Optimization method of mechanical parameters of large-scale rock mass based on multi-scale test data

Abstract

The application discloses a large-scale rock mass mechanical parameter optimization method based on multi-scale test data and relates to the technical field of parameter optimization. Test process data of multiple rock mass test scales are collected, energy evolution analysis is performed on the test process data, a multi-scale abnormal path database is obtained, energy path characteristic data of a rock mass simulation engineering scale are collected for multi-feature correlation analysis, matching evaluation values of each rock mass test scale of the rock mass simulation engineering scale are obtained, deviation discriminant analysis is performed on the matching evaluation values of each rock mass test scale, energy path matching information of the rock mass simulation engineering scale is processed and obtained, then, a path correction constraint is performed on a mechanical parameter optimization process of the rock mass simulation engineering scale, path deviation analysis is performed on the path correction constraint process, path deviation characteristic values are obtained, and associated rock mass mechanical parameters are updated and optimized according to the path deviation characteristic values, so that the reliability and engineering safety of the rock mass mechanical parameter optimization are improved.

Description

Technical Field

[0001] This invention relates to the field of parameter optimization technology, specifically to a method for optimizing large-scale rock mechanics parameters based on multi-scale experimental data. Background Technology

[0002] In the construction and advancement of deep underground engineering, the mechanical parameters of large-scale rock masses are the fundamental inputs for conducting surrounding rock stability analysis, support structure design, and construction scheme optimization. Their rationality is directly related to engineering safety and long-term service performance. Since it is difficult to obtain the true mechanical parameters of engineering-scale rock masses directly through overall loading tests, engineering practice usually relies on multi-scale rock mass test data. Through parameter backpropagation and optimization methods, the results of indoor small-scale tests, field meso-scale tests, and engineering-scale numerical simulations are comprehensively utilized. The core task is to obtain a set of rock mass mechanical parameters that have physical consistency and engineering interpretability at the engineering scale, while ensuring full utilization of multi-scale data.

[0003] However, during deep engineering projects, as excavation disturbances continue to evolve, rock mass structural surfaces are gradually exposed, leading to significant changes in the stress and seepage fields. The stress state of the rock mass may undergo rapid rearrangement within a short period, shifting from a deformation mechanism dominated by continuous media to a failure mechanism dominated by structural surface slippage and crack opening. In this specific scenario, existing multi-scale parameter optimization algorithms, which primarily rely on the fitting effect of macroscopic response curves for parameter updates, lack the identification and constraint of energy dissipation paths and physical mechanisms implicit in tests at different scales. This results in inconsistent energy contribution relationships between data at different scales during parameter optimization. The fracture energy and damage evolution characteristics reflected in small-scale tests are abnormally amplified during parameter backpropagation, causing the constraint effect on the optimization results to dominate in a short period. The constraint ability of large-scale simulation on overall deformation and stability is significantly weakened, ultimately leading to abnormal rock mechanics parameters obtained through inversion. Rock mass stability assessments and support designs based on these parameters will deviate from actual mechanical behavior, making it difficult to accurately identify potential instability risks and potentially inducing structural failure during construction and operation, posing significant safety hazards. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for optimizing large-scale rock mechanics parameters based on multi-scale experimental data, which can effectively solve the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution: a method for optimizing large-scale rock mass mechanics parameters based on multi-scale test data, including collecting test process data at multiple rock mass test scales, performing energy evolution analysis on the test process data to obtain energy path characteristic data at each rock mass test scale, and performing anomaly correlation analysis on the energy path characteristic data at each rock mass test scale to obtain a multi-scale anomaly path database.

[0006] Energy path characteristic data at the engineering scale of rock mass simulation were collected. Combined with a multi-scale anomaly path database, multi-feature correlation analysis was performed on the energy path characteristic data at the engineering scale of rock mass simulation to obtain the matching evaluation values ​​of each rock mass test scale at the engineering scale of rock mass simulation.

[0007] Deviation discriminant analysis is performed on the matching evaluation values ​​of each rock mass test scale at the rock mass simulation engineering scale to obtain the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale. Based on the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale, the energy path matching information of the rock mass simulation engineering scale is obtained, and then the path correction constraint is applied to the mechanical parameter optimization process at the rock mass simulation engineering scale.

[0008] Path offset analysis is performed on the path correction constraint process to obtain the path offset characteristic value at the engineering scale of rock mass simulation, and the associated rock mass mechanical parameters are updated and optimized based on the path offset characteristic value at the engineering scale of rock mass simulation.

[0009] Furthermore, the method for energy evolution analysis of the test process data is as follows: the test process data of the multi-rock mass test scale includes mechanical response data, displacement data, internal pore pressure data and energy data of each rock mass test scale.

[0010] Stress-deformation evolution analysis was performed on the mechanical response data of each rock mass test scale to obtain the mechanical response curves of each rock mass test scale; displacement analysis was performed on the displacement data of each rock mass test scale to obtain the displacement response curves of each rock mass test scale; pore pressure variation analysis was performed on the internal pore pressure data of each rock mass test scale to obtain the internal pore pressure variation curves of each rock mass test scale; and energy dissipation integration analysis was performed on the energy data of each rock mass test scale to obtain the energy dissipation curves of each rock mass test scale.

[0011] Based on the mechanical response curves, displacement response curves, internal pore pressure change curves, and energy dissipation curves of each rock mass test scale, energy path characteristic data for each rock mass test scale are constructed.

[0012] Furthermore, the method for performing anomaly correlation analysis on energy path characteristic data of each rock mass test scale is as follows: extract energy path characteristic data of each rock mass test scale, analyze and obtain the anomaly path type of each rock mass test scale, and construct a multi-scale anomaly path database based on the energy path characteristic data of each rock mass test scale and the corresponding anomaly path type.

[0013] Furthermore, the energy path characteristic data at the engineering scale of the rock mass simulation includes the mechanical response curve, displacement response curve, internal pore pressure change curve, and energy dissipation curve at the engineering scale of the rock mass simulation.

[0014] Furthermore, the method for performing multi-feature correlation analysis on the energy path characteristic data of the rock mass simulation engineering scale is as follows: the energy path characteristic data of the rock mass simulation engineering scale is compared with the energy path characteristic data of all rock mass test scales in the multi-scale anomaly path database to obtain the matching evaluation value of each rock mass test scale and the corresponding anomaly path type of the rock mass simulation engineering scale. The matching evaluation value of each rock mass test scale of the rock mass simulation engineering scale is used to characterize the quantitative result of the consistency of energy paths of the rock mass simulation engineering scale under different rock mass test scales.

[0015] Furthermore, the method for deviation discrimination analysis of the matching evaluation values ​​of each rock mass test scale at the rock mass simulation engineering scale is as follows: the matching evaluation values ​​of each rock mass test scale at the rock mass simulation engineering scale are compared with the preset corresponding rock mass test scale matching evaluation threshold. If the matching evaluation value of the corresponding rock mass test scale at the rock mass simulation engineering scale is higher than the preset corresponding rock mass test scale matching evaluation threshold, the matching evaluation result of the corresponding rock mass test scale at the rock mass simulation engineering scale is marked as qualified for energy path matching; otherwise, the matching evaluation result of the corresponding rock mass test scale at the rock mass simulation engineering scale is marked as unqualified for energy path matching.

[0016] Furthermore, the method for obtaining energy path matching information at the rock mass simulation engineering scale is as follows: extract the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale and the corresponding abnormal path types. If there are two or more abnormal path types corresponding to the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale that are qualified for energy path matching, then the energy path matching information at the rock mass simulation engineering scale is marked as cross-scale matching abnormal; otherwise, no marking is performed.

[0017] Furthermore, the method for path correction constraints on the mechanical parameter optimization process at the rock mass simulation engineering scale is as follows: extract the energy path matching information at the rock mass simulation engineering scale; if the energy path matching information at the rock mass simulation engineering scale is an anomaly across scales, sort the corresponding matching evaluation values ​​of each rock mass test scale in descending order, mark the anomaly path type corresponding to the highest rock mass test scale matching evaluation value as the dominant energy path type of that rock mass simulation engineering scale, and then apply path correction constraints to the mechanical parameter optimization process at the rock mass simulation engineering scale.

[0018] Furthermore, the method for performing path offset analysis on the path correction constraint process is as follows: performing path offset analysis on the path correction constraint process to obtain path offset characteristic data at the rock mass simulation engineering scale. The path offset characteristic data at the rock mass simulation engineering scale includes stress data offset coefficient, displacement data offset coefficient, internal pore pressure data offset coefficient, and energy data offset coefficient at the rock mass simulation engineering scale.

[0019] A comprehensive analysis of path offset characteristic data at the engineering scale of rock mass simulation is conducted to obtain path offset characteristic values ​​at the engineering scale of rock mass simulation. These path offset characteristic values ​​at the engineering scale of rock mass simulation are used to quantify the severity of matching deviations between different energy paths during parameter optimization.

[0020] Furthermore, the method for updating and optimizing the associated rock mass mechanical parameters based on the path offset characteristic value at the rock mass simulation engineering scale is as follows: the path offset characteristic value at the rock mass simulation engineering scale is compared with a preset path offset characteristic threshold. If the path offset characteristic value at the rock mass simulation engineering scale is higher than the preset path offset characteristic threshold, the path offset result at the rock mass simulation engineering scale is marked as a path anomaly; otherwise, the path offset result at the rock mass simulation engineering scale is marked as a path normal. If the path offset result at the rock mass simulation engineering scale is a path anomaly, the associated rock mass mechanical parameters are updated and optimized.

[0021] The present invention has the following beneficial effects: This invention substantially solves the fundamental flaw of existing multi-scale parameter back-inference methods, which only fit at the macroscopic response level and ignore the differences in physical mechanisms, by introducing a multi-scale consistency constraint mechanism with energy paths as the core. This invention constructs a multi-scale abnormal path database, explicitly models and quantifies the destruction physical mechanisms and energy dissipation paths corresponding to each scale of the experiment, and performs multi-feature correlation analysis and matching evaluation in engineering-scale simulation. This can effectively avoid the problem of implicit folding of multi-scale energy constraints and sudden jumps in parameter evolution paths when key operating condition thresholds are crossed, thus improving the authenticity and stability of parameter inversion results from a physical perspective.

[0022] This invention introduces cross-scale matching anomaly discrimination, dominant energy path type identification, and path correction constraint mechanisms to achieve proactive intervention and risk pre-control in the parameter optimization process. When the engineering-scale simulation behavior simultaneously meets the matching qualification conditions of multiple abnormal path types, this invention can promptly identify high-risk states of multi-steady-state competition or mechanism overlap. By constraining the parameter value range and update step size, the optimization process is forcibly guided to a safe evolution channel consistent with the dominant physical mechanism, thereby effectively suppressing large-scale jumps in numerical iteration within the critical region and reducing the possibility of systemic failure risks in overall stability assessment and support design due to parameter distortion.

[0023] This invention constructs a path offset analysis and path offset characteristic value determination mechanism to monitor the consistency of the entire parameter optimization process in multiple physics fields, and provides a safe recovery means for parameter reset in extreme cases. It can not only quantitatively evaluate the comprehensive deviation of multi-dimensional responses such as stress, displacement, internal pore pressure and energy at different scales, but also trigger timely updates and optimization configurations when constraints fail or local ill-conditions occur, so as to avoid the optimization process from falling into an irreversible erroneous path. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Please see Figure 1 As shown, this embodiment of the invention provides a technical solution: a method for optimizing large-scale rock mass mechanical parameters based on multi-scale test data, including collecting test process data at multiple rock mass test scales, performing energy evolution analysis on the test process data to obtain energy path characteristic data at each rock mass test scale, and performing anomaly correlation analysis on the energy path characteristic data at each rock mass test scale to obtain a multi-scale anomaly path database.

[0027] Energy path characteristic data at the engineering scale of rock mass simulation were collected. Combined with a multi-scale anomaly path database, multi-feature correlation analysis was performed on the energy path characteristic data at the engineering scale of rock mass simulation to obtain the matching evaluation values ​​of each rock mass test scale at the engineering scale of rock mass simulation.

[0028] Deviation discriminant analysis is performed on the matching evaluation values ​​of each rock mass test scale at the rock mass simulation engineering scale to obtain the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale. Based on the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale, the energy path matching information of the rock mass simulation engineering scale is obtained, and then the path correction constraint is applied to the mechanical parameter optimization process at the rock mass simulation engineering scale.

[0029] Path offset analysis is performed on the path correction constraint process to obtain the path offset characteristic value at the engineering scale of rock mass simulation, and the associated rock mass mechanical parameters are updated and optimized based on the path offset characteristic value at the engineering scale of rock mass simulation.

[0030] Specifically, the method for energy evolution analysis of the test process data is as follows: the test process data of the multi-rock mass test scale includes mechanical response data, displacement data, internal pore pressure data and energy data of each rock mass test scale.

[0031] Stress-deformation evolution analysis was performed on the mechanical response data of each rock mass test scale to obtain the mechanical response curves of each rock mass test scale; displacement analysis was performed on the displacement data of each rock mass test scale to obtain the displacement response curves of each rock mass test scale; pore pressure variation analysis was performed on the internal pore pressure data of each rock mass test scale to obtain the internal pore pressure variation curves of each rock mass test scale; and energy dissipation integration analysis was performed on the energy data of each rock mass test scale to obtain the energy dissipation curves of each rock mass test scale.

[0032] It should be added that mechanical response data refers to the stress and strain changes of the rock mass during loading tests. Real-time monitoring of the rock mass is achieved using mechanical sensors and strain gauges, recording stress changes and corresponding strain data under different load tests to obtain a mechanical response curve at the test scale, representing the elastic strain energy of the rock mass. Displacement data refers to the displacement response of the rock mass during loading, reflecting its deformation characteristics. Precise measurements are taken on the rock mass surface using displacement sensors, collecting data at multiple points and recording the displacement value of each point over time. Analysis of the displacement data yields a displacement response curve at the test scale, and averaging the displacement data at each point provides overall deformation information, reflecting the rock mass's displacement capacity under different loads. Internal pore pressure data refers to the pressure changes of the fluid within the pores of the rock mass, characterizing the response of the pore medium under stress. Pore pressure sensors are used to measure and record pore pressure changes under different loading test conditions. Based on the pore pressure change data, pressure change analysis is performed to obtain an internal pore pressure change curve, reflecting the changes in pore pressure within the rock mass and its contribution to energy loss. Energy data refers to the energy changes of rock mass under different loading test conditions. Energy sensors are used to measure the energy input and release of rock mass under different mechanical responses. Through energy dissipation integration analysis, energy dissipation curves are obtained, showing the energy loss of rock mass under damaged or non-damaged conditions.

[0033] Based on the mechanical response curves, displacement response curves, internal pore pressure change curves, and energy dissipation curves of each rock mass test scale, energy path characteristic data for each rock mass test scale are constructed.

[0034] This step defines the specific physical meaning and construction method of the energy path, serving as the data foundation of the entire methodology. Its beneficial effect lies in integrating traditional, isolated mechanical parameter curves into a comprehensive energy path curve cluster that can simultaneously reflect stress, deformation, seepage, and energy dissipation. This allows for a more complete characterization of energy transfer and transformation behavior during rock mass failure. The underlying principle is that energy is a unified scale connecting different scales and physical processes. By constructing energy paths, microscopic crack propagation energy, mesoscopic frictional slip energy, and macroscopic deformation energy and seepage energy can be correlated and analyzed within the same framework, providing the possibility of achieving true cross-scale energy consistency constraints.

[0035] Specifically, the method for anomaly correlation analysis of energy path characteristic data at each rock mass test scale is as follows: extract energy path characteristic data at each rock mass test scale, analyze and obtain the anomaly path types at each rock mass test scale, the anomaly path types include brittle fracture-dominated type, ductile flow-dominated type, structural plane slip-dominated type, pore compression-dominated type and mixed transition type, and construct a multi-scale anomaly path database based on the energy path characteristic data at each rock mass test scale and the corresponding anomaly path types.

[0036] This step completes the digital archiving of physical mechanisms. Its beneficial effect is that it creates a structured physical mechanism-energy path feature mapping knowledge base, making data-driven mechanism identification possible. Its principle is to establish a stable correlation between empirical and qualitative failure modes in the field of rock mechanics, such as brittleness and ductility, and measurable and quantitative energy evolution data features, such as curve shape, inflection point, and energy consumption rate, thereby precipitating domain knowledge into computable and reusable data.

[0037] It should be noted that the energy path characteristic data of each rock mass test scale are input into the trained anomaly path classification model to obtain the anomaly path type of each rock mass test scale.

[0038] Training methods for abnormal path classification models include: An energy path labeling dataset was constructed for training an anomaly path classification model. This dataset includes complete energy path feature data sequences collected at multiple rock mass test scales under different loading conditions, along with corresponding anomaly path type labels. In a laboratory environment, standard rock mass samples covering different lithologies and joint development levels were subjected to various mechanical loading paths, including uniaxial compression, triaxial confining pressure, cyclic loading and unloading, and loading at different rates. Energy path feature data sequences of the rock mass samples throughout the loading to failure process were synchronously acquired and analyzed using high-precision sensors. Subsequently, rock mechanics experts, combining macroscopic and microscopic morphological analysis of the samples after failure, labeled each energy path feature data sequence with corresponding anomaly path type labels. The set of anomaly path type labels includes brittle fracture-dominated, ductile flow-dominated, structural plane slip-dominated, pore compression-dominated, and mixed transitional types.

[0039] A Long Short-Term Memory (LSTM) network capable of handling multi-channel temporal features was selected as the basic architecture of the anomaly path classification model. The input of this model is energy path feature data with a fixed time step and multiple feature dimensions. The energy path feature data sequentially includes mechanical response curves, displacement response curves, internal pore pressure change curves, and energy dissipation curves at various rock mass test scales within the same time window. The output of the model is the probability distribution vector of the anomaly path type label corresponding to the time window. The probability distribution vector represents the likelihood that the input energy path feature sequence is identified as one of the five categories: brittle fracture-dominated, ductile flow-dominated, structural plane slip-dominated, pore compression-dominated, and mixed transition type. The constructed energy path label dataset is divided into a training subset and a validation subset. The anomaly path classification model is trained in a supervised manner using the training subset. During the training process, the predicted type probability is calculated through forward propagation. The cross-entropy loss function is used to measure the difference between the predicted probability distribution and the true anomaly path type label. The backpropagation algorithm and an optimizer, such as the Adam optimizer, are used to update the model parameters to minimize the loss function value.

[0040] After the complete training cycle is completed, the training subset is used to perform a final performance evaluation on the trained abnormal path classification model. When the overall classification accuracy of the model for the five types of abnormal paths—brittle fracture-dominated, ductile flow-dominated, structural surface slip-dominated, pore compression-dominated, and mixed transitional—is stably higher than the preset classification accuracy threshold, the abnormal path classification model is considered to have completed training, and a well-trained abnormal path classification model is obtained.

[0041] In this implementation plan, the construction of a multi-scale anomaly path database helps to unify experimental data from different scales and reflecting different physical mechanisms, such as brittle fracture and structural surface slip, into quantifiable and comparable energy path characteristic data, and classify them into a multi-scale anomaly path database. This provides an objective and standard comparison benchmark for subsequent judgment of which physical mechanism the engineering-scale simulation behavior is consistent with. Its principle is to transform the abstract problem of physical mechanism consistency into a specific data matching problem, and establish a mapping bridge from physical essence to data characteristics.

[0042] Specifically, the energy path characteristic data at the engineering scale of rock mass simulation includes mechanical response curves, displacement response curves, internal pore pressure change curves, and energy dissipation curves at the engineering scale of rock mass simulation.

[0043] Specifically, the method for performing multi-feature correlation analysis on energy path characteristic data at the rock mass simulation engineering scale is as follows: the energy path characteristic data at the rock mass simulation engineering scale is compared with the energy path characteristic data at all rock mass test scales in the multi-scale anomaly path database to obtain the matching evaluation value of each rock mass test scale at the rock mass simulation engineering scale and the corresponding anomaly path type. The matching evaluation value of each rock mass test scale at the rock mass simulation engineering scale is used to characterize the quantitative result of the consistency of energy paths at different rock mass test scales at the rock mass simulation engineering scale.

[0044] In this embodiment, the matching evaluation values ​​of various rock mass test scales at the engineering scale of rock mass simulation can be obtained through the following analysis method, with the specific analysis conditions as follows: ; ; In the formula, This represents the k-th rock mass test scale matching evaluation value at the engineering scale of the rock mass simulation. The correlation coefficient represents the relationship between the j-th characteristic curve at the engineering scale of rock mass simulation and the j-th characteristic curve at the k-th rock mass test scale. This represents the value corresponding to the i-th data point of the j-th characteristic curve at the engineering scale of the rock mass simulation. This represents the numerical average of the j-th characteristic curve at the engineering scale of the rock mass simulation. This represents the value corresponding to the i-th data point of the j-th characteristic curve at the k-th rock mass test scale in the rock mass simulation engineering scale. This represents the numerical average of the j-th characteristic curve at the k-th rock mass test scale in the rock mass simulation engineering scale. This represents the matching evaluation factor corresponding to the j-th characteristic curve of the k-th rock mass test scale in the preset rock mass simulation engineering scale. k represents the number of each rock mass test scale, k=1, 2, 3, ..., p, p represents the total number of rock mass test scales, j represents the number of each characteristic curve, j=1, 2, 3, ..., m, j represents the total number of characteristic curves, and i represents the number of each data point, i=1, 2, 3, ..., n, n represents the total number of data points.

[0045] It should be added that the correlation coefficients between the characteristic curves of the simulated engineering scale and the corresponding characteristic curves of the rock mass test scale indicate the high degree of consistency between the simulated engineering scale behavior and the physical mechanism analyzed at the rock mass test scale. For example, the simulated fracture process is highly consistent with the brittle fracture characteristics of indoor rock samples. This provides an interpretable, statistically based quantitative tool that can clearly illustrate the similarity between the simulation results and a certain test mechanism. Its principle is that by comparing the characteristic curves one by one and calculating the correlation coefficients, the degree of consistency between the simulated engineering scale and the target test scale mechanism in multiple dimensions such as stress response, deformation mode, seepage effect and energy release can be carefully evaluated.

[0046] In this implementation plan, a comprehensive analysis is conducted to obtain matching evaluation values ​​for various rock mass test scales at the engineering scale of rock mass simulation. This can quantitatively assess the overall degree of agreement between the large-scale rock mass mechanical behavior predicted by numerical simulation and the typical failure physical mechanisms observed and summarized at different test scales from micro to macro. This provides cross-scale, multi-dimensional objective evidence for judging the physical authenticity of the set of mechanical parameters used in the current simulation. By integrating the consistency of multiple characteristic curves such as mechanical response, displacement response, internal pore pressure change, and energy dissipation between the engineering scale of rock mass simulation and specific rock mass test scales, the matching evaluation value transforms the originally implicit and potentially conflicting inter-scale mechanism correlations into explicit and comparable quantitative indicators.

[0047] It should be noted that the ability to evaluate in real time and quantitatively the degree of agreement between the engineering-scale mechanical behavior predicted by the current numerical simulation and various physical experiments in the experimental database reveals whether the physical mechanisms hidden behind the simulation parameters are consistent with the actual or expected failure modes. Its principle is to transform the fuzzy judgment of whether the mechanism matches into a precise numerical assessment by calculating quantitative indicators such as correlation coefficients, thus providing data support for identifying the risk of mechanism mismatch.

[0048] Specifically, the method for deviation discrimination analysis of the matching evaluation values ​​of each rock mass test scale at the rock mass simulation engineering scale is as follows: compare the matching evaluation values ​​of each rock mass test scale at the rock mass simulation engineering scale with the preset corresponding rock mass test scale matching evaluation threshold. If the matching evaluation value of the corresponding rock mass test scale at the rock mass simulation engineering scale is higher than the preset corresponding rock mass test scale matching evaluation threshold, then the matching evaluation result of the corresponding rock mass test scale at the rock mass simulation engineering scale is marked as qualified for energy path matching; otherwise, the matching evaluation result of the corresponding rock mass test scale at the rock mass simulation engineering scale is marked as unqualified for energy path matching.

[0049] Specifically, the method for processing the energy path matching information obtained at the rock mass simulation engineering scale is as follows: extract the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale and the corresponding abnormal path types. If there are two or more abnormal path types corresponding to the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale that are qualified for energy path matching, then the energy path matching information at the rock mass simulation engineering scale is marked as cross-scale matching anomaly; otherwise, no marking is performed.

[0050] This implementation scheme can accurately identify the most dangerous multistable competition or mechanism fuzzy critical states in the optimization process, which are direct precursors to parameter path jumps. Its principle is that when engineering simulation results simultaneously meet the qualification criteria of two or more different physical mechanisms, it indicates that the current parameter combination is in a non-monotonic, physically ambiguous overlapping region. The optimization algorithm is extremely sensitive in this region; even small numerical perturbations or changes in iteration step size can cause it to rapidly sway towards one mechanism, completely abandoning the data constraints corresponding to another mechanism, thus triggering a sudden change in the system state. Marking cross-scale matching anomalies is precisely to provide early warning before such a sudden change occurs.

[0051] Specifically, the method for path correction constraints on the optimization process of mechanical parameters at the rock mass simulation engineering scale is as follows: extract the energy path matching information at the rock mass simulation engineering scale; if the energy path matching information at the rock mass simulation engineering scale is an anomaly across scales, sort the corresponding matching evaluation values ​​of each rock mass test scale in descending order, mark the anomaly path type corresponding to the highest rock mass test scale matching evaluation value as the dominant energy path type of that rock mass simulation engineering scale, and then apply path correction constraints to the optimization process of mechanical parameters at the rock mass simulation engineering scale.

[0052] It should be added that path correction constraints are applied to the mechanical parameter optimization process at the rock mass simulation engineering scale. The specific process is as follows: based on the dominant energy path type at the rock mass simulation engineering scale, the constraint rule set corresponding to the dominant energy path type is called from the preset path type-parameter constraint rule library. The constraint rule set at the rock mass simulation engineering scale includes the allowed value range of mechanical parameters and the maximum allowed update step size of different parameters in the optimization iteration. The called constraint rule set is applied to the ongoing mechanical parameter optimization process at the rock mass simulation engineering scale.

[0053] In this implementation plan, this step is a direct means of implementing proactive intervention and preventing sudden jumps. Its beneficial effect is that, after detecting a risk, it proactively and forcibly restricts the search direction and step size to a safe channel that conforms to the most likely physical mechanism. Its principle is that it is equivalent to installing a physical guide and damper for the optimization algorithm. By shrinking the parameter space, such as adjusting the range of values ​​and suppressing the rate of change, such as limiting the update step size, it greatly reduces the possibility of large-scale, jump-like searches near the critical point in the optimization process. It forces the iteration to approach the optimal solution consistent with the dominant mechanism in a smoother and more controllable way, thereby realizing the transformation from possible sudden jumps to a smooth transition.

[0054] Specifically, the method for path offset analysis of the path correction constraint process is as follows: perform path offset analysis on the path correction constraint process to obtain path offset characteristic data at the rock mass simulation engineering scale. The path offset characteristic data at the rock mass simulation engineering scale includes stress data offset coefficient, displacement data offset coefficient, internal pore pressure data offset coefficient, and energy data offset coefficient at the rock mass simulation engineering scale.

[0055] The stress data offset coefficient reflects the degree of deviation between the stress value at the simulated engineering scale of rock mass and the corresponding experimental stress value under multi-scale experimental conditions. By using stress sensors to monitor the stress distribution data of the rock mass in real time at the simulated engineering scale, the stress value is measured at each sampling point and compared with the stress value of the corresponding experimental data at the experimental scale to calculate the offset. By averaging the offsets from multiple sampling points, an average offset value is obtained, which is the stress data offset coefficient. This coefficient characterizes the stress deviation between the simulated engineering scale and the experimental scale of rock mass.

[0056] The displacement data offset coefficient is used to describe the degree of offset between displacement data at the engineering scale of rock mass simulation and displacement data at the test scale of rock mass. Displacement sensors, such as laser rangefinders or displacement sensors, are used to measure the displacement at different locations at the engineering scale of rock mass simulation in real time, and the data is compared with the displacement data at the corresponding test scale of rock mass. The displacement deviation of each monitoring point is calculated, and the displacement data offset coefficient is obtained by averaging the displacement deviations of all monitoring points. It reflects the degree of difference in displacement data of rock mass at different scales.

[0057] The internal pore pressure data offset coefficient is used to reflect the degree of deviation between pore pressure data at the rock mass simulation engineering scale and experimental scale data. By installing pore pressure sensors, such as gas pressure sensors or liquid pressure sensors, at different locations at the rock mass simulation engineering scale and the corresponding experimental scale, pore pressure data at each monitoring point is obtained. Then, the difference between the pore pressure data at the rock mass simulation engineering scale and the experimental scale is calculated. By averaging the offset values ​​of all data points, the pore pressure data offset coefficient is obtained. This internal pore pressure data offset coefficient describes the degree of deviation of pore pressure data at different scales.

[0058] The energy data offset coefficient characterizes the offset between energy release, absorption, and other energy data at the simulated engineering scale and the experimental scale of rock mass. Through energy monitoring equipment, energy sensors record the energy release curve at the simulated engineering scale of rock mass and compare it with the energy data in multi-scale experiments. The energy difference at each sampling point is measured, and the overall energy offset value is obtained through statistical analysis. By calculating the mean of all measured energy deviations, the energy data offset coefficient is obtained. This energy data offset coefficient reflects the difference in energy path between the simulated engineering scale and the experimental scale of rock mass.

[0059] A comprehensive analysis of path offset characteristic data at the engineering scale of rock mass simulation is conducted to obtain path offset characteristic values ​​at the engineering scale of rock mass simulation. These path offset characteristic values ​​at the engineering scale of rock mass simulation are used to quantify the severity of matching deviations between different energy paths during parameter optimization.

[0060] In this embodiment, the path offset characteristic value of the rock mass simulation engineering scale can be obtained through the following analysis method, with the specific analysis conditions as follows: ; In the formula, The path offset characteristic value represents the engineering scale of the rock mass simulation. The stress data offset coefficient represents the engineering scale of the rock mass simulation. This represents the path offset influence factor corresponding to the set unit stress data offset coefficient. The offset coefficient represents the displacement data at the engineering scale of the rock mass simulation. This represents the path offset influence factor corresponding to the set unit displacement data offset coefficient. This represents the offset coefficient of internal pore pressure data at the engineering scale of rock mass simulation. This represents the path offset influence factor corresponding to the set unit internal pore pressure data offset coefficient. Energy data offset coefficient representing the engineering scale of rock mass simulation. This represents the path offset impact factor corresponding to the set unit energy data offset coefficient.

[0061] It should be added that the path offset influence factors corresponding to the unit stress data offset coefficient, unit displacement data offset coefficient, unit internal pore pressure data offset coefficient, and unit energy data offset coefficient are used to adjust the importance of the data in the path offset feature data at the rock mass simulation engineering scale in the process of obtaining path offset feature values. For example, a mapping relationship between the path offset feature data and the path offset influence factors at the rock mass simulation engineering scale is set in the parameter optimization database. The path offset influence factors corresponding to the real-time path offset feature data can be matched through the pre-set mapping relationship. By matching the path offset feature data with the pre-set mapping relationship respectively, the path offset influence factors corresponding to the unit stress data offset coefficient, unit displacement data offset coefficient, unit internal pore pressure data offset coefficient, and unit energy data offset coefficient can be obtained.

[0062] In this implementation plan, the stress data offset coefficient, displacement data offset coefficient, internal pore pressure data offset coefficient, and energy data offset coefficient at the engineering scale of rock mass simulation are correlated and do not exist independently. For example, an abnormal increase in the stress data offset coefficient often directly leads to a synchronous increase in the displacement data offset coefficient, because an unrealistic stress distribution will inevitably cause a deformation response that deviates from the measured data. Significant changes in the internal pore pressure data offset coefficient will, by altering the effective stress field, transmit and amplify the deviation between the stress and displacement data offset coefficients. As a macroscopic criterion for the overall system equilibrium, anomalies in the energy data offset coefficient usually indicate that the aforementioned local data offsets of stress, displacement, and pore pressure have not yet been fully realized. It can satisfy the concentrated embodiment of energy conservation. The path deviation characteristic value of the rock mass simulation at the engineering scale obtained by comprehensive analysis can quantitatively assess the overall deviation between the comprehensive mechanical response predicted by numerical simulation and the real rock mass behavior reflected by multi-source field monitoring data in the current parameter optimization iteration step, in the multi-physics field dimension. This helps to identify the critical state of the parameter evolution path that is about to become unstable or has already "jumped" in the optimization algorithm in a timely manner. It provides accurate decision-making basis for dynamically triggering parameter constraints, adjusting the optimization step size or switching the dominant mechanism, thereby guiding the parameter optimization process back to a physically consistent stable path, and ultimately improving the reliability of the large-scale rock mass mechanical parameter inversion results and the safety of engineering applications.

[0063] Specifically, the method for updating and optimizing the associated rock mass mechanics parameters based on the path offset characteristic value at the rock mass simulation engineering scale is as follows: the path offset characteristic value at the rock mass simulation engineering scale is compared with a preset path offset characteristic threshold. If the path offset characteristic value at the rock mass simulation engineering scale is higher than the preset path offset characteristic threshold, the path offset result at the rock mass simulation engineering scale is marked as a path anomaly; otherwise, the path offset result at the rock mass simulation engineering scale is marked as a path normal. If the path offset result at the rock mass simulation engineering scale is a path anomaly, the associated rock mass mechanics parameters are updated and optimized. When the path offset characteristic value exceeds this threshold, it indicates that even after the aforementioned path correction constraint, the overall imbalance of the multiphysics field is still too severe.

[0064] It should be noted that the process of updating and optimizing the associated rock mass mechanics parameters is as follows: if the path offset result at the engineering scale of the rock mass simulation is abnormal, then the associated rock mass mechanics parameters at the engineering scale of the rock mass simulation are reset.

[0065] In this implementation scheme, the beneficial effect of this step is that even after constraints are applied, the overall health of the optimization process is continuously monitored, and the parameter reset optimization configuration is executed when abnormal indicators are detected. Its principle is to acknowledge and handle the extreme case that the optimization algorithm may fall into constraint failure, and to provide a mechanism for safe recovery from such a state. It is equivalent to installing a reset button for the entire optimization system, ensuring the reliability of the optimization method under extreme conditions.

[0066] It should be noted that the large-scale rock mechanics parameter optimization method based on multi-scale test data also includes a parameter optimization database, which stores the matching evaluation factors corresponding to the characteristic curves of each rock mass test scale at each rock mass simulation engineering scale obtained by analyzing historical data, the corresponding rock mass test scale matching evaluation thresholds, the path type-parameter constraint rule library, the path offset influence factor corresponding to the unit stress data offset coefficient, the path offset influence factor corresponding to the unit displacement data offset coefficient, the path offset influence factor corresponding to the unit internal pore pressure data offset coefficient, the path offset influence factor corresponding to the unit energy data offset coefficient, and the path offset feature threshold.

[0067] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0068] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A method for optimizing large-scale rock mass mechanical parameters based on multi-scale experimental data, characterized in that, include: Data from the test process at multiple rock mass test scales were collected. Energy evolution analysis was performed on the test process data to obtain energy path characteristic data at each rock mass test scale. Anomaly correlation analysis was then performed on the energy path characteristic data at each rock mass test scale to obtain a multi-scale anomaly path database. The test process data of the multi-rock mass test scale includes mechanical response data, displacement data, internal pore pressure data and energy data of each rock mass test scale. The energy path characteristic data includes mechanical response curves obtained based on the mechanical response data, displacement response curves obtained based on the displacement data, internal pore pressure change curves obtained based on the internal pore pressure data, and energy dissipation curves obtained based on the energy data. The method for performing anomaly correlation analysis on energy path characteristic data at various rock mass test scales is as follows: Energy path characteristic data of each rock mass test scale were extracted, and the abnormal path types of each rock mass test scale were analyzed. Based on the energy path characteristic data of each rock mass test scale and the corresponding abnormal path types, a multi-scale abnormal path database was constructed. Energy path characteristic data at the rock mass simulation engineering scale are collected. Combined with a multi-scale anomaly path database, multi-feature correlation analysis is performed on the energy path characteristic data at the rock mass simulation engineering scale to obtain matching evaluation values ​​for each rock mass test scale at the rock mass simulation engineering scale. The matching evaluation values ​​for each rock mass test scale at the rock mass simulation engineering scale are used to quantify the degree of consistency of energy paths at different rock mass test scales in the rock mass simulation engineering scale. Deviation discriminant analysis is performed on the matching evaluation values ​​of each rock mass test scale at the rock mass simulation engineering scale to obtain the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale. Based on the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale, the energy path matching information of the rock mass simulation engineering scale is obtained. Then, the path correction constraint is applied to the mechanical parameter optimization process at the rock mass simulation engineering scale. The matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale include qualified energy path matching and unqualified energy path matching. The energy path matching information is used to characterize the matching relationship between the energy path feature data of each rock mass test scale and the energy path feature data of the rock mass simulation engineering scale. The path correction constraint is used to constrain the mechanical parameter optimization process at the rock mass simulation engineering scale. Path offset analysis is performed on the path correction constraint process to obtain the path offset characteristic value at the rock mass simulation engineering scale. The associated rock mass mechanical parameters are then updated and optimized based on the path offset characteristic value at the rock mass simulation engineering scale. The path offset characteristic value is used to characterize the degree of offset of the energy path characteristic data at the rock mass simulation engineering scale relative to the energy path characteristic data at the rock mass test scale.

2. The method for optimizing large-scale rock mass mechanical parameters based on multi-scale experimental data according to claim 1, characterized in that, The method for performing energy evolution analysis on the experimental process data is as follows: Stress-deformation evolution analysis was performed on the mechanical response data of each rock mass test scale to obtain the mechanical response curves of each rock mass test scale; displacement analysis was performed on the displacement data of each rock mass test scale to obtain the displacement response curves of each rock mass test scale; pore pressure variation analysis was performed on the internal pore pressure data of each rock mass test scale to obtain the internal pore pressure variation curves of each rock mass test scale; and energy dissipation integration analysis was performed on the energy data of each rock mass test scale to obtain the energy dissipation curves of each rock mass test scale. Based on the mechanical response curves, displacement response curves, internal pore pressure change curves, and energy dissipation curves of each rock mass test scale, energy path characteristic data for each rock mass test scale are constructed.

3. The method for optimizing large-scale rock mass mechanical parameters based on multi-scale experimental data according to claim 1, characterized in that, The energy path characteristic data at the engineering scale of the rock mass simulation includes the mechanical response curve, displacement response curve, internal pore pressure change curve, and energy dissipation curve at the engineering scale of the rock mass simulation.

4. The method for optimizing large-scale rock mass mechanical parameters based on multi-scale experimental data according to claim 3, characterized in that, The method for performing multi-feature correlation analysis on energy path characteristic data at the engineering scale of rock mass simulation is as follows: By performing multi-feature correlation analysis on the energy path characteristic data of the rock mass simulation engineering scale and the energy path characteristic data of all rock mass test scales in the multi-scale anomaly path database, the matching evaluation value of each rock mass test scale and the corresponding anomaly path type of the rock mass simulation engineering scale are obtained.

5. The method for optimizing large-scale rock mass mechanical parameters based on multi-scale test data according to claim 4, characterized in that, The method for deviation discrimination analysis of the matching evaluation values ​​of various rock mass test scales at the simulated engineering scale is as follows: The matching evaluation value of each rock mass test scale at the rock mass simulation engineering scale is compared with the preset matching evaluation threshold of the corresponding rock mass test scale. If the matching evaluation value of the corresponding rock mass test scale at the rock mass simulation engineering scale is higher than the preset matching evaluation threshold, the matching evaluation result of the corresponding rock mass test scale at the rock mass simulation engineering scale is marked as qualified for energy path matching; otherwise, the matching evaluation result of the corresponding rock mass test scale at the rock mass simulation engineering scale is marked as unqualified for energy path matching.

6. The method for optimizing large-scale rock mass mechanical parameters based on multi-scale experimental data according to claim 5, characterized in that, The method for obtaining energy path matching information at the engineering scale of rock mass simulation is as follows: Extract the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale and the corresponding abnormal path types. If there are two or more abnormal path types corresponding to the matching evaluation results of each rock mass test scale at the rock mass simulation engineering scale that are qualified for energy path matching, then mark the energy path matching information of the rock mass simulation engineering scale as cross-scale matching anomaly; otherwise, do not mark it.

7. The method for optimizing large-scale rock mass mechanical parameters based on multi-scale experimental data according to claim 6, characterized in that, The method for path correction constraints in the optimization process of mechanical parameters at the engineering scale of rock mass simulation is as follows: Extract energy path matching information at the rock mass simulation engineering scale. If the energy path matching information at the rock mass simulation engineering scale is an anomaly across scales, sort the corresponding rock mass test scale matching evaluation values ​​in descending order. Mark the anomaly path type corresponding to the highest rock mass test scale matching evaluation value as the dominant energy path type at that rock mass simulation engineering scale, and then apply path correction constraints to the mechanical parameter optimization process at the rock mass simulation engineering scale.

8. The method for optimizing large-scale rock mass mechanical parameters based on multi-scale experimental data according to claim 1, characterized in that, The method for performing path offset analysis on the path correction constraint process is as follows: Path offset analysis is performed on the path correction constraint process to obtain path offset characteristic data at the rock mass simulation engineering scale. The path offset characteristic data at the rock mass simulation engineering scale includes stress data offset coefficient, displacement data offset coefficient, internal pore pressure data offset coefficient, and energy data offset coefficient at the rock mass simulation engineering scale. A comprehensive analysis of path offset characteristic data at the engineering scale of rock mass simulation is conducted to obtain path offset characteristic values ​​at the engineering scale of rock mass simulation. These path offset characteristic values ​​at the engineering scale of rock mass simulation are used to quantify the severity of matching deviations between different energy paths during parameter optimization.

9. The method for optimizing large-scale rock mass mechanical parameters based on multi-scale test data according to claim 8, characterized in that, The method for updating and optimizing the associated rock mass mechanical parameters based on the path offset characteristic values ​​at the rock mass simulation engineering scale is as follows: The path offset characteristic value at the rock mass simulation engineering scale is compared with the preset path offset characteristic threshold. If the path offset characteristic value at the rock mass simulation engineering scale is higher than the preset path offset characteristic threshold, the path offset result at the rock mass simulation engineering scale is marked as a path anomaly. Otherwise, the path offset result at the rock mass simulation engineering scale is marked as a path normal. If the path offset result at the rock mass simulation engineering scale is a path anomaly, the associated rock mass mechanical parameters are updated and optimized.

Images