Multi-source data fusion three-dimensional crustal stress field dynamic inversion method and system based on intelligent optimization algorithm

Abstract

The invention provides a multi-source data fusion three-dimensional crustal stress field dynamic inversion method and system based on an intelligent optimization algorithm, and the method comprises the steps: collecting multi-source geological data containing crustal stress components and rock mass mechanical parameters, and constructing a three-dimensional geomechanical model; setting model boundary gravity and multi-direction tectonic stress load working conditions, calculating single-working-condition stress distribution and extracting stress calculation values of measuring points; constructing an inversion objective function based on the stress calculation value and the measured value; an improved intelligent optimization algorithm is adopted, and an optimal boundary load combination coefficient is inversed by taking a target function as a fitness function; and linearly superposing the optimal coefficient and the corresponding working condition stress distribution to synthesize a global three-dimensional initial crustal stress field. The problems that a traditional method is low in inversion efficiency, prone to falling into local optimum and insufficient in stress component precision under complex geological conditions are solved, and rapid and accurate acquisition of the three-dimensional crustal stress field of the engineering area is achieved.

Description

Technical Field

[0001] This invention relates to the fields of geotechnical engineering, engineering geology and computational intelligence, specifically a method and system for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms. Background Technology

[0002] The initial geostress field of a rock mass refers to the stress state existing within the rock mass under natural conditions. It is a fundamental environmental load that must be considered in the design and construction of all underground engineering projects (such as tunnels, mines, underground power plants, and energy storage facilities). Accurately understanding the distribution law of the initial geostress field in the engineering area is of irreplaceable importance for evaluating the stability of the surrounding rock, predicting geological hazards such as rock bursts and large deformations, and optimizing engineering design to achieve the goals of safety and economy.

[0003] Currently, methods for obtaining geostress fields mainly fall into two categories: direct field measurement and numerical simulation inversion analysis. Direct field measurement methods (such as hydraulic fracturing and stress relief methods) can directly obtain the stress state at the measurement point, providing intuitive and reliable data, and are the cornerstone of geostress research. However, these methods generally have inherent limitations, such as high cost, long testing cycles, high requirements for site conditions, and the ability to obtain only spatially discrete "point" information. For engineering areas with complex geological conditions and wide coverage, relying solely on a limited number of measurement points is insufficient to comprehensively and accurately reflect the continuous distribution and heterogeneity of the stress field in three-dimensional space. Numerical simulation inversion analysis, as an effective supplement and extension to field measurements, establishes a numerical model (such as a finite element model or finite difference model) that reflects the generalized geological characteristics of the engineering area. It uses limited field-measured stress data as target constraints and infers the stress field distribution throughout the study area through mathematical physics inversion theory. This method can provide full-field, continuous three-dimensional stress information and has therefore been widely used in engineering practice.

[0004] Traditional numerical inversion methods are mostly based on linear optimization theories such as the least squares method. The basic process is as follows: First, a series of basic boundary load conditions reflecting the causes of geostress (such as self-weight, horizontal tectonic forces, etc.) are set. Then, a set of load combination coefficients is optimized through the least squares criterion so that the sum of squared errors between the stress at the measurement points calculated by the numerical model and the stress measured in the field is minimized under this combination of coefficients.

[0005] However, as engineering practices extend into deeper and more complex geological environments, the limitations of traditional inversion methods have become increasingly apparent, mainly in the following three aspects:

[0006] First, the inversion algorithm lacks global optimization capability and is prone to getting trapped in local optima. Geostress inversion is essentially a complex nonlinear optimization problem. The relationship between its objective function (stress error) and the parameters to be determined (load coefficient) is often highly complex, with multiple local extrema. Traditional local search algorithms, such as the least squares method, heavily rely on the selection of initial values, easily getting trapped in local optima and failing to find the global optimum. This may result in the inverted stress field, while fitting reasonably well at the measurement points, being severely distorted in its overall distribution, failing to accurately reflect the macroscopic laws of the regional stress field, and misleading engineering decisions.

[0007] Second, the model fails to fully utilize multi-source geological information, resulting in poor geological fit. Traditional inversion methods typically use only measured stress data as the sole fitting target, neglecting the vast amount of other geological information accumulated during engineering surveys, such as high-precision topographic data, the spatial distribution of geological structures like faults and folds, zoning of different lithologies, and wave velocity or resistivity anomalies revealed by geophysical exploration. This multi-source data contains rich indirect information closely related to the distribution of the geostress field. Ignoring this information is tantamount to discarding valuable prior knowledge, leading to a lack of sufficient geological constraints in the inversion model, and casting doubt on its physical rationality and extrapolation prediction capabilities.

[0008] Third, the inversion model is static and lacks a dynamic updating mechanism. Large-scale engineering projects have long construction cycles and are usually carried out in stages. During excavation and tunneling, new geological phenomena (such as unknown small faults and dikes) are constantly revealed, and new monitoring or testing data (such as convergence deformation and supplementary in-situ stress measurements) are obtained. Traditional inversion processes are one-off, and the initial stress field model is static, making it impossible to effectively integrate these dynamically acquired new information into the model for correction and updating. This leads to designs based on initial inversion results that are difficult to adapt to changes in actual geological conditions revealed during construction, making it impossible to realize the modern engineering management concept of "dynamic design and information-based construction."

[0009] Therefore, developing a three-dimensional geostress field inversion method and system that can overcome local optima, effectively integrate multi-source data, and has dynamic updating capabilities has become an urgent technical need in the field of geotechnical engineering. Summary of the Invention

[0010] The purpose of this invention is to address the shortcomings and deficiencies of the prior art by providing a more advanced, reliable, and practical method and system for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms.

[0011] The primary objective of this invention is to fundamentally overcome the technical challenge of traditional inversion methods being prone to local optima by introducing an intelligent optimization algorithm with powerful global search capabilities, thereby ensuring that the inversion results can closely approximate the true global optimal stress state and significantly improving the reliability and accuracy of the inversion results.

[0012] Another objective of this invention is to construct a comprehensive inversion objective function that integrates multi-source geological information, combining stress data fitting with geological law constraints and mechanical principle constraints, so that the inversion process does not rely solely on sparse measurement point data, but is based on a richer and more solid geological and physical foundation, thereby significantly improving the geological rationality and extrapolation prediction capability of the inversion model.

[0013] Another objective of this invention is to design a complete dynamic feedback update mechanism that enables the geostress field model to evolve and continuously optimize itself as the project progresses, achieving a leap from "static snapshot" to "dynamic movie," providing real-time and accurate geological environment information support for the entire life cycle of the project, and ultimately serving the goals of safe, economical, and efficient construction.

[0014] To achieve the above objectives, the present invention adopts the following technical solution:

[0015] A dynamic inversion method for three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms is characterized by the following steps:

[0016] Step S1: Multi-source geological data acquisition and 3D geomechanical model construction: Through field testing and geological exploration, multi-source geological data of the engineering area are acquired. The multi-source geological data includes the 3D stress components of the measured stress points, topography, geological structure, lithology distribution, and rock mass mechanical parameters; a 3D geomechanical model of the engineering area is constructed based on the multi-source geological data.

[0017] Step S2: Boundary load case setting and single-case stress field calculation: Based on the boundary of the three-dimensional geomechanical model constructed in step S1, the benchmark calculation case is constructed. The benchmark calculation case includes each single load case corresponding to the applied gravity load and the tectonic stress load in multiple directions. The stress distribution of the entire model under each single load case is calculated, and the stress calculation value at the measuring point is extracted.

[0018] Step S3: Construction of the inversion objective function: Based on the stress calculation value extracted in step S2 and the measured value of the three-dimensional stress components in step S1, an inversion objective function is constructed. The inversion objective function is the weighted sum of squares of the difference between the measured stress value and the stress calculation value at the measurement point, and different weighting coefficients are introduced for the normal stress term and the shear stress term to reflect the difference in their data reliability.

[0019] Step S4: Global Inversion of Intelligent Optimization Algorithm: Using an improved intelligent optimization algorithm, the optimal boundary load combination coefficients are determined by inversion using the inversion objective function established in Step S3 as the fitness function.

[0020] Step S5: Synthesis of three-dimensional initial stress field: The optimal boundary load combination coefficients obtained in step S4 are linearly superimposed with the stress distribution of the entire model under the corresponding load condition in step S2 to synthesize and calculate the three-dimensional initial geostress field of the entire domain.

[0021] Furthermore, it also includes:

[0022] Step S6: Dynamic Update of Stress Field: During the engineering construction process, when new multi-source geological data or stress monitoring data is obtained, the update process is triggered; the new multi-source geological information includes newly revealed geological structures, lithological distributions, or rock mass mechanical parameters; the stress monitoring data includes supplementary measured geostress data; the new multi-source geological data or stress monitoring data are incorporated into the original database, the updated data is used to reconstruct or correct the inversion objective function, and the boundary load combination coefficients are rapidly inverted based on the improved intelligent optimization algorithm. The optimal coefficients obtained from the inversion are used to dynamically correct the three-dimensional initial geostress field, generating an updated three-dimensional geostress field.

[0023] Furthermore, the three-dimensional stress components in step S1 are obtained by the hollow inclusion stress relief method or the hydraulic fracturing method, and the rock mass mechanical parameters are obtained by the bearing plate load test, including the rock mass elastic modulus.

[0024] Furthermore, the structural stress loads in multiple directions in step S2 include: a unit compressive load along the horizontal primary structural direction, a unit compressive load along the horizontal secondary structural direction, and a unit shear load at the bottom.

[0025] Furthermore, the inversion objective function model in step S3 is as follows:

[0026] (1);

[0027] in, The objective function value is retrieved, where m is the total number of measurement points. , These are the measured values ​​of normal stress and shear stress at the k-th measuring point, respectively. , These are the calculated values ​​of the corresponding normal stress and shear stress, respectively. and These are the maximum values ​​of all measured normal and shear stresses, used for normalization. is the normal stress weighting coefficient, and its value range is [0,1].

[0028] Furthermore, the improved intelligent optimization algorithm in step S4 is an improved particle swarm optimization algorithm, the improvements of which include: adopting an adaptive inertia weight that decreases linearly with the number of iterations, and introducing an asynchronously changing learning factor, focusing on individual learning in the early stage of iteration to maintain population diversity, and focusing on social learning in the later stage of iteration to promote convergence to the global optimum.

[0029] Furthermore, step S4 also includes a convergence judgment mechanism: when the change in the fitness function value within a continuously set number of iterations is less than the preset tolerance, the optimization calculation process is automatically terminated to improve the inversion efficiency.

[0030] A multi-source data fusion dynamic inversion system for three-dimensional geostress field based on intelligent optimization algorithms includes:

[0031] The multi-source geological data acquisition and three-dimensional geomechanical model construction module is used to acquire multi-source geological data of the engineering area through field testing and geological exploration. The multi-source geological data includes three-dimensional stress components, topography, geological structure, lithology distribution and rock mass mechanical parameters of the measured stress points; and constructs a three-dimensional geomechanical model of the engineering area based on the multi-source geological data.

[0032] The boundary load case setting and single-case stress field calculation module is used to construct the benchmark calculation case based on the boundary of the constructed three-dimensional geomechanical model. The benchmark calculation case includes each single load case corresponding to the applied gravity load and the tectonic stress load in multiple directions. It calculates the stress distribution of the entire model under each single load case and extracts the stress calculation value at the measuring point.

[0033] The inversion objective function construction module is used to construct an inversion objective function based on the extracted calculated stress values ​​and the measured values ​​of the three-dimensional stress components. The inversion objective function is the weighted sum of squares of the difference between the measured stress value and the calculated stress value at the measurement point, and different weight coefficients are introduced for the normal stress term and the shear stress term to reflect the difference in their data reliability.

[0034] The intelligent optimization algorithm global inversion module is used to determine the optimal boundary load combination coefficients by using an improved intelligent optimization algorithm with the established inversion objective function as the fitness function.

[0035] The three-dimensional initial stress field synthesis module is used to linearly superimpose the optimal boundary load combination coefficients with the stress distribution of the entire model under the corresponding load conditions to synthesize and calculate the three-dimensional initial geostress field of the entire domain.

[0036] Furthermore, it also includes: a stress field dynamic update module, used to trigger the update process when new multi-source geological data or stress monitoring data is obtained during the engineering construction process; the new multi-source geological information includes newly revealed geological structures, lithological distributions, or rock mass mechanical parameters; the stress monitoring data includes supplementary measured geostress data; the new multi-source geological data or stress monitoring data are incorporated into the original database, the updated data is used to reconstruct or correct the inversion objective function, and the boundary load combination coefficients are quickly inverted based on an improved intelligent optimization algorithm. The optimal coefficients obtained from the inversion are used to dynamically correct the three-dimensional initial geostress field to generate an updated three-dimensional geostress field.

[0037] Furthermore, the intelligent inversion analysis module has a built-in parameter adaptive unit and a convergence judgment unit, which automatically terminates the optimization calculation process when the change in the fitness function value within a continuously set number of iterations is less than the preset tolerance, so as to improve the inversion efficiency.

[0038] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0039] 1. The inversion results are globally optimal, and the accuracy and reliability are significantly improved: By adopting an improved intelligent optimization algorithm for global search, the problem of solution distortion caused by getting trapped in local optima in traditional methods is effectively avoided. This makes the stress field obtained by inversion more realistically reflect the macroscopic stress distribution law in complex geological environments, and provides a more reliable data foundation for engineering decision-making.

[0040] 2. Deep integration of multi-source information, solid physical foundation of the model: Breaking through the limitations of traditional inversion relying solely on stress data, it creatively incorporates multi-source information such as topography, geological structure, and geophysics as constraints into the inversion process. This makes the inversion model no longer a purely mathematical fit, but a numerical reproduction with clear geological and physical significance, greatly enhancing the geological rationality of the model and its predictive ability in unexplored areas.

[0041] 3. Possesses dynamic evolution capabilities, serving the entire engineering construction process: An innovative dynamic stress field update mechanism is designed, enabling the inversion model to "grow" alongside the engineering progress. It continuously absorbs and utilizes new information revealed during construction, achieving self-correction and improvement of the model. This provides real-time, accurate, and continuous geological environment support for dynamic design, risk warning, and information-based construction, representing the development direction of geotechnical engineering analysis from static to dynamic.

[0042] 4. High degree of automation and integration, significantly improving analysis efficiency: It integrates data management, model building, numerical calculation, intelligent inversion, visualization and dynamic updates into one, forming a complete analysis pipeline, which greatly reduces manual intervention and repetitive work, and improves the standardization and work efficiency of geostress field analysis. Attached Figure Description

[0043] Figure 1 This is a detailed implementation diagram of a multi-source data fusion dynamic inversion method for three-dimensional geostress field based on intelligent optimization algorithm according to an embodiment of the present invention;

[0044] Figure 2 This is a schematic diagram of a finite difference numerical model according to an embodiment of the present invention;

[0045] Figure 3 This is a schematic diagram of the stress field formation under self-weight load according to an embodiment of the present invention;

[0046] Figure 4 This is a schematic diagram of the stress field formation under a horizontal x-direction load according to an embodiment of the present invention;

[0047] Figure 5 This is a schematic diagram of the stress field formation under a horizontal y-direction load according to an embodiment of the present invention;

[0048] Figure 6 This is a schematic diagram of the stress field formation under shear load according to an embodiment of the present invention;

[0049] Figure 7 This is a schematic diagram of the contour lines of self-weight stress along the factory building according to an embodiment of the present invention;

[0050] Figure 8 This is a schematic diagram of the contour lines of the maximum principal stress along the factory building according to an embodiment of the present invention;

[0051] Figure 9 This is a schematic diagram of the minimum principal stress contour lines along the factory building according to an embodiment of the present invention;

[0052] Figure 10 This is a flowchart of a method for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms, according to an embodiment of the present invention. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0054] Example: Application of initial stress field inversion engineering in the deep underground powerhouse area of ​​a large hydropower station in western Sichuan.

[0055] The project background and geological conditions are introduced as follows:

[0056] This application example focuses on the underground powerhouse area of ​​a large hydropower station in western Sichuan. Located on the eastern edge of the Qinghai-Tibet Plateau, the power station is situated in a highly tectonic compression environment with deep valleys, steep terrain, high measured in-situ stress values, and significant rockburst risks. Accurately obtaining the three-dimensional initial in-situ stress field of this area is the primary task in assessing the stability and rockburst risk of the underground cavern complex.

[0057] The exposed strata in the project area are mainly thick- to medium-thick-bedded metamorphic sandstone interbedded with thin-bedded slate from the Middle Triassic, with the rock mass quality primarily classified as Class II and III. The geological structure is complex, with several faults of varying sizes, including F1, F2, and F3. F1 is the main fault, with an attitude of N35°E / SE∠65° and a fracture zone width of approximately 3-5 meters, significantly controlling the regional stress field distribution. The initial in-situ stress field in the plant area is generally controlled by a combination of self-weight and regional NE-SW trending strong tectonic compressive stress.

[0058] Please see Figure 1 and Figure 10 This invention provides a method for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms, comprising the following steps:

[0059] S1: Multi-source geological data acquisition and 3D geomechanical model construction

[0060] To construct a high-precision inversion model, the system collected and integrated the following multi-source data:

[0061] (1) Topographic and geological structure data: A high-precision digital elevation model with a scale of 1:1000 was used to accurately depict the surface morphology. Through geological mapping, borehole exploration, and adit logging, the spatial orientation and extent of stratigraphic boundaries, major faults (F1, F2, F3), and the distribution of dikes were accurately obtained. All these geological interfaces were digitized and constructed into three-dimensional surfaces.

[0062] (2) Assignment of rock mass mechanical parameters: Based on the field rock mass quality classification (RQD, RMR) and the results of indoor rock mechanics tests, the model rock mass was divided into four mechanical parameter zones: Class II fresh rock mass zone, Class III1 slightly weathered unloaded upper zone, Class III2 slightly weathered unloaded lower zone, and Class IV fault influence zone. The mechanical parameters used in each zone are shown in Table 1.

[0063] Table 1 Recommended values ​​for rock mass mechanical parameters

[0064]

[0065] (3) Measured ground stress data: Three-dimensional ground stress measurements were conducted at seven measuring points within exploration adits (PD1, PD2, PD3) at different elevations within the factory area, using the hollow inclusion stress relief method. The measurement results provide the three principal stress values ​​for each measuring point. , , ) and its direction, and the six stress components ( , , , , , The values ​​of ) are the core basis for inverting the fitting of the objective function.

[0066] Using professional 3D modeling software (such as Rhino), the previously processed terrain surfaces, stratigraphic interfaces, and fault planes were "stitched together" to establish a 3D solid model that can accurately reflect the geological structure of the engineering area.

[0067] Subsequently, the solid model was imported into the finite difference analysis software FLAC3D for mesh generation. To improve computational efficiency and ensure accuracy in critical areas, a non-uniform mesh generation technique was employed: a sparser mesh was used in areas far from the plant, while the mesh was refined around the plant's cavern complex and near faults. The final generated finite difference model is shown below. Figure 2 As shown, the model contains a total of approximately 520,000 hexahedral elements and 650,000 nodes.

[0068] Finally, based on the mechanical parameters in Table 1, different material groups are assigned to the corresponding regions in the model, completing the conversion from a geological model to a computable mechanical model.

[0069] S2: Boundary Load Case Setting and Single-Case Stress Field Calculation. Based on the analysis of the causes of geostress, the main factors affecting the stress field in this project area are determined to be self-weight and tectonic activity. Accordingly, the following four basic boundary load cases are set to simulate the composition of the stress field:

[0070] Condition 1: Self-weight stress field. A vertically downward gravitational acceleration (g = 9.81 m / s²) is applied to the entire model. Through elastic calculations, the stress distribution generated solely by the rock mass's self-weight is obtained, and its σz component contour plot is shown below. Figure 3 As shown, the linear increase in stress with depth is clearly demonstrated.

[0071] Condition 2: Tectonic stress field in the horizontal X direction. A uniformly distributed pressure is applied along the X direction (approximately NE-SW, i.e., the direction of the maximum principal stress in the region) on the model boundary. The tectonic compression in this direction is simulated by applying a certain amount of boundary displacement while maintaining the displacement constraint mode on the boundary. The calculated stress distribution under this condition is as follows: Figure 4As shown, its characteristic is that it produces obvious compressive stress concentration along the loading direction.

[0072] Condition 3: Horizontal Y-direction structural stress field. A uniformly distributed pressure along the Y-direction (approximately NW-SE direction) is applied to the model boundary to simulate lateral constraint perpendicular to the main structural direction. The resulting stress field is as follows: Figure 5 As shown.

[0073] Condition 4: Shear Tectonic Stress Field. To simulate potential shear forces or residual tectonic stresses at depth, a pair of balanced shear loads are applied to the bottom boundary of the model. The stress field distribution dominated by shear stress under this condition is as follows: Figure 6 As shown.

[0074] In FLAC3D, elastic solutions were performed for the four working conditions described above. The six stress components of all elements across the entire model range were calculated and derived for each working condition, forming four basic stress field databases. }, { }, { }, { }

[0075] S3: Construction of the Inversion Objective Function. One of the core ideas of this invention is that the actual initial stress field is a linear superposition of the above-mentioned basic load conditions in a specific ratio. Let the load combination coefficient vector to be determined be... Then the calculated stress tensor of any element is { } can be represented as:

[0076]

[0077] The goal of inversion is to find an optimal set of coefficients. This ensures that the calculated stress is closest to the measured stress at the seven measured points. Therefore, the following inversion objective function is constructed. :

[0078]

[0079] in, , These are the measured values ​​of normal stress and shear stress at the k-th measuring point, respectively. , These are the calculated values ​​of the corresponding normal stress and shear stress, respectively. and These represent the maximum values ​​of all measured normal and shear stresses, used for normalization. In this example, based on the understanding that normal stress data from the engineering area is relatively more reliable, the weighting coefficients are... The value is 0.7.

[0080] S4: Global Inversion Algorithm for Intelligent Optimization: To solve the problem that traditional inversion methods are prone to getting trapped in local optima, this invention uses a particle swarm optimization algorithm for global search.

[0081] (1) Algorithm improvement strategies:

[0082] Nonlinear decreasing inertia weight: A linearly decreasing inertia weight is adopted. The weight is larger in the early stage of the iteration (w_max=0.9), which is beneficial for global exploration; the weight is smaller in the later stage of the iteration (w_min=0.4), which is beneficial for local fine search.

[0083] Asynchronous learning factors: Individual learning factor c1 and social learning factor c2 change with the number of iterations. In the early stage, c1 > c2, which encourages particle diversity; in the later stage, c2 > c1, which promotes the convergence of particles to the global optimum.

[0084] (2) Setting and executing inversion parameters:

[0085] Particle population size: set to 100.

[0086] Search space: The search range for the four load factors k1 to k4 is set to cover all possible combinations of tensile and compressive loads.

[0087] Maximum number of iterations: set to 600.

[0088] Fitness function: ,in The value is a very small positive number to prevent division by zero. The optimization objective is to maximize the fitness, i.e., minimize the objective function F(K).

[0089] The optimization process is automatically executed by the system's intelligent inversion engine. At the start of the computation, the particle swarm is randomly initialized within the search space. In each iteration, each particle updates its velocity and position based on its own historical best position and the swarm's historical best position. After approximately 450 iterations, the fitness function curve stabilizes, indicating that the algorithm has converged, at which point the global optimum is obtained. .

[0090] S5: Synthesis of the Three-Dimensional Initial Stress Field: By substituting the optimal load combination coefficient K* obtained from the PSO algorithm inversion into the linear superposition formula, the three-dimensional initial stress field of the entire engineering area can be synthesized. To verify the inversion accuracy, the calculated stress and the measured stress were compared. The results show that the average relative error of the main stress components is less than 15%, proving the reliability of the inversion results.

[0091] To visually demonstrate the inversion results and directly serve engineering design, the distribution of key stress components was extracted along the axis of the underground powerhouse and plotted as contour maps.

[0092] Figure 7 The contour lines of self-weight stress (σz) along the plant's axis are displayed. It can be seen that the σz value is mainly controlled by the burial depth, and it gradually increases from the entrance section of the plant to the tail section. The distribution pattern conforms to general laws, verifying the rationality of the model in terms of self-weight field.

[0093] Figure 8 The map displays the contour lines of the maximum principal stress (σ1). It clearly reveals that in areas of high geostress, σ1 values ​​are generally high (generally greater than 20 MPa). Of particular note is the significant density of contour lines near the footwall of the F1 fault, indicating a high stress concentration zone with σ1 exceeding 28 MPa. This provides crucial information for subsequent cavern layout and rockburst risk assessment.

[0094] Figure 9 The contour lines of the minimum principal stress (σ3) are shown. Their distribution is also closely related to geological structures, exhibiting obvious heterogeneity at fault zones and lithological changes.

[0095] These inversion results provide quantitative scientific basis for the fine-tuning of the axis of underground powerhouse cavern groups, the dynamic design of support parameters, and the formulation of rockburst prevention measures in key sections.

[0096] S6: Dynamic Stress Field Update: This system has dynamic update capabilities. After the excavation of the first floor of the plant was completed, two new measured stress data points were obtained in the newly added drainage tunnel. The dynamic update control module was triggered, executing the following logic:

[0097] Data assimilation: Incorporating new data into the existing measured database.

[0098] Update mode determination: Since the amount of new data is small and within the scope of the original model, the system determines to adopt the "local fast update" mode.

[0099] Fast inversion: Using the optimal solution K* obtained from the global inversion as the new initial population center, the PSO algorithm is run within a narrowed search range for a small number of iterations (e.g., 100 times) to quickly obtain the updated load factor K**.

[0100] Model update: The updated 3D stress field is synthesized using K** and all visualization results are automatically updated.

[0101] This mechanism ensures that the geostress field model can evolve dynamically along with the engineering construction, always providing the latest and most accurate geological environment information for engineering decisions, and truly realizing information-based construction and dynamic design.

[0102] This invention relates to a three-dimensional geostress field inversion method and its supporting automated analysis system. This method is based on an improved intelligent optimization algorithm, deeply integrates multi-source heterogeneous geological exploration data (including topography, geological structure, lithology distribution, geophysical characteristics, and field stress measurement data), and possesses the ability to dynamically update based on newly revealed information during the engineering process. This invention is particularly suitable for solving the challenge of high-precision geostress field assessment in major engineering projects such as high mountain and canyon areas, deep-buried long tunnels, and large underground hydropower station powerhouse complexes, providing crucial technical support for project site selection, tunnel axis optimization, support design, and disaster early warning.

[0103] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms, characterized in that: Includes the following steps: Step S1: Multi-source geological data acquisition and 3D geomechanical model construction: Through field testing and geological exploration, multi-source geological data of the engineering area are acquired. The multi-source geological data includes the 3D stress components of the measured stress points, topography, geological structure, lithology distribution, and rock mass mechanical parameters; a 3D geomechanical model of the engineering area is constructed based on the multi-source geological data. Step S2: Boundary load case setting and single-case stress field calculation: Based on the boundary of the three-dimensional geomechanical model constructed in step S1, the benchmark calculation case is constructed. The benchmark calculation case includes each single load case corresponding to the applied gravity load and the tectonic stress load in multiple directions. The stress distribution of the entire model under each single load case is calculated, and the stress calculation value at the measuring point is extracted. Step S3: Construction of the inversion objective function: Based on the stress calculation value extracted in step S2 and the measured value of the three-dimensional stress components in step S1, an inversion objective function is constructed. The inversion objective function is the weighted sum of squares of the difference between the measured stress value and the stress calculation value at the measurement point, and different weighting coefficients are introduced for the normal stress term and the shear stress term to reflect the difference in their data reliability. Step S4: Global Inversion of Intelligent Optimization Algorithm: Using an improved intelligent optimization algorithm, the optimal boundary load combination coefficients are determined by inversion using the inversion objective function established in Step S3 as the fitness function. Step S5: Synthesis of three-dimensional initial stress field: The optimal boundary load combination coefficients obtained in step S4 are linearly superimposed with the stress distribution of the entire model under the corresponding load condition in step S2 to synthesize and calculate the three-dimensional initial geostress field of the entire domain.

2. The method for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms according to claim 1, characterized in that, Also includes: Step S6: Dynamic Update of Stress Field: During the engineering construction process, when new multi-source geological data or stress monitoring data is obtained, the update process is triggered; the new multi-source geological information includes newly revealed geological structures, lithological distributions, or rock mass mechanical parameters; the stress monitoring data includes supplementary measured geostress data; the new multi-source geological data or stress monitoring data are incorporated into the original database, the updated data is used to reconstruct or correct the inversion objective function, and the boundary load combination coefficients are rapidly inverted based on the improved intelligent optimization algorithm. The optimal coefficients obtained from the inversion are used to dynamically correct the three-dimensional initial geostress field, generating an updated three-dimensional geostress field.

3. The method for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms according to claim 1, characterized in that, The three-dimensional stress components in step S1 are obtained by the hollow inclusion stress relief method or the hydraulic fracturing method. The rock mass mechanical parameters are obtained by the bearing plate load test. The rock mass mechanical parameters include the rock mass elastic modulus.

4. The method for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms according to claim 1, characterized in that, The structural stress loads in multiple directions in step S2 include: unit compressive load along the horizontal main structural direction, unit compressive load along the horizontal secondary structural direction, and unit shear load at the bottom.

5. The method for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms according to claim 1, characterized in that, The inversion objective function model in step S3 is: （1）； in, The objective function value is retrieved, where m is the total number of measurement points. , These are the measured values ​​of normal stress and shear stress at the k-th measuring point, respectively. , These are the calculated values ​​of the corresponding normal stress and shear stress, respectively. and These are the maximum values ​​of all measured normal and shear stresses, used for normalization. is the normal stress weighting coefficient, and its value range is [0,1].

6. The method for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms according to claim 1, characterized in that, The improved intelligent optimization algorithm in step S4 is an improved particle swarm optimization algorithm. The improvements include: adopting an adaptive inertia weight that decreases linearly with the number of iterations and introducing an asynchronously changing learning factor. In the early stage of iteration, the algorithm focuses on individual learning to maintain population diversity, and in the later stage of iteration, it focuses on social learning to promote convergence to the global optimum.

7. The method for dynamic inversion of three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms according to claim 6, characterized in that, Step S4 also includes a convergence judgment mechanism: when the change in the fitness function value within a set number of consecutive iterations is less than the preset tolerance, the optimization calculation process is automatically terminated to improve the inversion efficiency.

8. A dynamic inversion system for three-dimensional geostress field based on multi-source data fusion using intelligent optimization algorithms, characterized in that: include: The multi-source geological data acquisition and three-dimensional geomechanical model construction module is used to acquire multi-source geological data of the engineering area through field testing and geological exploration. The multi-source geological data includes three-dimensional stress components, topography, geological structure, lithology distribution and rock mass mechanical parameters of the measured stress points; and constructs a three-dimensional geomechanical model of the engineering area based on the multi-source geological data. The boundary load case setting and single-case stress field calculation module is used to construct the benchmark calculation case based on the boundary of the constructed three-dimensional geomechanical model. The benchmark calculation case includes each single load case corresponding to the applied gravity load and the tectonic stress load in multiple directions. It calculates the stress distribution of the entire model under each single load case and extracts the stress calculation value at the measuring point. The inversion objective function construction module is used to construct an inversion objective function based on the extracted calculated stress values ​​and the measured values ​​of the three-dimensional stress components. The inversion objective function is the weighted sum of squares of the difference between the measured stress value and the calculated stress value at the measurement point, and different weight coefficients are introduced for the normal stress term and the shear stress term to reflect the difference in their data reliability. The intelligent optimization algorithm global inversion module is used to determine the optimal boundary load combination coefficients by using an improved intelligent optimization algorithm with the established inversion objective function as the fitness function. The three-dimensional initial stress field synthesis module is used to linearly superimpose the optimal boundary load combination coefficients with the stress distribution of the entire model under the corresponding load conditions to synthesize and calculate the three-dimensional initial geostress field of the entire domain.

9. The multi-source data fusion three-dimensional geostress field dynamic inversion system based on intelligent optimization algorithm according to claim 8, characterized in that, Also includes: The stress field dynamic update module is used to trigger an update process during engineering construction when new multi-source geological data or stress monitoring data is obtained. The new multi-source geological information includes newly revealed geological structures, lithological distributions, or rock mass mechanical parameters. The stress monitoring data includes supplementary measured geostress data. The new multi-source geological data or stress monitoring data is incorporated into the original database, and the updated data is used to reconstruct or correct the inversion objective function. Based on an improved intelligent optimization algorithm, the boundary load combination coefficients are quickly inverted. The optimal coefficients obtained from the inversion are used to dynamically correct the three-dimensional initial geostress field, generating an updated three-dimensional geostress field.

10. The multi-source data fusion three-dimensional geostress field dynamic inversion system based on intelligent optimization algorithm according to claim 8, characterized in that, The intelligent inversion analysis module has a built-in parameter adaptive unit and a convergence judgment unit, which automatically terminates the optimization calculation process when the change in the fitness function value within a set number of consecutive iterations is less than the preset tolerance, so as to improve the inversion efficiency.

Images