Complex stratum tunnel blasting parameter intelligent design optimization method

Abstract

The invention discloses a complex stratum tunnel blasting parameter intelligent design optimization method, and particularly relates to the technical field of tunnel blasting, and the method comprises the following steps: obtaining initial full-section blast hole design, detonation network design and explosive load design information, and constructing an initial mathematical representation data set; based on the tunnel sampling data, establishing a tunnel three-dimensional model, and inputting the initial data into a trained blasting simulation model to obtain a preliminary simulation result; performing deviation analysis by comparing a simulation result with an expected result, and judging whether a blasting parameter optimization mechanism is started or not; if so, carrying out sensitivity analysis and cycle analysis to confirm all sensitive parameters. Utilizing a genetic algorithm to optimize blasting parameters, replacing corresponding parameters in the initial data set, and finally obtaining an optimized mathematical representation data set; according to the method, blasting parameters can be accurately adjusted, unnecessary blasting material waste is reduced, and secondary correction work caused by excessive blasting or insufficient blasting can be avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel blasting, and more specifically, to an intelligent design and optimization method for tunnel blasting parameters in complex strata. Background Art

[0002] Under the requirements of the "dual carbon" strategy, the requirements for the mechanization, intelligence, and greening levels of tunnel construction are constantly increasing. Limited by technical levels and control means, when the tunnel body passes through complex strata conditions such as developed joint fissures, fault fracture zones, and high ground stress, large overbreak often occurs during tunnel blasting excavation, and even collapse problems may occur.

[0003] Tunnel controlled blasting is a controlled blasting technology that rationally utilizes explosive energy. The tunnels excavated by this method have regular shapes, no obvious blasting cracks on the rock walls, can maintain the integrity of the surrounding rock, improve its stability and self-bearing capacity, and reduce the amount of overbreak and underbreak in general blasting construction. Controlled blasting technology originated in Sweden in the early 1960s, and the blasting sequence is to carry out cut blasting first and then perimeter blasting.

[0004] In the prior art, when designing tunnel blasting, first consider the full-section size of the tunnel and the drilling conditions of the mechanical jumbo, and propose a parametric design method for the full-section blast holes of the tunnel based on the blasting design principle (first cut holes, then perimeter holes, and finally auxiliary holes); then, according to the rock mass explosibility grading method, propose a differential design method for the full-section blast hole design, and finally complete the refined design of the full-section blast holes; secondly, from the theoretical perspectives of the specific charge of the rock mass and the broken rock burden volume, propose a calculation method for the charge per single blast hole, and at the same time realize the automatic statistics of the charge of the blast holes; finally, based on the initiation rule from the inside to the outside of the full-section blast holes, propose an automatic initiation network configuration method for the full-section blast holes. However, before actual application, it is still necessary to explore the adaptability of the tunnel to be blasted, and when necessary, seek an optimized parameter design that is more suitable for the tunnel to be blasted. Therefore, the present invention proposes an intelligent design and optimization method for tunnel blasting parameters in complex strata, in order to help blasting personnel better achieve the established blasting goals. Summary of the Invention

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An intelligent design and optimization method for tunnel blasting parameters in complex strata, comprising the following steps:

[0007] Obtain the information of the initial full-section blast hole design, initiation network design, and charge design, and then organize them into an initial mathematical representation data set;

[0008] Construct a 3D tunnel model based on tunnel sampling information, and substitute the initial mathematical representation dataset into a pre-trained tunnel blasting simulation model for simulation to obtain the initial tunnel blasting simulation results;

[0009] Based on the initial tunnel blasting simulation results and the expected tunnel blasting simulation results, perform deviation analysis to determine whether to activate the blasting parameter optimization mechanism;

[0010] After the blasting parameter optimization mechanism is activated, first perform sensitivity analysis to find the first-order sensitive parameters, and then perform cyclic analysis until all sensitive parameters are found and summarized into a sensitive parameter group;

[0011] Based on the sensitive parameter group, use the genetic algorithm to find the optimized parameter design data, and replace the corresponding initial parameter design data in the initial mathematical representation dataset with the optimized parameter design data to obtain the optimized mathematical representation dataset.

[0012] In a preferred embodiment, the deviation analysis refers to:

[0013] Divide the 3D tunnel model into multiple 3D regions, and respectively obtain the blasting instantaneous pressure data, displacement data, and broken area range data of each 3D region, as well as the blasting instantaneous pressure data, displacement data, and broken area range data in the expected tunnel blasting simulation results of each region. Calculate the difference of the corresponding data items, and perform absolute value processing on the results of the difference calculation to obtain three types of absolute values corresponding to the blasting instantaneous pressure, displacement, and broken area range. Perform weighted summation on the three types of absolute values to obtain the comprehensive deviation value.

[0014] In a preferred embodiment, determining whether to activate the blasting parameter optimization mechanism refers to:

[0015] Extract the maximum value among all the comprehensive deviation values, and then compare it with a preset activation threshold. If the maximum value among all the comprehensive deviation values is greater than the preset activation threshold, generate an abnormal signal. If the maximum value among all the comprehensive deviation values is less than or equal to the preset activation threshold, generate a normal signal. When the abnormal signal is generated, activate the blasting parameter optimization mechanism.

[0016] In a preferred embodiment, performing sensitivity analysis refers to:

[0017] Use single-factor analysis to obtain the influence degree of each individual parameter factor in the simulation model on the simulation results, and then compare it with a preset first-gradient sensitivity threshold. Retain all the individual parameter factors greater than the preset first-gradient sensitivity threshold and use them as the first-order sensitive parameters.

[0018] In a preferred embodiment, the Sobol method or the Morris method is used for cycle analysis.

[0019] In a preferred embodiment, the judgment logic for using the Sobol method or the Morris method is as follows: obtain the total number of first-order sensitive parameters and the acceptance time limit of cycle analysis, then substitute the total number of first-order sensitive parameters and the acceptance time limit of cycle analysis together as input variables, and use the type of method used in cycle analysis as the output variable. Fuzzify the input variables, convert the values of the input variables into fuzzy sets, fuzzify the output variable, convert the output variable into a fuzzy set, formulate fuzzy rules to describe the adaptability of the two methods under different combinations of data types, and infer the type of method used in cycle analysis by reasoning the fuzzified input variables through the fuzzy rules.

[0020] In a preferred embodiment, during cycle analysis, all the next-order sensitive parameters are searched for in each cycle. There is a preset interaction effect threshold corresponding to each order. In each cycle, the parameter corresponding to the maximum interaction effect of the previous-order sensitive parameters is obtained, and then the parameter corresponding to the maximum interaction effect of the previous-order sensitive parameters is compared with the interaction effect threshold corresponding to the next order. If the parameter corresponding to the maximum interaction effect of the previous-order sensitive parameters is greater than the interaction effect threshold corresponding to the next order, the parameter corresponding to the maximum interaction effect is determined as the next-order sensitive parameter. Repeat this cycle until all sensitive parameters are found and summarized into a sensitive parameter group.

[0021] In a preferred embodiment, when using the genetic algorithm to find optimized parameter design data, the fitness evaluation formula is used to obtain the reciprocal of the sum of all comprehensive deviation values as the fitness evaluation value, and the maximum value among all comprehensive deviation values satisfies being less than or equal to a preset activation threshold. The finally optimized parameter data combination is the optimized parameter design data.

[0022] The technical effects and advantages of the present invention:

[0023] The intelligent design in the present invention ensures a more precise blasting process by automatically adjusting and selecting appropriate blasting parameters (such as charge amount, initiation time, etc.), thereby optimizing the blasting force and range and reducing the impact caused by excessive or insufficient blasting. During tunnel construction, due to complex geological conditions and uncertain factors, there are relatively high safety risks in blasting operations. The intelligent design method utilizes simulation technology to adjust the blasting design, effectively avoiding accident risks and ensuring the safety of the construction process. Traditional blasting designs often rely on manual experience and repeated debugging, resulting in waste of resources and increased time costs. The intelligent design optimization method can precisely adjust blasting parameters, reducing unnecessary waste of blasting materials (such as explosives, detonating tubes, etc.). At the same time, the intelligent system can also optimize equipment and labor input, avoiding secondary correction work caused by excessive or insufficient blasting, further reducing labor and equipment costs. By improving construction efficiency, the intelligent design can also effectively shorten the construction period and reduce the overall cost of the project.

[0024] Blasting construction will generate environmental problems such as vibrations and flying rocks, which may all have an impact on the ecological environment. The intelligent design can, through precise control of blasting parameters, adjust parameters such as the charge amount in the present invention to avoid excessive vibrations. Through precise design and control, the intelligent design optimization of blasting parameters can not only ensure the project quality but also meet the requirements of green construction, reducing interference to the surrounding environment and improving the acceptability of construction personnel and the environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] For the convenience of those skilled in the art to understand, the present invention will be further described below in conjunction with the accompanying drawings;

[0026] Figure 1 It is the schematic diagram of the intelligent design optimization method for blasting parameters of tunnels in complex strata in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present invention.

[0028] Refer to Figure 1 The following embodiments are obtained:

[0029] Embodiment 1: The intelligent design optimization of blasting parameters for tunnels in complex strata refers to precisely optimizing various parameters (such as explosive amount, blasting method, blasting time, etc.) of tunnel blasting design through intelligent technology under complex geological conditions to achieve efficient, safe, and economical tunnel construction.

[0030] Challenges of Tunnels in Complex Strata: In tunnel engineering in complex strata, the geological conditions vary greatly, and factors such as groundwater level, rock type, and fault zones can all affect the blasting effect. The traditional design of blasting parameters relies on the experience of engineers, which is prone to errors and inaccuracies, leading to an increase in construction risks or delays in construction progress. To overcome these problems, it is particularly important to adopt an intelligent design method.

[0031] Advantages of Intelligent Design: Through in-depth analysis of geological exploration data and the combination of artificial intelligence and big data technologies, intelligent design optimization can accurately calculate the optimal values of each blasting parameter. By training historical blasting data with machine learning algorithms, it can predict and optimize blasting parameters, thus avoiding empirical errors in traditional design methods, intelligently optimizing and adjusting parameters, reducing construction costs, and enhancing safety.

[0032] Blasting parameters refer to the key parameters that need to be set during tunnel blasting, such as explosive type, explosive quantity, blasting time interval, charging depth, blasting method, etc. These parameters are crucial for the blasting effect and the smooth progress of tunnel construction.

[0033] The intelligent design optimization method for blasting parameters in tunnels in complex strata includes the following steps:

[0034] Obtain the information of the initial full-face blast hole design, initiation network design, and charge quantity design, and then organize it into an initial mathematical representation data set; this process is the starting point of the entire optimization process. By collecting and organizing the design information of tunnel blasting (such as blast hole layout, initiation sequence, and charge quantity) and converting it into mathematical data representation, the initial data set for the optimization process is constructed. These data will be used for subsequent simulation and optimization calculations. Ensure that there is a clear starting point for the optimization process and provide a standardized data format for subsequent calculations.

[0035] Construct a three-dimensional tunnel model based on the tunnel sampling information, and substitute the initial mathematical representation data set into the pre-trained tunnel blasting simulation model for simulation to obtain the initial tunnel blasting simulation results; By constructing a three-dimensional tunnel model through tunnel sampling information, the actual geological and physical environment of the tunnel can be more realistically simulated. Then substitute the initial design data into the trained blasting simulation model for blasting simulation to obtain the preliminary blasting effect (such as blasting range, etc.). The simulation results provide a basis for subsequent analysis and optimization and are the key to judging whether the blasting effect meets expectations.

[0036] Based on the deviation analysis of the initial tunnel blasting simulation results and the expected tunnel blasting simulation results, judge whether to activate the blasting parameter optimization mechanism; evaluate the effectiveness of the current design by comparing the simulation results and the expected results. If there is a large deviation between the two, it indicates that the current blasting parameter design may not be ideal and needs to be further optimized. This step determines whether to enter the optimization stage and is the key to judging whether to activate the optimization mechanism.

[0037] After the blasting parameter optimization mechanism is activated, first conduct a sensitivity analysis to find the first-stage sensitive parameters, and then conduct a cyclic analysis until all sensitive parameters are found and summarized into a sensitive parameter group; the purpose of the sensitivity analysis is to find the parameters that have the greatest impact on the blasting effect. Through the sensitivity analysis of each design parameter, gradually determine which parameters have a greater impact on the change of the tunnel blasting effect, and then screen out a group of sensitive parameters. These parameters will become the key objects of optimization, helping to narrow the optimization scope, focus on the most influential parameters, improve the optimization efficiency and accuracy. The cyclic analysis is to find other parameters with a greater degree of interaction involved in the optimization process of the first-stage sensitive parameters for joint subsequent optimization, to avoid only optimizing the first-stage sensitive parameters and still having an excessive impact on other parameters with a greater degree of interaction involved. Finding all sensitive parameters and summarizing them into a sensitive parameter group can complete the overall optimization at one time.

[0038] Based on the sensitive parameter group, use the genetic algorithm to find the optimized parameter design data, and replace the corresponding initial parameter design data in the initial mathematical representation dataset with the optimized parameter design data to obtain the optimized mathematical representation dataset. The genetic algorithm is an effective global optimization algorithm that can find the optimal solution among a large number of possible parameter combinations. By optimizing the sensitive parameter group through the genetic algorithm, a new set of blasting parameters can be generated. These optimized parameters are expected to make the blasting effect closer to the expected. This step optimizes the sensitive parameters through the genetic algorithm to ensure that the blasting design better meets the actual requirements. The optimized parameters will replace the original parameters to form the final optimized design dataset and complete the design optimization.

[0039] Deviation analysis refers to: dividing the tunnel three-dimensional model into multiple three-dimensional regions, respectively obtaining the blasting instantaneous pressure data, displacement data, and broken area range data of each three-dimensional region, as well as the blasting instantaneous pressure data, displacement data, and broken area range data in the expected tunnel blasting simulation results of each region, calculating the difference of the corresponding data items, and taking the absolute value of the result of the difference calculation to obtain three types of absolute values corresponding to the blasting instantaneous pressure, displacement, and broken area range, and performing a weighted sum of the three types of absolute values to obtain the comprehensive deviation value. The specific steps are as follows:

[0040] Model Division and Data Collection: First, divide the tunnel three-dimensional model into multiple three-dimensional regions. In each region, collect the instantaneous pressure data, displacement data, and the range data of the broken area obtained during the actual blasting process.

[0041] Simulation Data Acquisition: Obtain the expected instantaneous pressure data, displacement data, and the range data of the broken area for each region through tunnel blasting simulation.

[0042] Difference Calculation: Compare the actual data with the simulation data and calculate the difference of each item of data (pressure, displacement, broken area).

[0043] Absolute Value Processing: Take the absolute value of the calculated difference to ensure that the deviation value is not affected by the sign.

[0044] Weighted Summation: According to the importance of different data items, assign different weights and perform weighted summation on the three types of absolute values (pressure, displacement, broken area) to obtain a comprehensive deviation value.

[0045] Through deviation analysis, the difference between the tunnel blasting simulation results and the expected blasting results can be quantitatively evaluated, thereby judging the accuracy of the input data of the simulation model. If the deviation is large, it can reveal the problems existing in the design parameters of the simulation model, providing a basis for the optimization and adjustment of subsequent designs. By deeply analyzing the deviations in different regions, it is possible to better understand the abnormal phenomena that may occur during the blasting process, thereby providing a more accurate reference for the actual construction process and avoiding potential safety risks. Deviation analysis can also be used as a feedback mechanism to gradually improve the quality and effect of the blasting project through continuous monitoring and correction.

[0046] Judging whether to activate the blasting parameter optimization mechanism means: Extract the maximum value among all the comprehensive deviation values, and then compare it with the preset activation threshold. If the maximum value among all the comprehensive deviation values is greater than the preset activation threshold, an abnormal signal is generated. If the maximum value among all the comprehensive deviation values is less than or equal to the preset activation threshold, a normal signal is generated. When an abnormal signal is generated, the blasting parameter optimization mechanism is activated.

[0047] For the comprehensive deviation values of each three-dimensional region during the tunnel blasting process (these values are calculated by comparing the differences between the actual data and the simulation data), extract the maximum value among them. This maximum value reflects the place with the largest deviation in each region. Compare the above maximum deviation value with a preset activation threshold. The activation threshold is usually a value determined through experience, experimental data, or design requirements, representing the maximum tolerance range of the allowable deviation.

[0048] If the maximum deviation value is greater than the activation threshold, it indicates that there is a large deviation in the blasting process, which may affect the blasting effect or safety, and the system generates an abnormal signal. If the maximum deviation value is less than or equal to the activation threshold, it means that the blasting effect is within an acceptable range, and the system generates a normal signal. When an abnormal signal is generated, the blasting parameter optimization mechanism is automatically triggered for adjustment and optimization. For example, parameters such as blasting charge, delay control, and blasthole arrangement are modified to reduce the deviation and improve the blasting accuracy and effect.

[0049] This mechanism monitors the deviation situation in the blasting process in real time and automatically determines whether to adjust the blasting parameters according to the magnitude of the deviation. This automated feedback and adjustment mechanism can detect problems in a timely manner during the blasting process and take optimization measures to avoid delays caused by human intervention. Excessive deviation may indicate problems in the blasting design or implementation process, which may lead to blasting failure, geological disasters, or structural damage. By activating the optimization mechanism when the deviation is too large, potential safety hazards can be avoided, ensuring the safety of the blasting process. By adjusting the blasting parameters, the blasting effect can be optimized, making the blasting process more accurate and reasonable, avoiding resource waste and multiple blasting situations, and improving the efficiency of tunnel excavation. By optimizing the blasting parameters, unnecessary adjustments are avoided, which can effectively reduce the additional costs caused by deviation during the blasting process, such as re-blasting and repair work. Adjusting the blasting parameters can ensure the efficiency and stability of tunnel excavation, thereby improving the quality of the entire tunnel project and enhancing the controllability of blasting operations in the face of complex geological conditions.

[0050] Performing sensitivity analysis means: using single-factor analysis to obtain the degree of influence of each individual parameter factor on the simulation result in the simulation model, and then comparing it with the preset first-gradient sensitivity threshold. All individual parameter factors greater than the preset first-gradient sensitivity threshold are retained and used as the first-tier sensitive parameters.

[0051] Single-factor analysis (also known as "single-factor sensitivity analysis") is a common method of sensitivity analysis. It analyzes the impact of the change of one input parameter on the model output while ignoring the changes of other input parameters. This can help understand how the change of a single factor affects the results of the system or model. The steps of single-factor analysis are as follows:

[0052] Define the input parameter range: Determine the possible range of each input parameter, usually setting the minimum and maximum values of the parameter according to actual data or domain knowledge.

[0053] Change a single input parameter: Keep other parameters fixed and only change the value of one input parameter, and observe the change of the model output. This method helps to identify how a certain input parameter individually affects the output result.

[0054] Record output changes: Record the output results after each parameter change. The relationship between different input values and output results can be displayed in the form of tables, graphs, etc.

[0055] Calculate sensitivity indicators: For example, the first gradient (i.e., the ratio of output change caused by parameter change) can be calculated to quantify the impact of input parameter change on the output. If the gradient is large, it indicates that the input parameter has a greater impact on the model result.

[0056] Compare with thresholds: Compare the sensitivity of each input parameter with a preset sensitivity threshold (such as the first gradient sensitivity threshold) to screen out those parameters that have a greater impact on the output result. These parameters are called "first-level sensitive parameters". By changing each input parameter one by one and observing the change of the output result, the importance of the parameter can be measured. Sensitivity analysis can help identify which parameters have a significant impact on the simulation results and make reasonable decisions or optimizations accordingly.

[0057] When performing cyclic analysis, the Sobol method or Morris method is used. The Sobol method is a global sensitivity analysis method based on variance decomposition, which is used to evaluate the main effects (the influence of a single parameter) and interaction effects (the interaction between multiple parameters on the output) of input variables. It quantifies the impact of each parameter on the result by decomposing the output variance into the contributions of each input parameter and their combinations. The core idea of the Sobol method is to decompose the variance of the model output (i.e., the variability of the output) into the independent contributions of each input variable and their interaction contributions. By calculating the "contribution share" of each input variable, it can be determined which variables have the greatest impact on the model result.

[0058] The Sobol method quantifies its impact on the output by calculating the sensitivity index (Sobol Index) of each parameter: The main effect Sobol index Si represents the contribution ratio of the input variable xi to the output variance of the model, while the interaction effect Sobol index Sij represents the contribution ratio of the interaction contribution of xi and xj to the output variance of the model. xi and xj represent the variables corresponding to the indexes i and j respectively. In cyclic analysis, if the main effect Sobol index Si is the sensitive parameter of the previous level, such as the second-level sensitive parameter, then if the input variable xj in the interaction effect Sobol index Sij corresponding to the maximum value is greater than the interaction effect threshold corresponding to the third level, then the input variable xj is the third-level sensitive parameter.

[0059] The Morris method is a local sensitivity analysis method used to evaluate the relative impact of model input parameters on the output. It calculates the contribution of each input variable to the change in the model output by "jumping" the values of the input variables and evaluates the sensitivity based on these contributions. The Morris method is particularly suitable for high-dimensional problems and has a smaller computational cost than the Sobol method. In the Morris method, it can be a combination of input variables. For example, in the combination, one is a known second-order sensitive parameter xi, and the other is another parameter. Based on the final result, the other parameters are sorted. If the interaction effect of the parameter xj ranked first is greater than the interaction effect threshold corresponding to the third order, then the parameter xj ranked first is the third-order sensitive parameter corresponding to the second-order sensitive parameter. If the interaction effect of the parameter xj ranked first is less than or equal to the interaction effect threshold corresponding to the third order, then the known second-order sensitive parameter xi has no corresponding third-order sensitive parameter.

[0060] A more detailed example is as follows: In the standard Morris method, the main effect only reflects the impact of a single variable on the output, while the interaction effect reveals the joint impact of the two. If you want to analyze the combined input variables, you can take the following steps: Select the combination of input variables: Suppose there are multiple variables. Take a known second-order sensitive parameter xi and another variable xj. If you want to evaluate the joint sensitivity of these two variables, you can consider creating the variable combination (xi, xj) and then analyze the impact of these combinations on the output result. Usually, the following strategy can be adopted: To evaluate the sensitivity of the variable combination (xi, xj), it is necessary to generate the values of the combined variables through the Latin Hypercube Sampling (LHS) method or other suitable sampling methods. For each variable combination (xi, xj), the interaction effect reflects the impact of the joint change of the two variables on the output of the simulation model. A multiple regression model of the variable combination (xi, xj) can be constructed, and then the least squares regression method can be used to estimate the coefficients of each variable and its interaction term. The coefficient of the interaction term is the degree of the impact of the joint change of the two variables on the model output. The output result of the simulation model is defined as the sum of all comprehensive deviation values.

[0061] The judgment logic using the Sobol method or the Morris method is as follows:

[0062] Obtain the total number of first - stage sensitivity parameters and the acceptance time limit for loop analysis. Then, substitute the total number of first - stage sensitivity parameters and the acceptance time limit for loop analysis together as input variables, and use the type of method adopted during loop analysis as the output variable. Perform fuzzy processing on the input variables, convert the values of the input variables into fuzzy sets, perform fuzzy processing on the output variable, and convert the output variable into a fuzzy set. Formulate fuzzy rules to describe the adaptability of the two methods under different combinations of data types. Infer the type of method adopted during loop analysis by applying the fuzzy rules to the fuzzified input variables.

[0063] Input variables: Total number of first - order sensitivity parameters: This is an indicator to measure the influence degree of model input variables on the output. Generally, the more there are, the greater the sensitivity of the model input variables to the output.

[0064] Acceptance time limit for loop analysis: This refers to the upper limit of the calculation or running time set during sensitivity analysis. A shorter time limit may require choosing a method with higher computational efficiency, while a longer time limit may allow the use of a method with a larger amount of calculation. These two input variables will be fuzzified. The purpose of fuzzification is to convert these two variables with definite numerical values into fuzzy sets, so as to handle uncertainty and fuzziness. For example, the "total number of first - order sensitivity parameters" may be converted into the following fuzzy sets: few: e.g., 0 - 10; moderate: e.g., 10 - 50; many: e.g., above 50. Similarly, the acceptance time limit for loop analysis may be fuzzified as: short: e.g., 0 - 10 hours; moderate: e.g., 10 - 50 hours; long: e.g., above 50 hours.

[0065] Definition and fuzzification of the output variable: The output variable refers to the type of method to be selected during this decision - making process, usually the Sobol method or the Morris method. We also perform fuzzification on this output variable. The possible fuzzy sets include: Sobol method and Morris method.

[0066] The fuzzification of the output variable can be classified as follows: preferred: indicating that a certain method is more applicable under given conditions; moderate: indicating that the difference between the two methods is not significant and either can be chosen arbitrarily; not applicable: indicating that the adaptability of a certain method is poor under given conditions.

[0067] Formulation of fuzzy rules: Fuzzy rules are used to describe the adaptability of the two methods under different combinations of input variables. These rules will be constructed based on experience or expert judgment. For example, the rules can be as follows:

[0068] Rule 1: If the "number of first-order sensitivity parameters" is "few" and the "acceptance time limit" is "short", the output is "preferred Morris method". Rule 2: If the "number of first-order sensitivity parameters" is "moderate" and the "acceptance time limit" is "moderate", the output is "moderately select Sobol method and Morris method". Rule 3: If the "number of first-order sensitivity parameters" is "many" and the "acceptance time limit" is "long", the output is "preferred Sobol method".

[0069] Fuzzy inference: The fuzzy inference process is based on fuzzy rules and fuzzy sets. In this process, the input variables (the fuzzified "number of first-order sensitivity parameters" and "acceptance time limit") are inferred through fuzzy rules to derive the output variable (i.e., the applicable method). The specific steps are as follows:

[0070] Input fuzzification: Map the actual numerical input to the fuzzy set through the membership function. For example, if the number of first-order sensitivity parameters is 30, its membership degree in the "moderate" fuzzy set may be 0.7, and the membership degree in the "few" set is 0.3.

[0071] Rule inference: Use fuzzy rules for inference. Through inference, we can obtain the membership degrees of various possible outputs. For example, if the input is "moderate" number of sensitivity parameters and "short" acceptance time limit, the rule may infer that the membership degree of the output method "moderately select Sobol method and Morris method" is 0.8.

[0072] Defuzzification: Convert the fuzzy output into a specific choice according to the inference result. For example, defuzzification methods such as the centroid method can be used to obtain the finally selected method.

[0073] Finally, through the result of fuzzy inference, the most applicable cycle analysis method (Sobol method or Morris method) can be determined under the given input conditions. This method not only considers the numerical values of the input variables but also takes into account uncertainty and fuzziness, making the decision-making more flexible and intelligent.

[0074] During loop analysis, all next-step sensitive parameters are searched for in each loop. It is assumed that there is an interaction effect threshold corresponding to each step. In each loop, the parameter corresponding to the maximum interaction effect of the sensitive parameters in the previous step is obtained, and then the parameter corresponding to the maximum interaction effect of the sensitive parameters in the previous step is compared with the interaction effect threshold corresponding to the next step. If the parameter corresponding to the maximum interaction effect of the sensitive parameters in the previous step is greater than the interaction effect threshold corresponding to the next step, the parameter corresponding to the maximum interaction effect is determined as the sensitive parameter of the next step. This process is repeated until all sensitive parameters are found and summarized into a sensitive parameter group. It should be noted that during loop analysis, the sensitive parameters that have already been screened out should be excluded to avoid loop pressure, and finally, they can be summarized. By gradually analyzing the sensitivity parameters of each step, the most influential and critical parameters in each stage can be identified more accurately. The "sensitive parameters" in each stage refer to those factors that have significant interaction effects in that stage. This method ensures that the identified parameters are not only based on the changes of a single factor, but also consider the interaction effects between multiple parameters, and can evaluate the impacts of various parameters more comprehensively. By comparing with the "interaction effect threshold" in each loop, it can be ensured that the screened sensitive parameters are significant to a certain extent. The introduction of the threshold helps to reduce the interference of redundant factors and only retain the sensitive parameters that have a greater impact on the system output, avoiding affecting the final analysis result due to the interference of too many irrelevant factors. Through staged iterative analysis, the selection of sensitive parameters in each stage can be made more targeted and logical. Each round of selection is based on the analysis results of the previous round. This method can accumulate and gradually filter out unimportant parameters, ensuring that the selection of sensitive parameters in each step is more refined and accurate. Finally, through multiple rounds of screening and analysis, the set of all sensitive parameters is summarized. This method can systematically identify the sensitive parameters under multiple stages or multiple interaction effects, providing an accurate basis for decision-making. It is especially applicable to complex systems with multiple interacting factors, where it is necessary to gradually analyze the interaction effects between various factors and their impacts on the system behavior. This process of gradually screening sensitive parameters can have important practical application value in the field of multi-parameter optimization.

[0075] When using a genetic algorithm to find optimized parameter design data, the fitness evaluation formula is used to obtain the reciprocal of the sum of all comprehensive deviation values as the fitness evaluation value, and it is ensured that the maximum value among all comprehensive deviation values is less than or equal to a preset activation threshold. The finally optimized parameter data combination is the optimized parameter design data. To ensure the quality of the solution, an activation threshold is usually set, that is, the maximum value among all comprehensive deviation values must be less than or equal to the preset activation threshold. This activation threshold ensures that the optimized solution does not have excessive errors, thus avoiding solutions that deviate too much from the target. For example, if the maximum deviation value exceeds the activation threshold, it means that the solution does not meet the design requirements or quality standards, so it should not be selected as the winner in the genetic algorithm. The genetic algorithm gradually optimizes the parameter design data by simulating the natural selection process. In each generation, the algorithm performs selection, crossover, and mutation operations based on the fitness of individuals to generate a new population. Individuals with higher fitness (i.e., parameter combinations with smaller comprehensive deviations) are more likely to be selected to enter the next generation. Through multiple generations of iteration, the finally obtained parameter combination can meet the requirements of the activation threshold and minimize the sum of all comprehensive deviation values, thereby obtaining the optimized parameter design data. Through the optimization process of the genetic algorithm, the parameter design can be effectively adjusted to make the final solution meet the set performance requirements. The fitness evaluation formula uses the reciprocal of all comprehensive deviation values and controls the quality of the result through the activation threshold, thereby ensuring that the optimized result is not only theoretically optimal but also feasible and meets the requirements in practical applications.

[0076] The above formulas are all dimensionless and take their numerical calculations. The formula is obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0077] It should be understood that in various embodiments of the present application, the magnitudes of the sequence numbers of the above processes do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0078] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraint conditions of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0079] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0080] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present application, and all should be covered by the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. An intelligent design and optimization method for tunnel blasting parameters in complex strata, characterized in that: The following steps are involved: Obtain the initial full-section blasthole design, detonation network design, and charge design information, and then organize them into an initial mathematical representation data set; A three-dimensional tunnel model is constructed based on tunnel sampling information, and the initial mathematical representation data set is substituted into the pre-trained tunnel blasting simulation model for simulation to obtain the initial tunnel blasting simulation results; Based on the initial tunnel blasting simulation results and the expected tunnel blasting simulation results, deviation analysis is performed to determine whether to activate the blasting parameter optimization mechanism; After the blasting parameter optimization mechanism is activated, a sensitivity analysis is first performed to find the first-step sensitive parameters, and then a cyclic analysis is performed until all sensitive parameters are found and summarized into a sensitive parameter group; Based on the sensitive parameter group, a genetic algorithm is used to find optimized parameter design data, and the optimized parameter design data is used to replace the corresponding initial parameter design data in the initial mathematical characterization data set to obtain the optimized mathematical characterization data set.

2. The intelligent design and optimization method for complex stratum tunnel blasting parameters according to claim 1 is characterized in that: Deviation analysis refers to: The tunnel three-dimensional model is divided into multiple three-dimensional areas, and the blasting instantaneous pressure data, displacement data, and crushing area range data of each three-dimensional area, as well as the blasting instantaneous pressure data, displacement data, and crushing area range data in the expected tunnel blasting simulation results of each area are obtained respectively. The corresponding data items are difference calculated, and the results of the difference calculation are processed by absolute value to obtain three types of absolute values ​​corresponding to the blasting instantaneous pressure, displacement, and crushing area range. The three types of absolute values ​​are weighted and summed to obtain a comprehensive deviation value.

3. The intelligent design and optimization method for complex stratum tunnel blasting parameters according to claim 2 is characterized in that: Determining whether to activate the blasting parameter optimization mechanism refers to: The maximum value of all comprehensive deviation values ​​is extracted and then compared with the preset activation threshold. If the maximum value of all comprehensive deviation values ​​is greater than the preset activation threshold, an abnormal signal is generated. If the maximum value of all comprehensive deviation values ​​is less than or equal to the preset activation threshold, a normal signal is generated. When an abnormal signal is generated, the blasting parameter optimization mechanism is activated.

4. The intelligent design and optimization method for complex stratum tunnel blasting parameters according to claim 3 is characterized in that: Conducting a sensitivity analysis means: Using single factor analysis, the influence of each individual parameter factor on the simulation results in the simulation model is obtained, and then compared with the preset first gradient sensitivity threshold, and all individual parameter factors greater than the preset first gradient sensitivity threshold are retained and used as the first step sensitivity parameters.

5. The intelligent design and optimization method for complex stratum tunnel blasting parameters according to claim 4 is characterized in that: The Sobol method or Morris method was used for the circular analysis.

6. The intelligent design and optimization method for complex stratum tunnel blasting parameters according to claim 5 is characterized in that: The judgment logic using the Sobol method or the Morris method is: Obtain the total number of first-step sensitive parameters and the acceptance time limit of the loop analysis, then substitute the total number of first-step sensitive parameters and the acceptance time limit of the loop analysis as input variables, take the type of method used in the loop analysis as the output variable, fuzzify the input variables, convert the values ​​of the input variables into fuzzy sets, fuzzify the output variables, convert the output variables into fuzzy sets, formulate fuzzy rules, describe the adaptability of the two methods under different data type combinations, reason with the fuzzified input variables through fuzzy rules, and determine the type of method used in the loop analysis.

7. The intelligent design and optimization method for complex stratum tunnel blasting parameters according to claim 6 is characterized in that: During the loop analysis, all the next-step sensitive parameters are searched for in each loop. The interaction effect threshold corresponding to each step is preset. In each loop, the parameter corresponding to the maximum value of the interaction effect of the previous-step sensitive parameter is obtained, and then the parameter corresponding to the maximum value of the interaction effect of the previous-step sensitive parameter is compared with the interaction effect threshold corresponding to the next step. If the parameter corresponding to the maximum value of the interaction effect of the previous-step sensitive parameter is greater than the interaction effect threshold corresponding to the next step, the parameter corresponding to the maximum value of the interaction effect is determined as the next-step sensitive parameter. This cycle is repeated until all sensitive parameters are found and summarized into a sensitive parameter group.

8. The intelligent design and optimization method for complex stratum tunnel blasting parameters according to claim 7 is characterized in that: When using a genetic algorithm to find optimized parameter design data, the fitness evaluation formula is used to obtain the inverse of the sum of all comprehensive deviation values ​​as the fitness evaluation value, and the maximum value of all comprehensive deviation values ​​is less than or equal to the preset activation threshold. The parameter data combination finally obtained is the optimized parameter design data.

Images