, the innovation of the scheme described in this embodiment lies in the feedback loss. Therefore, the settings for the first three losses will not be discussed in detail here. Specifically, the feedback loss... The following design is adopted: in, To provide feedback on losses, M represents the number of performance indicators used in the comprehensive evaluation of blasting effectiveness, namely, seven indicators: profile quality index, geological difficulty coefficient, over-excavation control quality index, block size quality index, half-hole mark quality index, average damage depth, and vibration control index. Here, M=7. To indicate the degree of importance of the indicators, The expected indicator value, These are the actual evaluation indicator values.

[0046] In one or more embodiments, the above loss function Adjustment parameters The settings are dynamically adjusted based on the comprehensive effect evaluation results, specifically in the following manner: in, In the current model loss function initial value, For the comprehensive effect evaluation results, To adjust the parameters.

[0047] It should be noted that the parameters are adjusted here. The dynamic adjustment has a certain triggering mechanism, if and only if When this dynamic adjustment is triggered, the borehole parameter adaptive optimization model will perform incremental training of the model based on the new loss function and combined with the current unqualified blasting data.

[0048] Example 2 Based on the solution described in Embodiment 1 above, as follows Figure 6 As shown, this embodiment provides a system for optimizing the parameters of blasting boreholes in the surrounding rock zone of a tunnel face, including: The data acquisition module is used to acquire multi-source features of the tunnel face, including surface morphology features, joint density features, resistivity features, rebound strength features, and historical blasting features. The partitioning module is used to partition the current face of the tunnel based on the obtained multi-source features using a pre-built deep learning-based partitioning model, and obtain the partitioning results. The parameter optimization unit is used to obtain optimized blasting borehole parameters based on the obtained partitioning results, historical blasting parameters, effect evaluation data, and engineering environment parameters, through a pre-built deep learning-based adaptive optimization model for borehole parameters. Specifically, after blasting using the blasting borehole parameters, on-site data is acquired and a comprehensive effect evaluation result is obtained through a preset effect evaluation strategy. Based on the obtained comprehensive effect evaluation result, the loss function of the borehole parameter adaptive optimization model is dynamically adjusted in conjunction with a preset triggering mechanism, and incremental training of the borehole parameter adaptive optimization model is performed.

[0049] It should be noted that the system described in this embodiment corresponds to the method described in Embodiment 1, and its technical details have been described in detail in Embodiment 1, so they will not be repeated here.

[0050] Example 3: This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on it. When the processor executes the program, it implements the method for optimizing the parameters of blasting boreholes in the surrounding rock of a tunnel face as described in Embodiment 1.

[0051] Example 4: This embodiment provides a non-transitory computer-readable storage medium storing a computer program that, when executed by a processor, implements the method for optimizing blast hole parameters in zoned blasting of surrounding rock at a tunnel face as described in Embodiment 1.

[0052] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications or equivalent changes made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for optimizing the parameters of blasting boreholes in zoned blasting of surrounding rock at a tunnel face, characterized in that, include: The tunnel face is subjected to multi-source features, including surface morphology features, joint density features, resistivity features, rebound strength features, and historical blasting features. Based on the obtained multi-source features, the current working face is partitioned using a pre-built deep learning-based partitioning model to obtain the partitioning results; Based on the obtained partitioning results, historical blasting parameters, effect evaluation data, and engineering environment parameters as input, the optimized blasting borehole parameters are obtained through a pre-built deep learning-based adaptive optimization model for borehole parameters. Specifically, after blasting using the blasting borehole parameters, on-site data is acquired and a comprehensive effect evaluation result is obtained through a preset effect evaluation strategy. Based on the obtained comprehensive effect evaluation result, the loss function of the borehole parameter adaptive optimization model is dynamically adjusted in conjunction with a preset triggering mechanism, and incremental training of the borehole parameter adaptive optimization model is performed. The deep learning-based adaptive optimization model for borehole parameters specifically performs the following processing steps: The zoning results, historical blasting parameters, effect evaluation data, and engineering environmental parameters are used as inputs. Based on the feature tensor and partition adjacency matrix in the partitioning results, the spatial topological relationship of each partition is obtained through the partition feature extraction branch. Based on historical blasting parameters and effect evaluation data, the long-term dependencies of blasting parameters are obtained through historical data coding branches; Based on engineering environmental parameters, the corresponding engineering constraint codes are obtained through environmental parameter coding branches; Based on the obtained spatial topological relationships, long-term dependencies of blasting parameters, and engineering constraint encoding, combined with an attention mechanism, cross-modal feature fusion is achieved to obtain fused features; Based on the obtained fusion features, the optimized blasting parameters for different partitions are obtained through a preset prediction head.

2. The method for optimizing blast hole parameters in zoned blasting of surrounding rock at a tunnel face as described in claim 1, characterized in that, The preset effect evaluation strategy includes the following processing steps: The following data were acquired: three-dimensional point cloud data of the working face before and after blasting, multi-view image data of the blast pile, sound wave velocity test data before and after blasting, vibration monitoring time history data, and high-speed photography recording data. Based on the obtained data, three-dimensional forming quality assessment, blast block distribution assessment, blast damage assessment, and blast dynamic response assessment of the working face after blasting were performed respectively, and the assessment results were obtained. Based on the obtained evaluation results, a comprehensive effect evaluation result is obtained through fusion processing.

3. The method for optimizing blasting hole parameters in zoned blasting of surrounding rock at a tunnel face as described in claim 2, characterized in that, A geological difficulty coefficient is introduced into the effectiveness evaluation, and the expected benchmark of the effectiveness evaluation results is dynamically adjusted based on the geological difficulty coefficient. The geological difficulty coefficient is expressed as follows: in, For actual rock strength, As a reference rock strength, This represents the actual joint density. As the reference joint density, The rock mass integrity coefficient. The baseline rock mass integrity coefficient, These are the weighting coefficients for each dimension.

4. The method for optimizing blasting hole parameters in zoned blasting of surrounding rock at a tunnel face as described in claim 1, characterized in that, The calculation of the overall effect evaluation result adopts the following processing procedure: in, For the comprehensive effect evaluation results, The profile quality index. Geological difficulty coefficient, The quality index for over-excavation control. This is the block size quality index. The quality index of the half-hole mark. The average damage depth, The vibration control index, These are the weighting coefficients for different evaluation dimensions.

5. The method for optimizing blasting hole parameters in zoned blasting of surrounding rock at a tunnel face as described in claim 1, characterized in that, The loss function of the adaptive optimization model for borehole parameters is specifically expressed as follows: in, Predict the loss for the parameters. Due to the loss of geological consistency, For engineering constraint losses, To provide feedback on the losses, To adjust the parameters.

6. The method for optimizing blasting hole parameters in zoned blasting of surrounding rock at a tunnel face as described in claim 5, characterized in that, The adjustment parameter of the feedback loss in the loss function is dynamically adjusted in the following manner: in, These are the initial values ​​for the feedback loss adjustment parameters in the current model's loss function. For the comprehensive effect evaluation results, To adjust the parameters.

7. A system for optimizing blasting borehole parameters in zoned blasting of surrounding rock at a tunnel face, characterized in that, include: The data acquisition module is used to acquire multi-source features of the tunnel face, including surface morphology features, joint density features, resistivity features, rebound strength features, and historical blasting features. The partitioning module is used to partition the current face of the tunnel based on the obtained multi-source features using a pre-built deep learning-based partitioning model, and obtain the partitioning results. The parameter optimization unit is used to obtain optimized blasting borehole parameters based on the obtained partitioning results, historical blasting parameters, effect evaluation data, and engineering environment parameters, through a pre-built deep learning-based adaptive optimization model for borehole parameters. Specifically, after blasting using the blasting borehole parameters, on-site data is acquired and a comprehensive effect evaluation result is obtained through a preset effect evaluation strategy. Based on the obtained comprehensive effect evaluation result, the loss function of the borehole parameter adaptive optimization model is dynamically adjusted in conjunction with a preset triggering mechanism, and incremental training of the borehole parameter adaptive optimization model is performed. The deep learning-based adaptive optimization model for borehole parameters specifically performs the following processing steps: The zoning results, historical blasting parameters, effect evaluation data, and engineering environmental parameters are used as inputs. Based on the feature tensor and partition adjacency matrix in the partitioning results, the spatial topological relationship of each partition is obtained through the partition feature extraction branch. Based on historical blasting parameters and effect evaluation data, the long-term dependencies of blasting parameters are obtained through historical data coding branches; Based on engineering environmental parameters, the corresponding engineering constraint codes are obtained through environmental parameter coding branches; Based on the obtained spatial topological relationships, long-term dependencies of blasting parameters, and engineering constraint encoding, combined with an attention mechanism, cross-modal feature fusion is achieved to obtain fused features; Based on the obtained fusion features, the optimized blasting parameters for different partitions are obtained through a preset prediction head.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running thereon, characterized in that, When the processor executes the program, it implements the method for optimizing the parameters of blasting boreholes in the surrounding rock zone of a tunnel face as described in any one of claims 1-6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the method for optimizing the parameters of blasting boreholes in the surrounding rock zone of a tunnel face as described in any one of claims 1-6.

Images