Method for determining advance grouting reinforcement parameters of tunnel fault fracture zone

Abstract

The invention relates to the technical field, in particular to a method for determining advance grouting reinforcement parameters of a tunnel fault fracture zone. The grouting pressure, the rate, the water cement ratio, the diffusion radius, the gelation time and the grouting amount per unit length are comprehensively covered through parameter grid structure division, the consistency of numerical values of all nodes is guaranteed through rock core strength and anchor rod tension data checking, and a combination with the error exceeding a threshold value generates a new solution through variation and crossover operation in a differential evolution algorithm. The global search capability is improved, combinations which do not conform to conditions are quickly eliminated, the global search capability can be enhanced, a low-quality solution can be quickly replaced, pheromone concentration is directly generated based on combination fitness, the path selection probability is calculated according to a fitness ratio, and a high-quality path is adaptively enhanced by utilizing an ant colony algorithm. The convergence speed is improved, the risk of falling into local optimum is reduced, iteration strengthening and dynamic search direction adjustment of a high-quality solution are achieved, and the convergence speed and result stability of the optimal solution are improved.

Description

Technical Field

[0001] This invention relates to the field of parameter optimization calculation technology, and in particular to a method for determining parameters for pre-grouting reinforcement of fault fracture zones in tunnels. Background Technology

[0002] In the field of parameter optimization computation technology, under multivariable and multi-constraint conditions, mathematical modeling and computational algorithms are used to determine the optimal combination of parameters that makes the objective function optimal, thereby improving system performance, efficiency, or stability. The core lies in constructing the functional relationship between input parameters and output effects, and using optimization algorithms to perform global or local searches of the parameter space to obtain the optimal parameter scheme that meets the preset objectives.

[0003] The purpose of a method for determining parameters for pre-grouting reinforcement in fault fracture zones of tunnels is to ensure the continuity and density of the reinforced body formation and effectively control the deformation of the surrounding rock by accurately calculating and selecting key control parameters in the grouting process in complex geological structures. This improves the overall stability of the tunnel structure and reduces the risk of geological disasters such as sudden inrush and collapse. The method also aims to formulate scientific construction parameters from a quantitative perspective, reduce empirical intervention, and improve the reliability and repeatability of the reinforcement effect.

[0004] Parameter optimization calculations, under multivariable and multi-constraint conditions, rely on mathematical modeling and a single optimization algorithm for searching. The lack of a comprehensive and discretized system for the input-output relationship leads to uneven parameter space coverage. Inappropriate combinations cannot be dynamically eliminated during global and local searches, requiring extensive empirical intervention to correct the iteration direction. This makes it difficult to guarantee the consistency and reliability of the calculation results. Furthermore, the lack of a tiered handling mechanism for errors in key parameters and the inability to autonomously adjust threshold judgments can easily lead to deviations in complex geological environments and cause uncontrolled deformation of the surrounding rock. The parameter confirmation process does not incorporate quantitative comparison with real-time on-site monitoring curves, resulting in a significant discrepancy between the final solution and actual construction feedback. The lack of multiple screening and multi-algorithm collaboration mechanisms results in insufficient repeatability and predictability of reinforcement effects, increasing the risk of geological disasters and unstable construction efficiency. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method for determining the parameters of advanced grouting reinforcement in tunnel fault fracture zones.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for determining the parameters of pre-grouting reinforcement in tunnel fault fracture zones, comprising the following steps:

[0007] S1: Grid the grouting pressure, grouting rate, water-cement ratio, diffusion radius, gelation time, and grouting volume per unit length. Verify the consistency of node values ​​based on core strength and anchor tension data, and establish a parametric grid structure.

[0008] S2: Using the node values ​​in the parameter grid structure as the initial population code, the error value is calculated by using the field fracture permeability coefficient of the fault fracture zone and the result of each parameter combination. For combinations where the error exceeds the threshold, the differential evolution algorithm is called to perform differential mutation and crossover operations to generate new combinations to replace the eliminated items, thus obtaining the parameter fitness matrix.

[0009] S3: Using the ant colony algorithm, the fitness value of each combination in the fitness matrix is ​​directly used as the initial pheromone concentration, and the path selection probability is calculated based on the ratio of each fitness value to the total. The pheromone concentration is then updated according to a fixed evaporation ratio.

[0010] S4: Based on the pheromone concentration set, extract the top N combinations and high-concentration path combinations from the fitness matrix, pair them by position and compare parameter values, call the detection porosity and rock parent material compressive strength, perform double screening on the corresponding items, retain the combinations that meet the conditions at the same time, and form a parameter candidate set.

[0011] S5: Calculate the Euclidean distance between each set of parameters in the candidate parameter set and the on-site construction monitoring curve values, and compare them with multiple thresholds. By reading the lining deformation data, fitting the grouting rebound curve, and performing minimum distance statistics, a parameter scheme is generated.

[0012] As a further aspect of the present invention, the specific steps for generating the parametric mesh structure are as follows:

[0013] Based on the grouting pressure, grouting rate, water-cement ratio, diffusion radius, gel time, and grouting volume per unit length, the grid is divided. For each parameter type, the parameter values ​​are traversed by numerical step size and cross-combined to generate a node list. The nodes are numbered and attribute fields are filled to generate a grid node parameter set.

[0014] Based on the set of grid node parameters, the rock core strength and anchor bolt tension records are extracted for each node and the numerical difference is calculated. The difference is compared with the threshold and nodes that do not match are removed to establish a parameter grid structure.

[0015] As a further aspect of the present invention, the specific steps for generating the parametric mesh structure are as follows:

[0016] Based on the node values ​​in the parameter grid structure, the grouting pressure, grouting rate, water-cement ratio, diffusion radius, gel time, and grouting volume per unit length are constructed into a six-dimensional vector, combined in order into an ordered array, node numbers are added and combination indices are marked, all node combination data are summarized, and a parameter combination code set is generated.

[0017] Based on the parameter combination encoding set, the coordinates corresponding to each combination number are extracted. The differential evolution algorithm is used to read the permeability coefficient of the fracture zone of the fault. The grouting pressure is multiplied by the permeability coefficient to obtain the predicted diffusion radius. The absolute value of the difference between the predicted value and the original value is calculated. The number and error value pairs are summarized to generate an error value list.

[0018] Based on the error value list, all combinations with error values ​​greater than a set threshold are filtered out. The structure is copied from the low-error combinations to generate new combinations to replace the eliminated combinations. The combinations are then renumbered and the complete matrix is ​​restored to complete the combination update and obtain the parameter fitness matrix.

[0019] As a further aspect of the present invention, the differential evolution algorithm first uses the constructed parameter combination encoding set as the initial population, with each combination corresponding to a six-dimensional vector node. For each parameter combination, three distinct combinations are selected from the population, and new mutated combinations are generated by superimposing the differences. Then, the mutated combinations are cross-crossed with the original combinations, and the parameters are retained or replaced according to the set cross-cross probability to obtain experimental combinations. Subsequently, the grouting pressure parameters in the experimental combinations are read and multiplied with the permeability coefficient of the fault fracture zone to obtain the corresponding predicted grouting diffusion radius. The error value is compared with the diffusion radius value of the original combination. When the error of the experimental combination is smaller than that of the original combination, the original combination is replaced. Finally, through multiple rounds of iterative updates, combinations are eliminated according to the error magnitude to generate an optimized parameter fitness matrix.

[0020] As a further aspect of the present invention, the differential evolution algorithm is based on the formula:

[0021]

[0022] in: This indicates the predicted grouting diffusion radius. This indicates the volume of slurry injected per unit time. Indicates the dynamic viscosity of the slurry. This indicates the connectivity of fractures within a fault fracture zone. The adjustment coefficient representing the fracture connectivity. This indicates the saturation level of the residual water in the fissure. The adjustment coefficient represents the residual water saturation. This represents the permeability coefficient of the fracture zone in the fault. This indicates the grouting pressure applied during grouting construction. This represents the original pore water pressure in the formation. This represents the constant value of pi.

[0023] As a further aspect of the present invention, the specific steps for generating the parametric mesh structure are as follows:

[0024] Based on the fitness matrix, the ant colony algorithm is used to extract the combination number and fitness value of each group, calculate the sum of fitness values, normalize the fitness of each group according to the ratio, establish a path number and concentration pairing table, write the initialization record, and generate the pheromone initialization set.

[0025] Based on the pheromone initialization set, the path numbers are sorted by concentration value, and the corresponding selection probability intervals for each group are accumulated. The mapping relationship between paths and intervals is constructed, the probability range of each group of paths is marked, and a path selection probability set is generated.

[0026] Based on the path selection probability set, the current path concentration is read, the evaporation value is calculated and subtracted from the concentration to generate an updated concentration value, the difference between the old and new concentrations is calculated and written into the change table, the status of each path is recorded, and a pheromone concentration set is generated.

[0027] As a further aspect of the present invention, the ant colony algorithm first reads the initial pheromone concentration of each path in the fitness matrix, normalizes all fitness values ​​by summing them to obtain the concentration of each path, then calculates the path selection probability based on the ratio of the concentration of each path to its heuristic pheromone value, uses a roulette wheel algorithm to determine the ant's path, then deducts the current pheromone concentration according to a set evaporation rate, and allocates pheromone increments according to the fitness of each path. Finally, the increments are merged into the remaining pheromone concentration, and the process is repeated until a preset number of iterations is reached and all path concentrations converge, outputting the final pheromone concentration set.

[0028] As a further aspect of the present invention, the ant colony algorithm is based on the formula:

[0029]

[0030] in: Representing a path The normalized pheromone concentration value, Representing a path The initial pheromone concentration, Representing a path Heuristic information values, Representing a path fitness value, Representing a path The connectivity stability coefficient, Representing a path The curvature influence coefficient, This indicates that pheromone concentration affects the weighting coefficient. This indicates that heuristic information influences the weighting coefficients. The adjustment weight represents the connectivity stability coefficient. The adjustment weight represents the curvature influence coefficient. This represents the total number of paths.

[0031] As a further aspect of the present invention, the specific steps for generating the parameter candidate set are as follows:

[0032] Based on the top N parameter combinations in the parameter fitness matrix and the high-scoring paths in the optimized parameter path set, elements from the two sets are extracted and paired according to their positions. Field values ​​and combination numbers are merged in parallel to form a list of combination pairs and generate a path parameter set.

[0033] Based on the path parameter set, the porosity value and the compressive strength data of the parent rock are read and compared one by one, and combinations that simultaneously meet the two conditions are selected to form a parameter candidate set.

[0034] As a further aspect of the present invention, the specific steps for generating the parameter determination scheme are as follows:

[0035] Based on each set of parameters in the candidate parameter set and the on-site construction monitoring curve data, the grouting parameter sequence and monitoring curve points are extracted, and the distance is calculated by taking the square root of the sum of the square differences. Multiple threshold filtering is applied to select combinations and generate a low difference combination set.

[0036] Based on the low difference combination set, the lining deformation record and the grouting rebound curve sampling points are read and fitted and compared. The combined error values ​​are sorted and the minimum term is selected to generate a definite parameter scheme.

[0037] Compared with the prior art, the advantages and positive effects of the present invention are:

[0038] In this invention, the grouting pressure, rate, water-cement ratio, diffusion radius, gelation time, and grouting volume per unit length are fully covered by the parameter grid structure. The consistency of the values ​​of each node is ensured by checking the rock core strength and anchor tension data, so that the parameter space presents a regular grid for easy subsequent calculation.

[0039] In this invention, combinations whose errors exceed the threshold generate new solutions through mutation and crossover operations in the differential evolution algorithm, thereby improving the global search capability and quickly eliminating combinations that do not meet the conditions. This can enhance the global search capability and quickly replace low-quality solutions, thereby improving optimization efficiency and the rationality of parameter combinations.

[0040] In this invention, pheromone concentration is directly generated based on combined fitness and path selection probability is calculated based on fitness ratio. Ant colony algorithm is used to adaptively strengthen high-quality paths, improve convergence speed and reduce the risk of getting trapped in local optima, realize iterative strengthening of high-quality solutions and dynamic search direction adjustment, and improve the convergence speed and result stability of optimal solutions.

[0041] In this invention, the high-concentration path and the top N combinations are screened by both porosity and rock compressive strength. This ensures that the retained schemes simultaneously meet the requirements of grout continuity, density and surrounding rock deformation control. The reliability of the results is checked from multiple dimensions, which improves the stability of the tunnel structure, significantly reduces the risk of geological disasters, and makes the formulation of construction parameters more scientific, repeatable and controllable. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the main steps of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0044] Example 1 See also Figure 1 This invention provides a technical solution: a method for determining parameters for pre-grouting reinforcement of fault fracture zones in tunnels, comprising the following steps:

[0045] S1: Grid the grouting pressure, grouting rate, water-cement ratio, diffusion radius, gelation time, and grouting volume per unit length. Verify the consistency of node values ​​based on core strength and anchor tension data, and establish a parametric grid structure.

[0046] S2: Using the node values ​​in the parameter grid structure as the initial population code, the error value is calculated by using the field fracture permeability coefficient of the fault fracture zone and the result of each parameter combination. For combinations whose errors exceed the threshold, the differential evolution algorithm is called to perform differential mutation and crossover operations to generate new combinations to replace the eliminated items, thus obtaining the parameter fitness matrix.

[0047] S3: The ant colony algorithm is adopted. The fitness value of each combination in the parameter fitness matrix is ​​directly used as the initial pheromone concentration. The path selection probability is calculated based on the ratio of each fitness value to the total. The pheromone concentration is updated according to a fixed evaporation ratio.

[0048] S4: Based on the pheromone concentration set, extract the top N combinations and high-concentration path combinations from the fitness matrix, pair them by position and compare parameter values, call the detection porosity and rock parent material compressive strength, perform double screening on the corresponding items, retain the combinations that meet the conditions at the same time, and form a parameter candidate set.

[0049] S5: Calculate the Euclidean distance between each set of parameters in the candidate parameter set and the on-site construction monitoring curve values, and compare them with multiple thresholds. By reading the lining deformation data, fitting the grouting rebound curve, and performing minimum distance statistics, a parameter scheme is generated.

[0050] The specific steps for generating the parametric mesh structure are as follows:

[0051] Based on the grouting pressure, grouting rate, water-cement ratio, diffusion radius, gel time, and grouting volume per unit length, the grid is divided. For each parameter type, the parameter values ​​are traversed by numerical step size and cross-combined to generate a node list. The nodes are numbered and attribute fields are filled to generate a grid node parameter set.

[0052] Based on the grid node parameter set, core strength and anchor tension records are extracted for each node and the numerical difference is calculated. The difference is compared with the threshold and nodes that do not match are removed to establish a parametric grid structure.

[0053] Based on the meshing results of grouting pressure, grouting rate, water-cement ratio, diffusion radius, gel time, and grout volume per unit length, a multi-nested loop generation method was used to traverse the parameter values ​​for each parameter type according to the numerical step size and cross-combine to generate a node list. The grouting pressure traversal step size was set to 0.1 MPa, with a value range of 0.3 MPa to 1.5 MPa; the grouting rate traversal step size was set to 0.1 L / min, with a value range of 1.0 L / min to 5.0 L / min; the water-cement ratio traversal step size was 0.1, with a value range of 0.6 to 1.2; and the diffusion radius traversal step size was... The value ranges from 0.5m to 3.0m. The gelation time is set in 1min increments, ranging from 10min to 60min. The grouting volume per unit length is traversed in 2kg / m increments, ranging from 10kg / m to 60kg / m. During execution, the six types of parameter values ​​are combined sequentially through nested loops from the outer layer to the inner layer to generate a unique combination node number for each group. The field filling method is used to write the index field, traversal batch field, and combination status field into the data structure of each node. All node combination information is summarized to generate a grid node parameter set.

[0054] Based on the grid node parameter set, the structural difference calculation method is used to extract the rock core strength and anchor tension records for each node and calculate the numerical difference. During the execution process, the corresponding rock core uniaxial compressive strength record and anchor prestress value are read by calling the node number mapping table to index the borehole monitoring data file. The strength difference calculation accuracy is set to 0.1kN. The difference calculation command is executed and an error value field is generated. The error threshold is set to 10kN. The error filtering method is used to mark nodes that exceed the threshold as invalid, remove the node number and attribute field data, retain the valid node combinations with errors within the threshold range and write them into the status field identifier, summarize the remaining node attribute structure information, and establish the parametric grid structure.

[0055] The specific steps for generating the parametric mesh structure are as follows:

[0056] Based on the node values ​​in the parametric grid structure, the grouting pressure, grouting rate, water-cement ratio, diffusion radius, gel time, and grouting volume per unit length are constructed into a six-dimensional vector, combined in order into an ordered array, node numbers are added and combination indices are marked, and all node combination data are summarized to generate a parameter combination code set.

[0057] Based on the parameter combination encoding set, the coordinates corresponding to each combination number are extracted. The differential evolution algorithm is used to read the permeability coefficient of the fracture zone of the fault. The grouting pressure is multiplied by the permeability coefficient to obtain the predicted diffusion radius. The absolute value of the difference between the predicted value and the original value is calculated. The number and error value pairs are summarized to generate an error value list.

[0058] Based on the error value list, all combinations with error values ​​greater than the set threshold are screened out. The structure is copied from the low-error combinations to generate new combinations to replace the eliminated combinations. The combinations are renumbered and the complete matrix is ​​restored to complete the combination update and obtain the parameter fitness matrix.

[0059] Based on the node values ​​in the parameter grid structure, an ordered vector coding method is used to combine grouting pressure, grouting rate, water-cement ratio, diffusion radius, gel time, and grouting volume per unit length. During execution, six parameter value vectors are first constructed according to the parameter reading order. The grouting pressure value is read as a floating-point number in MPa, the grouting rate value is read as a floating-point number in L / min, the water-cement ratio value is set to a decimal format, the diffusion radius is set in meters, the gel time is read in minutes, and the grouting volume per unit length is set in kilograms per meter. The six types of parameters are sequentially assembled using a parameter assembly command to form an ordered one-dimensional array structure. A position information weighted label is used to combine the indexes. The combined index field is composed of a prefix parameter identifier and an incrementing number. A node number field is added to each combination and written into the parameter structure in text format. A full node combination data merging command is executed to summarize all node coding content and parameter attribute fields to generate a parameter combination coding set.

[0060] Based on the parameter combination encoding set, the differential evolution algorithm is used to locate the coordinates corresponding to each combination number and perform parameter update operations. During the execution, the initial population size of the algorithm is set to 100, the control parameter F is set to 0.8, the crossover probability CR is set to 0.9, and the maximum number of iterations is set to 500. The grouting pressure value field in each combination is extracted and multiplied with the fracture permeability coefficient of the corresponding coordinate point. The fracture permeability coefficient is extracted from the fault fracture zone structure and read in floating-point format with units of m / s. The product value is used as the predicted grouting diffusion radius. The diffusion radius field in the original combination is extracted and the calculation accuracy is set to 0.01m. The difference calculation operation is performed to obtain the absolute difference between the predicted value and the original value. An error value field is generated and bound to the combination number. The number and error value are paired and written into the structure. All combination numbers and error values ​​are summarized to generate an error value list.

[0061] Based on the error value list, a target error-driven replacement method is used to filter out all combinations with error values ​​greater than a set threshold of 0.5m. A filtering command is executed on the error field to extract the numbers of all combinations that meet the threshold. The structural information of each combination is read and all parameter fields are copied. The positions of the previously removed numbers are replaced according to the population index position order. During the replacement process, the node numbers are updated and the combination index field is reset. All parameter field formats are kept consistent with the initial encoding. A command to restore the data structure matrix after replacement is executed, the mutation update label field is written and the combination status is synchronized, the node encoding structure is reordered in the original order and archived, and the combination update is completed to obtain the parameter fitness matrix.

[0062] The differential evolution algorithm first uses the constructed parameter combination encoding set as the initial population, with each combination corresponding to a six-dimensional vector node. For each parameter combination, three distinct combinations are selected from the population, and new mutant combinations are generated by superimposing the differences. Then, the mutant combinations are crossovered with the original combinations, and the parameters are retained or replaced according to the set crossover probability to obtain the experimental combinations. Subsequently, the grouting pressure parameters in the experimental combinations are read and multiplied with the permeability coefficient of the fault fracture zone to calculate the corresponding predicted grouting diffusion radius. The error value is compared with the diffusion radius value of the original combination. If the error of the experimental combination is smaller than that of the original combination, the original combination is replaced. Finally, through multiple rounds of iterative updates, combinations are eliminated according to the error magnitude to generate an optimized parameter fitness matrix.

[0063] Differential evolution algorithm, according to the formula:

[0064]

[0065] in: This indicates the predicted grouting diffusion radius. This indicates the volume of slurry injected per unit time. Indicates the dynamic viscosity of the slurry. This indicates the connectivity of fractures within a fault fracture zone. The adjustment coefficient representing the fracture connectivity. This indicates the saturation level of the residual water in the fissure. The adjustment coefficient represents the residual water saturation. This represents the permeability coefficient of the fracture zone in the fault. This indicates the grouting pressure applied during grouting construction. This represents the original pore water pressure in the formation. Pi is a constant.

[0066] Execution process: First, based on the constructed parameter combination encoding set, the coordinate information corresponding to each group number is extracted, and the grouting parameters at the corresponding location, including the grouting flow rate, are extracted from the 3D geological model. Dynamic viscosity of grouting material Fault fracture zone permeability coefficient Grouting pressure and formation pore water pressure Subsequently, the connectivity of regional fractures is obtained by calling the fault fracture network model. and residual water saturation within the fissures To enhance the model's ability to express the response characteristics of complex geological structures, an adjustment coefficient is introduced. and The influence of fracture connectivity and residual saturation on the slurry diffusion process was controlled separately. The adjustment coefficients were determined using a minimum error optimization strategy. First, the objective function of error was set as the absolute value of the difference between the measured diffusion radius and the predicted diffusion radius. Multiple combined samples were substituted into the formula to calculate the error between the preliminary prediction result and the measured value. Then, the error was adjusted iteratively through gradient descent. and The value of the optimized parameter is used to minimize the prediction error. Finally, all optimized parameters are substituted into the formula to calculate the predicted diffusion radius. The error values ​​are compared with the measured values ​​to generate a list of error values, providing a fitness evaluation basis for the differential evolution algorithm in selecting reinforcement parameters.

[0067] The specific steps for generating the parametric mesh structure are as follows:

[0068] Based on the parameter fitness matrix, the ant colony algorithm is used to extract the combination number and fitness value of each group, calculate the sum of fitness values, normalize the fitness of each group according to the ratio, establish a path number and concentration pairing table, write the initialization record, and generate the pheromone initialization set.

[0069] Based on the pheromone initialization set, the path numbers are sorted by concentration value, and the corresponding selection probability intervals for each group are accumulated. The mapping relationship between paths and intervals is constructed, the probability range of each group of paths is marked, and a path selection probability set is generated.

[0070] Based on the path selection probability set, the current path concentration is read, the evaporation value is calculated and subtracted from the concentration to generate the updated concentration value, the difference between the old and new concentrations is calculated and written into the change table, the status of each path is recorded, and a pheromone concentration set is generated.

[0071] Based on the parameter fitness matrix, the ant colony algorithm is used to extract the combination number and fitness value of each combination. During the execution process, the number of ants is set to 50, the maximum number of iterations is 200, the upper limit of the pheromone initial concentration is 1.0, and the lower limit is 0.01. The fitness field value of each combination number is read and summed. The fitness value is divided by the sum and normalized. The normalization precision is set to 0.0001. After processing, the combination number and the normalization result are bound as path number and concentration pairing items and written into the initial structure array. At the same time, the path index field, concentration weight field and state field are added. The structure array is saved and the initialization is marked as complete, generating the pheromone initialization set.

[0072] Based on the pheromone initialization set, the path numbers are sorted by concentration value using a concentration sorting mapping method. The concentration sorting method is in descending order. The sorting results are traversed and the concentration value of each group of paths is accumulated and converted into a percentage. The percentage calculation is retained to four decimal places. A mapping table between the path index and the corresponding cumulative concentration value is constructed. Each record includes the path number, the upper limit of cumulative concentration, the lower limit of cumulative concentration, and the original concentration field. The path mapping identifier field is used to divide each group of concentration intervals into non-overlapping intervals. After the path number is bound to the interval, it is written into the path probability structure array. The structure includes the path index, the upper limit of the path probability interval, the lower limit of the path probability interval, the path selectability status, and the cumulative concentration ratio field, generating a path selection probability set.

[0073] Based on the path selection probability set, the current path concentration field value is read using a concentration decrease update method. The evaporation ratio coefficient is set to 0.1. The path concentration value is read and multiplied by the evaporation coefficient to obtain the evaporation value. The evaporation value is retained to three decimal places in floating-point format. The updated concentration field is generated by subtracting from the original path concentration. The updated concentration field is written into the path record structure. The difference between the original concentration and the updated concentration is read and the difference is calculated to generate a concentration difference field, which is written into the concentration change record table. At the same time, the current status field corresponding to the path number is updated. The status field includes an update flag, a concentration change flag, and a path availability label. Finally, all path structure information is archived to generate a pheromone concentration set.

[0074] The ant colony algorithm first reads the initial pheromone concentration of each path in the fitness matrix, normalizes all fitness values ​​by summing them to obtain the concentration of each path, then calculates the path selection probability based on the ratio of the concentration of each path to its heuristic pheromone value, uses a roulette wheel algorithm to determine the ant's path, then deducts the current pheromone concentration according to a set evaporation rate, and allocates pheromone increments according to the fitness of each path. Finally, the increments are merged into the remaining pheromone concentration, and the process is repeated until the preset number of iterations is reached and the concentrations of all paths converge, outputting the final pheromone concentration set.

[0075] Ant colony algorithm, according to the formula:

[0076]

[0077] in: Representing a path The normalized pheromone concentration value, Representing a path The initial pheromone concentration, Representing a path Heuristic information values, Representing a path fitness value, Representing a path The connectivity stability coefficient, Representing a path The curvature influence coefficient, This indicates that pheromone concentration affects the weighting coefficient. This indicates that heuristic information influences the weighting coefficients. The adjustment weight represents the connectivity stability coefficient. The adjustment weight represents the curvature influence coefficient. Indicates the total number of paths;

[0078] Execution process: First, based on the constructed parameter fitness matrix, extract the fitness value corresponding to each combination number. Simultaneously, the initial pheromone concentration for each path is obtained from the path data structure. and heuristic information values The heuristic information is calculated based on geological parameters such as path length, fracture development degree, and material accessibility. Then, the structural stability of each path is analyzed in the three-dimensional model of the fault fracture zone, and the connectivity stability coefficient is extracted. Then, the curvature attribute of the path is obtained through the geological modeling system, and the curvature influence coefficient is calculated. The above five types of parameters constitute a comprehensive evaluation index for path quality, and then weighting coefficients are set. , , , The effects of pheromone concentration, heuristic information, connectivity stability, and path curvature on path selection probability were controlled separately. Weights were automatically optimized using the minimum deviation method. This method involved setting the objective function as the sum of squared differences between the measured path selection probability and the theoretically calculated probability. Multiple rounds of trial calculations were performed on the sample paths, and the weight coefficients were updated using gradient descent until the error function converged. Finally, all parameters were substituted into the normalized pheromone concentration calculation formula to obtain the normalized pheromone value for each path. A pairing table between path numbers and their corresponding pheromone concentrations is established as the basic pheromone distribution set for the initialization phase of the ant colony algorithm, providing input for subsequent pheromone evaporation updates and path probability scheduling.

[0079] The specific steps for generating the candidate set of parameters are as follows:

[0080] Based on the top N parameter combinations in the parameter fitness matrix and the high-scoring paths in the optimized parameter path set, elements from the two sets are extracted and paired according to their positions. Field values ​​and combination numbers are merged in parallel to form a list of combination pairs and generate a path parameter set.

[0081] Based on the path parameter set, the detection porosity value and the rock parent material compressive strength data are read and compared one by one, and the combination that meets both conditions is screened to form a parameter candidate set.

[0082] Based on the top N parameter combinations in the parameter fitness matrix and high-scoring paths in the optimized parameter path set, the position index pairing method is used to extract elements from the two sets and perform pairing operations according to position. During the execution process, N is set to 20, and the high-score threshold for the path is set to a fitness value of not less than 0.85. The top N parameter combination numbers in the parameter fitness matrix are read in sequence and a number list is generated. All path numbers and fitness fields in the optimized path set are read, and path numbers that meet the threshold conditions are filtered. A bidirectional indexing mechanism is used to pair the top N combinations with high-scoring paths in sequence. The pairing method adopts an index-based one-to-one mapping operation. After pairing, a parallel field splicing operation is performed on the field values ​​in the combination and the parameter fields in the path. The splicing fields include grouting pressure, grouting rate, water-cement ratio, diffusion radius, gel time, grouting volume per unit length, path number, and combination number. The field splicing uses a horizontal array expansion command to generate structured combination pair data units. The structure merging command is executed to summarize all combination pairs and write them into the combination pair record table to generate a path parameter set.

[0083] Based on the path parameter set, a dual-threshold difference filtering method is used to read the detected porosity values ​​and the compressive strength data of the parent material, and the differences are compared one by one. During the execution process, the upper limit of porosity filtering is first set to 3.5% and the lower limit to 1.0%, and the lower limit of compressive strength filtering is set to 30MPa. The porosity data corresponding to each combination number in the path parameter set is read. The data format is set to percentage with one decimal place. The porosity value of the corresponding core segment is extracted from the detection record by number index. The reading format is the 4th column of the CSV file. Then, the parent material compressive strength value is read. The data format is integer unit in MPa. It is extracted from the uniaxial compressive strength record dataset by path combination number. The reading format is TXT file with number keyword search. After reading the two data, the field difference calculation command is executed. The porosity difference is judged to see if it exceeds the set upper and lower limit range, and the compressive strength is judged to see if it is less than the set lower limit value. The combination that does not meet either condition is set as invalid combination and a rejection label field is added. The combination that meets both conditions is marked as retained and the status field is updated to "valid". The retained combination information is written into a new structure array to form a parameter candidate set.

[0084] The specific steps for generating a scheme with defined parameters are as follows:

[0085] Based on each set of parameters in the parameter candidate set and the on-site construction monitoring curve data, the grouting parameter sequence and the monitoring curve points are extracted, and the distance is obtained by summing the square differences and taking the square root. Multiple threshold filtering is applied to select combinations and generate a low difference combination set.

[0086] Based on the low difference combination set, the lining deformation record and grouting rebound curve sampling points are read and fitted and compared. The combination error values ​​are sorted and the minimum term is selected to generate a definite parameter scheme.

[0087] Based on the parameter candidate set and the on-site construction monitoring curve data, the sequence difference statistical method is used to read the grouting parameter sequence field in the monitoring curve CSV file and extract the grouting pressure, water-cement ratio and grouting volume per unit length according to the sampling order. The curve point field in the monitoring curve TXT file is read and the corresponding deformation is extracted according to the timestamp. The loop traversal operation is performed. The square difference accumulation command is used to calculate the square difference between each parameter sequence and the corresponding curve point sequence and accumulate the difference. The square root command is used to calculate the square root of the cumulative value. The multiple threshold filtering command is executed to compare and filter all calculated value fields. All combination numbers and error fields that meet the threshold conditions are extracted and summarized to generate a low difference combination set.

[0088] Based on the low-difference combination set, the rebound fitting comparison method is adopted. The deformation sampling point field in the lining deformation CSV file is read and the coordinates and deformation values ​​are extracted according to the sequence number. The sampling point field in the grouting rebound curve TXT file is read and the time and rebound value are extracted according to the sequence number. The least squares fitting command is executed and the fitting function type is set to linear regression. The parameters include order 1 and intercept flag as true. The fitting command is applied to the sampling point set to generate the fitting curve function and the fitting error field of each combination is calculated. The sorting command is executed to arrange all combination records in ascending order of fitting error. The combination number and parameter field corresponding to the first record are selected and summarized to generate a determined parameter scheme.

[0089] 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, equivalent changes, and modifications 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 determining parameters for pre-grouting reinforcement of fault fracture zones in tunnels, characterized in that, Includes the following steps: S1: Grid the grouting pressure, grouting rate, water-cement ratio, diffusion radius, gelation time, and grouting volume per unit length. Verify the consistency of node values ​​based on core strength and anchor tension data, and establish a parametric grid structure. S2: Using the node values ​​in the parameter grid structure as the initial population code, the error value is calculated by using the field fracture permeability coefficient of the fault fracture zone and the result of each parameter combination. For combinations where the error exceeds the threshold, the differential evolution algorithm is called to perform differential mutation and crossover operations to generate new combinations to replace the eliminated items, thus obtaining the parameter fitness matrix. S3: Using the ant colony algorithm, the fitness value of each combination in the fitness matrix is ​​directly used as the initial pheromone concentration, and the path selection probability is calculated based on the ratio of each fitness value to the total. The pheromone concentration is then updated according to a fixed evaporation ratio. S4: Based on the pheromone concentration set, extract the top N combinations and high-concentration path combinations from the fitness matrix, pair them by position and compare parameter values, call the detection porosity and rock parent material compressive strength, perform double screening on the corresponding items, retain the combinations that meet the conditions at the same time, and form a parameter candidate set. S5: Calculate the Euclidean distance between each set of parameters in the candidate parameter set and the on-site construction monitoring curve values, and compare them with multiple thresholds. By reading the lining deformation data, fitting the grouting rebound curve, and performing minimum distance statistics, a parameter scheme is generated.

2. The method for determining the parameters of pre-grouting reinforcement of tunnel fault fracture zones according to claim 1, characterized in that, The specific steps for generating the parametric mesh structure are as follows: Based on the grouting pressure, grouting rate, water-cement ratio, diffusion radius, gel time, and grouting volume per unit length, the grid is divided. For each parameter type, the parameter values ​​are traversed by numerical step size and cross-combined to generate a node list. The nodes are numbered and attribute fields are filled to generate a grid node parameter set. Based on the set of grid node parameters, the rock core strength and anchor bolt tension records are extracted for each node and the numerical difference is calculated. The difference is compared with the threshold and nodes that do not match are removed to establish a parameter grid structure.

3. The method for determining the parameters of pre-grouting reinforcement of tunnel fault fracture zones according to claim 1, characterized in that, The specific steps for generating the parametric mesh structure are as follows: Based on the node values ​​in the parameter grid structure, the grouting pressure, grouting rate, water-cement ratio, diffusion radius, gel time, and grouting volume per unit length are constructed into a six-dimensional vector, combined in order into an ordered array, node numbers are added and combination indices are marked, all node combination data are summarized, and a parameter combination code set is generated. Based on the parameter combination encoding set, the coordinates corresponding to each combination number are extracted. The differential evolution algorithm is used to read the permeability coefficient of the fracture zone of the fault. The grouting pressure is multiplied by the permeability coefficient to obtain the predicted diffusion radius. The absolute value of the difference between the predicted value and the original value is calculated. The number and error value pairs are summarized to generate an error value list. Based on the error value list, all combinations with error values ​​greater than a set threshold are filtered out. The structure is copied from the low-error combinations to generate new combinations to replace the eliminated combinations. The combinations are then renumbered and the complete matrix is ​​restored to complete the combination update and obtain the parameter fitness matrix.

4. The method for determining the parameters of pre-grouting reinforcement of tunnel fault fracture zones according to claim 3, characterized in that, The differential evolution algorithm first uses the constructed parameter combination encoding set as the initial population, with each combination corresponding to a six-dimensional vector node. For each parameter combination, three distinct combinations are selected from the population, and new mutated combinations are generated by superimposing the differences. Then, the mutated combinations are cross-crossed with the original combinations, and the parameters are retained or replaced according to the set cross-cross probability to obtain experimental combinations. Subsequently, the grouting pressure parameters in the experimental combinations are read and multiplied with the permeability coefficient of the fault fracture zone to calculate the corresponding predicted grouting diffusion radius. The error value is compared with the diffusion radius value of the original combination. If the error of the experimental combination is smaller than that of the original combination, the original combination is replaced. Finally, through multiple rounds of iterative updates, combinations are eliminated according to the error magnitude to generate an optimized parameter fitness matrix.

5. The method for determining the parameters of pre-grouting reinforcement of tunnel fault fracture zones according to claim 3, characterized in that, The differential evolution algorithm is based on the formula: ； in: This indicates the predicted grouting diffusion radius. This indicates the volume of slurry injected per unit time. Indicates the dynamic viscosity of the slurry. This indicates the connectivity of fractures within a fault fracture zone. The adjustment coefficient representing the fracture connectivity. This indicates the saturation level of the residual water in the fissure. The adjustment coefficient represents the residual water saturation. This represents the permeability coefficient of the fracture zone in the fault. This indicates the grouting pressure applied during grouting construction. This represents the original pore water pressure in the formation. This represents the constant value of pi.

6. The method for determining the parameters of pre-grouting reinforcement of tunnel fault fracture zones according to claim 1, characterized in that, The specific steps for generating the parametric mesh structure are as follows: Based on the fitness matrix, the ant colony algorithm is used to extract the combination number and fitness value of each group, calculate the sum of fitness values, normalize the fitness of each group according to the ratio, establish a path number and concentration pairing table, write the initialization record, and generate the pheromone initialization set. Based on the pheromone initialization set, the path numbers are sorted by concentration value, and the corresponding selection probability intervals for each group are accumulated. The mapping relationship between paths and intervals is constructed, the probability range of each group of paths is marked, and a path selection probability set is generated. Based on the path selection probability set, the current path concentration is read, the evaporation value is calculated and subtracted from the concentration to generate an updated concentration value, the difference between the old and new concentrations is calculated and written into the change table, the status of each path is recorded, and a pheromone concentration set is generated.

7. The method for determining the parameters of pre-grouting reinforcement of tunnel fault fracture zones according to claim 6, characterized in that, The ant colony algorithm first reads the initial pheromone concentration of each path in the fitness matrix, normalizes all fitness values ​​by summing them to obtain the concentration of each path, then calculates the path selection probability based on the ratio of the concentration of each path to its heuristic pheromone value, uses a roulette wheel algorithm to determine the ant's path, then deducts the current pheromone concentration according to a set evaporation rate, and allocates pheromone increments according to the fitness of each path. Finally, the increments are merged into the remaining pheromone concentration, and the algorithm is repeated until the preset number of iterations is reached and the concentrations of all paths converge, outputting the final pheromone concentration set.

8. The method for determining the parameters of pre-grouting reinforcement of tunnel fault fracture zones according to claim 6, characterized in that, The ant colony algorithm is based on the formula: ； in: Representing a path The normalized pheromone concentration value, Representing a path The initial pheromone concentration, Representing a path Heuristic information values, Representing a path fitness value, Representing a path The connectivity stability coefficient, Representing a path The curvature influence coefficient, This indicates that pheromone concentration affects the weighting coefficient. This indicates that heuristic information influences the weighting coefficients. The adjustment weight represents the connectivity stability coefficient. The adjustment weight represents the curvature influence coefficient. This represents the total number of paths.

9. The method for determining the parameters of pre-grouting reinforcement of tunnel fault fracture zones according to claim 1, characterized in that, The specific steps for generating the parameter candidate set are as follows: Based on the top N parameter combinations in the parameter fitness matrix and the high-scoring paths in the optimized parameter path set, elements from the two sets are extracted and paired according to their positions. Field values ​​and combination numbers are merged in parallel to form a list of combination pairs and generate a path parameter set. Based on the path parameter set, the porosity value and the compressive strength data of the parent rock are read and compared one by one, and combinations that simultaneously meet the two conditions are selected to form a parameter candidate set.

10. The method for determining the parameters of pre-grouting reinforcement of tunnel fault fracture zones according to claim 1, characterized in that, The specific steps for generating the determined parameter scheme are as follows: Based on each set of parameters in the candidate parameter set and the on-site construction monitoring curve data, the grouting parameter sequence and monitoring curve points are extracted, and the distance is calculated by taking the square root of the sum of the square differences. Multiple threshold filtering is applied to select combinations and generate a low difference combination set. Based on the low difference combination set, the lining deformation record and the grouting rebound curve sampling points are read and fitted and compared. The combined error values ​​are sorted and the minimum term is selected to generate a definite parameter scheme.

Images