Large underground cavern group crustal stress field inversion method under complex geological conditions

Abstract

The invention relates to a large underground cavern group crustal stress field inversion method under a complex geological condition, and provides a method for inversing a crustal stress field of a large underground cavern group by taking a river valley center line or a ridge line as an inversion numerical model boundary to solve the problems that underground cavern group crustal stress measured data is discrete and difficult to accurately predict under the background of deep river valley terrain and multi-fault development. And selecting actual measurement points which are far away from the fault and are in three-dimensional distribution. The method comprises the following steps: acquiring a displacement boundary and a gravitational acceleration magnitude of a three-dimensional numerical model through limited measured data by adopting stepwise multiple linear regression and evolutionary neural network joint inversion, positively loading the displacement boundary and the gravitational acceleration magnitude to a model boundary to obtain a regional crustal stress field, and extracting measured point data for preliminary verification; and simulating caving excavation, and analyzing whether the stress concentration area is consistent with the observed stress type damage position or not. The method integrates the advantages of the two methods, more optimal solutions can be selected, the problem that precision and calculated amount are difficult to consider in stress field prediction of the complex structure motion area is solved, and the method is suitable for crustal stress field prediction of the complex structure area.

Description

Technical Field

[0001] This invention relates to a method for predicting the geostress field of underground cavern groups, and particularly to a method for inverting the geostress field of large underground cavern groups under complex geological conditions. Background Technology

[0002] Multiple geological tectonic movements refer to the combined influence of various tectonic movements such as compression and shear on the formation of regional stress fields, in addition to the influence of self-weight stress.

[0003] The initial geostress field of a region serves as the foundation for analyzing the stability of the surrounding rock during excavation in engineering projects such as pumped storage power station underground cavern groups and ultra-deep underground laboratories. It plays a crucial role in simulating construction methods, designing support structures, and implementing disaster early warning and prevention measures. The formation of the geostress field in underground engineering areas is typically influenced by multiple factors, including geological tectonic movements, topography, faults, and fold structures. Existing inversion methods often fail to simultaneously meet the computational requirements of engineering projects. Therefore, it is essential to combine existing inversion algorithms with information such as historical tectonic features, statistical data from existing measurements, and damage characteristics to more accurately determine the distribution patterns of the geostress field in underground cavern groups under the influence of multiple factors. Summary of the Invention

[0004] The technical problem solved by this invention is to provide a method for predicting the geostress field of underground cavern groups under the combined influence of deep valley topography and multiple faults, which improves the problem that existing inversion methods cannot simultaneously meet the requirements of geostress field prediction accuracy and engineering calculation volume in complex tectonic movement areas.

[0005] The technical solution adopted by this invention to solve the above-mentioned technical problems is a method for predicting the in-situ stress field of underground cavern groups in complex tectonic movement zones, comprising the following steps: S01. Investigate the characteristics of historical geological tectonic movements in the region, the direction of principal stress in recent dynamic geological structures, and unloading effects, and roughly determine the composition of tectonic movements and macroscopic principal stress vector characteristics of the region.

[0006] S02. Based on the characteristic patterns of the effective measured ground stress data in the region, determine the stress field distribution characteristics. Simultaneously, compare this data with the regional macroscopic principal stress vector characteristics in S01 to analyze the reliability of the measured ground stress data and the inferred regional ground stress field characteristics. The results of both methods corroborate each other. The effective measured ground stress data in the region are field-measured stress data. First, data with significant errors due to operation or instrumentation are eliminated. Then, considering the relative positional relationship between the regional topography and the ground stress measuring points, and the stress distribution patterns, ground stress data that do not meet the criteria are also eliminated.

[0007] S03. Establish a three-dimensional geological model including regional geological structural features and underground cavern groups, and perform three-dimensional geostress field inversion using the measured geostress data selected in S02. Extract the inverted stress data at the measured point locations, and compare the vector characteristics of the inverted stress data at the measured point locations with those of the measured stress data to verify whether they are within the design error range, thus validating the reliability of the inversion results. The measured points suitable for geostress field inversion should be selected according to the following principles: the distance between the measured points and the structural surface should not be too close; the distribution of the measured points should not be too concentrated. The joint inversion algorithm obtains reasonable tectonic movement factors and their benchmark values ​​through stepwise multivariate regression, and then obtains a better initial geostress field inversion result through neural network inversion.

[0008] S04. Based on the initial geostress field obtained, simulate the excavation process of the exploratory tunnel, analyze the location of the stress concentration zone in the surrounding rock after the tunnel excavation, and compare whether the location of the stress concentration zone corresponds to the location of stress-type failure that occurred during the on-site exploratory tunnel excavation, further verifying the reliability of the inversion results. Beneficial effects of this invention: This invention provides a geostress field inversion method for underground cavern groups under complex geological conditions. This method mainly includes determining the direction of regional macroscopic principal stress, analyzing measured geostress data, establishing a three-dimensional geological model, performing a combined inversion of stepwise multiple linear regression and evolutionary neural network, and verifying the results based on the stress-type failure characteristics of the cavern surrounding rock. Starting from historical tectonic movements, this method constrains the multiple regression factors and their regression coefficients, improving the problem of excessive shear stress deviation. Through multi-source constraints and verification, such as historical geological structure analysis, measured geostress analysis, and stress-type failure characteristics of cavern groups, the accuracy of the geostress field inversion results is improved. Attached Figure Description

[0009] Figure 1 This invention relates to a method for inverting the geostress field of large underground cavern groups in complex geological structures. Figure 2 The inversion operation method of this invention is a BP neural network; Detailed Implementation

[0010] This invention provides a method for inverting the geostress field of large underground cavern groups in complex geological structures, such as... Figure 1 As shown, it includes the following steps: S01. Investigate the characteristics of historical geological tectonic movements in the region, the direction of principal stress in recent dynamic geological structures, and unloading effects, and roughly determine the composition of tectonic movements and macroscopic principal stress vector characteristics of the region.

[0011] Specifically, the direction of the regional horizontal principal stress is affected by plate interactions, while the dip angle of the principal stress is influenced by multiple factors such as topography, unloading of the upper rock mass, and tectonic movement.

[0012] S02. Based on the characteristic patterns of the effective measured stress data from the region, determine the stress field distribution characteristics. Simultaneously, compare this data with the regional macroscopic principal stress vector characteristics in S01 to analyze the reliability of the measured stress data and the inferred regional stress field characteristics. The results of both methods corroborate each other. The effective measured stress data for the region are obtained from on-site measured stress data. First, data with significant errors due to operation or instrumentation are eliminated. Then, considering the relative positional relationship between the regional topography and the stress measurement points, and the stress distribution patterns, data that do not meet the criteria are also eliminated.

[0013] Specifically, the following principles should be followed when selecting the measuring points applicable to geostress field inversion: (1) Keep away from faults: The shortest distance between the measuring point and a Class II fault (such as the main fault) should be ≥20m, and the shortest distance between the measuring point and a Class III fault (such as the secondary fault) should be ≥10m, so as to avoid the interference of stress change near the fault on the inversion results; (2) Uniform spatial distribution: The measuring points need to cover different horizontal burial depths (such as shallow 200-300m, deep 400-500m) and different orientations (such as upstream and downstream of the cavern group, left and right banks), and the distance between any two measuring points should not be less than 50m, to avoid the problem of collinearity of regression factors caused by the concentration of measuring points; (3) Data reliability screening: Remove measurement points with a deviation of more than 30% in the ratio of vertical stress to self-weight stress (K=σz / γh) (such as K<0.7 or K>1.3), as well as measurement points with a deviation of more than 20° between the principal stress direction and the macroscopic principal stress direction of the region, to ensure the representativeness of the data.

[0014] S03. Establish a three-dimensional geological model that includes regional geological structural features and underground cavern groups, following the modeling principles: (1) Model boundary selection: For deeply incised valley terrain, the model boundary should be arranged along the valley centerline or ridgeline to accurately reflect the unloading or concentration effect of the terrain on the regional stress field; if the valley centerline is too close to the cave group (less than 3 times the cave height), a straight line parallel to the valley direction should be selected as the boundary to ensure that the distance between the model boundary and the nearest sidewall of the cave group is 3-5 times the cave height, so as to avoid boundary effect interference. (2) Geological structure includes: The model should include Class II faults in the region and Class III faults that cross adjacent cavern groups. Class IV and lower small faults can be simplified to continuous media; (3) Topographic treatment: The original topographic undulations are preserved on the ground surface. The ultra-deep buried cavern group (buried depth > 800m) can be simplified to a horizontal ground surface, but the influence of topography needs to be corrected by adjusting the gravity acceleration.

[0015] Three-dimensional geostress field inversion was performed using the measured geostress data selected in S02. The inverted stress data at the measured point locations were extracted, and the vector characteristics of the inverted stress data at the measured point locations were compared with those of the measured stress data to verify the reliability of the inversion results. The measured points suitable for geostress field inversion should be selected according to the following principles: the distance between the measured points and the structural surface should not be too close; the distribution of the measured points should not be too concentrated. The joint inversion algorithm obtains reasonable tectonic motion factors and their benchmark values ​​through stepwise multivariate regression, and then obtains a better initial geostress field inversion result through neural network inversion.

[0016] Specifically, the joint inversion algorithm includes the following sub-steps: (1) Stepwise multiple regression stage: 1. Screening of inversion factors: Only the main factors that have a significant impact on the regional stress field (such as self-weight, horizontal compression in the X direction, horizontal compression in the Y direction, and shear tectonic movement in the XY plane) are selected, with a total number of factors ≤ 6, to avoid collinearity problems caused by too many factors; 2. Regression coefficient constraint: The regression coefficient must be ≥0, and factors with negative or insignificant coefficients are eliminated; Benchmark value determination: The regression coefficients and weights of each factor are obtained by least squares fitting, which serve as the benchmark for neural network sample design.

[0017] (2) Evolutionary neural network inversion stage: 1. Sample construction: Using the baseline value of stepwise regression as the center, expand at ±10% and ±20% levels to construct a mixed orthogonal sample (e.g., 3 factors × 3 levels = 27 groups of samples). 2. Network training: Input the boundary conditions of the sample (displacement, gravitational acceleration), output the stress components of the measurement point, optimize the network parameters through error backpropagation, and train until the mean square error is ≤5MPa; 3. Optimal Solution Extraction: Measured stress data is input into the trained network, which outputs the optimal combination of boundary conditions. This combination is then loaded into a 3D model to obtain the initial geostress field of the region, solving the problems of unclear sample benchmarks and excessively large solution spaces inherent in traditional neural networks. For example... Figure 2 S04. Based on the initial geostress field obtained, simulate the excavation process of the exploratory tunnel, analyze the location of the stress concentration zone in the surrounding rock after the exploratory tunnel is excavated, and compare whether the location of the stress concentration zone corresponds to the location of stress-type failure that occurred during the on-site exploratory tunnel excavation, so as to further verify the reliability of the inversion results.

[0018] Specifically, the verification step of comparing the vector characteristics of the inverted stress data and the measured stress data at the measurement points in S03 to determine whether they are within the design error range includes multi-dimensional verification to ensure that the inversion results conform to regional geological patterns and engineering realities: (1) Comparison and verification between calculation and actual measurement: Extract the stress components (σxx, σyy, σzz, τxy, etc.) and principal stresses (σ1, σ2, σ3) of the inverted measurement points, and compare them with the measured data. The stress component error is required to be ≤20%, the principal stress value error is required to be ≤15%, the azimuth deviation is required to be ≤10°, and the dip angle deviation is required to be ≤5°. (2) Verification of stress-induced failure in the exploratory tunnel: Simulate the exploratory tunnel excavation process, analyze the stress concentration zone of the surrounding rock of the exploratory tunnel, and compare the locations of stress-induced failures such as rock bursts and spalling during exploratory tunnel excavation. The overlap between the stress concentration zone and the failure location is required to be ≥80%. (3) Verification of measured data of disturbance stress: Collect stress monitoring data of the disturbance zone after the excavation of the cavern, and compare it with the inverted distribution of the disturbance stress field. The error of the disturbance stress increment is required to be ≤25%. (4) Verification of regional valley stress field characteristics: The inversion results must conform to the stress field law of deeply incised valley areas, that is, the principal stress increases linearly with vertical burial depth, the principal stress direction gently slopes towards the valley, and the deviation between the location of the maximum principal stress and the direction of the regional tectonic principal compressive stress is ≤15°. The stress-type failure refers to phenomena such as rock bursts and spalling that occur after the excavation of the cavern, and its characteristics include information such as the depth, range, and location of the failure.

Claims

1. A method for predicting the geostress field of underground cavern groups in complex tectonic movement zones, characterized in that, Includes the following steps: S01. Investigate the characteristics of historical geological tectonic movements in the region, the direction of principal stress in recent dynamic geological tectonic movements, and unloading effects, and roughly determine the composition of tectonic movements and macroscopic principal stress vector characteristics of the region. S02. Based on the characteristics of the effective measured stress data of the region, determine the stress field distribution characteristics. At the same time, compare with the characteristics of the macroscopic principal stress vector of the region in S01 to analyze the reliability of the measured stress data and the inferred regional stress field characteristics. The results of the two are mutually corroborated. The effective measured stress data of the region is taken from the field measured stress data. First, data with large errors caused by operation, instruments, etc. are eliminated. Then, considering the relative positional relationship between the regional topography and the stress measuring points, the stress distribution law, etc., the stress data that do not meet the requirements are eliminated. S03. Establish a three-dimensional geological calculation model including regional geological structural features and underground cavern groups, and use the measured geostress data selected in S02 to perform three-dimensional geostress field inversion. Extract the inverted stress data at the measured point locations, and compare the vector characteristics of the inverted stress data at the measured point locations with those of the measured stress data to verify whether they are within the design error range, thus verifying the reliability of the inversion results. The measured points suitable for geostress field inversion should be selected according to the following principles: the distance between the measured points and the structural surface should not be too close; the distribution of the measured points should not be too concentrated. The joint inversion algorithm obtains reasonable tectonic movement factors and their benchmark values ​​through stepwise multivariate regression, and then obtains better initial geostress field inversion results through neural network inversion. S04. Based on the obtained initial geostress field distribution, simulate the excavation process of the exploratory tunnel, analyze the location of the stress concentration zone in the surrounding rock after the exploratory tunnel is excavated, and compare whether the location of the stress concentration zone corresponds to the location of stress-type failure that occurred during the on-site exploratory tunnel excavation, so as to further verify the reliability of the inversion results.

2. The method for predicting the geostress field of underground cavern groups in complex tectonic movement zones according to claim 1, characterized in that, The method mainly includes determining the direction of regional macroscopic principal stress, analyzing measured geostress data, establishing a three-dimensional geological model, performing joint inversion of stepwise multiple linear regression and evolutionary neural network, and verifying the results based on the stress-type failure characteristics of the surrounding rock in the tunnel exploration.

3. The method for predicting the geostress field of underground cavern groups in complex tectonic movement zones according to claim 1, characterized in that, The analysis of the historical tectonic movements in the research area clarified the main factors considered in the multiple regression analysis. The weights of each factor were obtained through stepwise multiple linear regression, and at the same time, sample benchmark values ​​at different levels were provided for the inversion of evolutionary neural networks.

4. The method for predicting the geostress field of underground cavern groups in complex tectonic movement zones according to claim 1, characterized in that, The method of constructing a nonlinear relationship between ground stress and model boundary using an evolutionary neural network improves the problem of excessive shear stress deviation caused by stepwise multiple linear regression.

5. The method for predicting the geostress field of underground cavern groups in complex tectonic movement zones according to claim 1, characterized in that, The paper proposes selection principles for inversion based on measured geostress data, and constrains the selection of inversion points based on factors such as the distance between measuring points and faults of different grades, and the variation law of measuring point stress with burial depth.

6. The method for predicting the geostress field of underground cavern groups in complex tectonic movement zones according to claim 1, characterized in that, Regarding the verification of inversion results, the stress-type failure characteristics of the surrounding rock in exploratory tunnels or excavated small caverns are considered, ensuring the rationality of the in-situ stress inversion results.

Images