A method for indoor twin slope physical model test

Abstract

The present application relates to the technical field of slope physical model, and discloses a kind of indoor twin slope physical model test method, including through multidimensional sensing array collection slope prototype multi-source data, based on digital twin technology constructs 1:1 geometric mapping model;Adopt finite element and discrete element coupling algorithm, in combination with complex geological structure representation and adaptive grid optimization to carry out numerical simulation;Determine multiple similarity ratios according to similarity principle, select matching similar material to pour indoor physical model;Synchronous start numerical simulation and physical test, realize the consistent response characteristics of both through real-time data interaction interface and dynamic correction mechanism;Fusion analysis and verification screening are carried out to the collected data, and a stability evaluation database is constructed.The present application improves complex geological representation, optimizes model parameters and sensing arrangement, realizes the high-precision replication of slope mechanical behavior and deformation and failure process, provides reliable data support for slope engineering safety early warning, and improves the accuracy and timeliness of stability evaluation.

Description

Technical Field

[0001] This invention relates to the field of slope physical modeling technology, and in particular to an indoor twin slope physical model testing method. Background Technology

[0002] Slopes, as common geological structures in engineering construction, directly affect the safe operation of various projects such as transportation hubs, building construction, and water conservancy facilities. Landslides and collapses caused by slope instability often result in significant casualties and property losses. As engineering construction expands into complex geological areas, the geological environment of slopes becomes increasingly complex. Traditional physical model tests and single numerical simulation methods are no longer sufficient to meet the demands for high-precision, real-time stability evaluation. The rise of digital twin technology has provided a new path for slope engineering research. By constructing a mapping relationship between the physical world and digital space, dynamic monitoring and simulation of the entire life cycle of slopes has become an industry trend. The indoor twin slope physical model test method, developed against this backdrop, integrates multi-source sensing technology, numerical calculation methods, and similarity principles, aiming to establish a more realistic slope stability research system and overcome the limitations of traditional methods in complex geological conditions.

[0003] Existing technologies have two significant drawbacks: First, insufficient data interaction and collaboration. In traditional methods, numerical simulation and physical experiments are often conducted independently, lacking real-time data feedback and dynamic correction mechanisms. The parameter settings of numerical models are difficult to adjust in a timely manner based on the results of physical experiments, resulting in poor consistency between the two in terms of mechanical response and deformation patterns, and failing to fully leverage the synergistic advantages of digital twin technology. Second, inadequate model adaptability and accuracy. The numerical simulation process does not comprehensively characterize the complex geological structure of the slope, and the mesh generation and constitutive relation parameter selection lack targeted optimization. Furthermore, the similar material ratio design of the indoor physical model does not fully consider the nonlinear mechanical characteristics of the prototype soil and rock, and the sensor deployment is not precisely planned in conjunction with key stress areas, resulting in deviations between the experimental results and the actual prototype situation, making it difficult to effectively support accurate safety early warning for slope engineering. Summary of the Invention

[0004] In order to overcome the shortcomings and deficiencies of existing technologies, this invention provides an indoor twin slope physical model test method.

[0005] The technical solution adopted in this invention is an indoor twin slope physical model test method, comprising the following steps: S1, real-time acquisition of mechanical parameters, displacement field distribution, and pore water pressure changes of the slope prototype using a multi-dimensional sensor array to form an original dataset including multi-source heterogeneous data; S2, construction of a geometric mapping model of the slope prototype based on digital twin technology, obtaining the surface topology of the prototype through three-dimensional laser scanning, supplementing internal structural information with geological drilling data, and establishing a geometric digital representation that matches the prototype 1:1; S3, numerical simulation of the slope mechanical behavior using a coupled finite element and discrete element method, characterizing complex geological interfaces by dividing unstructured meshes, setting constitutive relation parameters that conform to the actual prototype, and performing slope deformation... The process of slope failure is dynamically replicated; S4, an indoor physical model test platform is built, and the geometric similarity ratio, mechanical similarity ratio, and time similarity ratio are determined according to the principle of similarity. Similar materials with the same mechanical properties as the prototype soil and rock are selected and poured to form an indoor slope model according to the design ratio; S5, numerical simulation and physical model test are started simultaneously, and the mechanical parameters, displacement response, and failure mode of the two are interactively fed back in real time through the data transmission interface. The parameters of the numerical model are adjusted based on the dynamic correction algorithm to make the response characteristics of the two consistent; S6, the numerical simulation data and physical test data collected during the test are fused and analyzed. The effective data are screened through the data verification mechanism to build a slope stability evaluation database, providing data support for slope engineering safety early warning.

[0006] Furthermore, the constitutive relation parameters in S3 are determined by the following expression: ,in, For shear stress, For cohesion, For normal stress, It is the internal friction angle. The nonlinear deformation coefficient is... For axial normal strain, This represents shear strain.

[0007] Furthermore, the proportion of similar materials in step S4 is calculated using the following expression: ,in, Let i be the mass percentage of the i-th component. The density of the prototype soil and rock mass. This represents the volume of the prototype rock and soil mass. Let be the mechanical property matching coefficient of the i-th component. Let be the density of the j-th similar material component. Let be the volume of the j-th similar material component, and n be the total number of similar material components.

[0008] Furthermore, the dynamic correction algorithm in S5 is implemented based on the following expression: ,in, For the corrected first Numerical model parameters at time t. Let be the numerical model parameters at time k. For the first Numerical model parameters at time t. This is the proportional correction factor. This is the inertial correction factor. is the integral correction coefficient, E is the response error between numerical simulation and physical experiment, and t is the experiment time.

[0009] Furthermore, the accuracy verification of the geometric mapping model in S2 is calculated using the following expression: ,in, M represents the geometric mapping accuracy error, and M represents the number of surface topological feature points. Let i be the coordinates of the i-th feature point on the model surface. Let N be the coordinates of the i-th feature point on the prototype surface, and N be the number of internal structural feature points. Let j be the coordinates of the j-th feature point inside the model. The coordinates of the j-th feature point inside the prototype.

[0010] Furthermore, the data fusion analysis in S6 is implemented through the following expression: ),in, For the merged data, For data weighting coefficients, For numerical simulation data, For physical experimental data, The attenuation coefficient is... This is the phase adjustment coefficient.

[0011] Further, S3 includes the following sub-steps: S31, obtaining spatial distribution information of faults and weak interlayers hidden structures inside the slope prototype through ground-penetrating radar detection, determining the differences in mechanical parameters of different geological layers by combining seismic wave test data, and establishing an initial numerical calculation model including complex geological structures; S32, using adaptive mesh refinement technology to refine the mesh of potential sliding surfaces and stress concentration areas of the slope, judging the rationality of mesh division through error estimation function, and readjusting the mesh density for areas that do not meet the calculation accuracy requirements; S33, based on the coupling interface of FLAC3D and PFC3D, performing collaborative calculation of the mechanical behavior of continuous and discrete media, setting the force transmission coefficient and displacement coordination conditions of the coupling interface to ensure the continuity of data interaction between different calculation modules; S34, introducing an adaptive time step adjustment mechanism to dynamically optimize the calculation time step according to the slope deformation rate. When the deformation rate exceeds a threshold, the time step is reduced; when the deformation rate is below the threshold, the time step is increased, improving calculation efficiency while ensuring calculation accuracy.

[0012] Further, S4 includes the following sub-steps: S41, determining the calibration influencing factors of similar materials through orthogonal experimental design, selecting aggregate particle size, binder content, moisture content, and compaction degree as experimental variables, designing multiple mix proportion schemes at different levels, and screening out the basic mix proportion with the highest matching degree with the prototype soil and rock mass through mechanical performance testing; S42, deriving the conversion relationship between geometric similarity ratio, stress similarity ratio, strain similarity ratio, and time similarity ratio based on the similarity principle, and determining the similarity ratios based on the spatial size limitations of the indoor test platform and the load range of the loading equipment. The specific value of the ratio; S43, the indoor physical model is constructed by layered casting method, and similar materials are laid in sequence according to the designed geological layering order. After each layer is laid, the compaction degree is controlled by vibration compaction equipment to ensure uniform density inside the model and avoid segregation; S44, multiple sets of sensing elements are preset inside the physical model, including strain gauges, displacement gauges and pore water pressure sensors. The placement and number of sensors are determined according to the calibration stress area predicted by numerical simulation. The sensor leads are led out of the model through sealed channels to avoid damage during the test.

[0013] Further, step S5 includes the following sub-steps: S51, real-time acquisition of strain, displacement, and pore water pressure data from the physical model experiment using a data acquisition card, conversion of analog signals to digital signals via A / D conversion, and transmission to the data processing center via Ethernet; S52, establishment of a communication connection between the numerical simulation system and the physical experiment system based on the TCP / IP protocol, setting the data transmission frequency and data format to ensure the real-time performance and accuracy of data transmission between the two; S53, use of the Kalman filter algorithm to filter out noise data during transmission, and prediction of the optimal estimated value of the data by establishing state equations and observation equations, eliminating the impact of environmental interference and equipment errors on data quality; S54, comparison and analysis of the filtered physical experiment data and numerical simulation results, calculation of the deviation between the two, and adjustment of constitutive parameters, boundary conditions, and initial stress field distribution in the numerical model based on the deviation feedback signal to keep the numerical simulation results synchronized with the physical experiment response.

[0014] An indoor twin slope physical model testing method is proposed, implemented through different units, including: a multi-source data acquisition unit for the slope prototype, composed of a 3D laser scanner, ground-penetrating radar, stress sensor, displacement sensor, and pore water pressure sensor, which comprehensively collects the geometric parameters and mechanical response data of the slope prototype through distributed deployment, and establishes a wired communication connection between the data output end and the data transmission unit; a slope digital twin modeling unit, equipped with 3D modeling software and a geological data processing module, which receives the raw data transmitted by the data acquisition unit, constructs a digital mapping model of the slope prototype through geometric reconstruction algorithms and geological information fusion technology, and outputs the modeling results to the numerical simulation unit; and a slope numerical simulation calculation unit, integrating a finite element-discrete element coupled calculation module and an adaptive solver, which performs slope mechanical behavior simulation calculations based on the digital twin model, and performs calculations through... The data interaction interface enables real-time data exchange with the physical model test unit. The indoor physical model test unit, including a similar material preparation device, model casting mold, loading system, and sensing module, constructs an indoor slope model according to the similarity principle and conducts mechanical tests. The test data is preprocessed and then transmitted to the data fusion analysis unit. The data fusion analysis unit is equipped with a multi-source data verification module, an error analysis module, and a database storage module. It receives numerical simulation data and physical test data, processes them through a data fusion algorithm to form a standardized dataset, and outputs it to the results application unit. The results application unit is equipped with a slope stability evaluation model and a data visualization module. It receives the fused and analyzed dataset, generates a slope safety assessment report, and displays it in chart form, providing technical support for engineering decisions. All units establish a data transmission link through an industrial Ethernet network for data interaction and collaborative work.

[0015] Compared with the prior art, the present invention has at least one of the following beneficial effects: This invention utilizes a multi-source sensor array to comprehensively collect prototype slope data, combines digital twin technology to construct a 1:1 geometric mapping model, achieves dynamic replication of mechanical behavior through a finite element and discrete element coupled algorithm, and simultaneously builds an indoor physical model based on the principle of similarity. A linkage mechanism between numerical simulation and physical experiment is established through a real-time data interaction interface, and the model parameters are continuously optimized through a dynamic correction algorithm. A stability evaluation database is constructed through multi-source data fusion analysis, forming a complete technical chain of data acquisition, modeling and simulation, physical experiment, collaborative correction, data fusion, and result application.

[0016] To address the shortcomings of traditional methods in terms of insufficient data synergy, this method establishes a real-time interactive feedback system between numerical simulation and physical experiment. Through dynamic correction and data fusion technology, it achieves precise matching of the response characteristics of the two, giving full play to the advantages of digital twin linkage. To address the issues of insufficient model adaptability and accuracy, this study improves the characterization of complex geological structures through ground-penetrating radar and seismic wave testing. It also optimizes the numerical model and similar material ratios by employing adaptive grid densification and orthogonal experiments, and precisely deploys sensing elements in key areas. This significantly improves the fit between experimental results and actual prototype conditions, providing more reliable technical support for slope engineering safety early warning and significantly enhancing the accuracy and timeliness of slope stability evaluation. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention. Figure 2 This is a flowchart of method step S3 of the present invention; Figure 3 This is a flowchart of method step S4 of the present invention; Figure 4 This is a flowchart of step S5 of the method of the present invention; Figure 5 This is a diagram showing the system unit composition of the present invention. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] like Figure 1 As shown, an indoor twin slope physical model test method includes the following steps: S1 uses a multi-dimensional sensor array to collect the mechanical parameters, displacement field distribution and pore water pressure changes of the slope prototype in real time, forming a raw dataset including multi-source heterogeneous data. Specifically, step S1 utilizes a multi-dimensional sensor array to comprehensively acquire multi-source data from the slope prototype. The sensor array includes stress sensors, displacement sensors, pore water pressure sensors, and ground-penetrating radar. The sensors are distributed at a density of 3-5 per square meter. The stress sensors have a range of 0-50 MPa and a measurement accuracy of ±0.1 MPa. The displacement sensors use laser displacement gauges with a measurement range of 0-500 mm and a resolution of 0.01 mm. The pore water pressure sensors cover a range of 0-2 MPa with a response time not exceeding 0.1 seconds. During data acquisition, the sampling frequency is set to 100 Hz, and the continuous acquisition time is no less than 72 hours to ensure the capture of the mechanical parameter fluctuations, dynamic distribution of the displacement field, and pore water pressure changes of the slope prototype under natural conditions. The acquired mechanical parameters include key indicators such as the compressive strength, shear strength, and elastic modulus of the soil and rock. The displacement field distribution data includes the spatial distribution characteristics of horizontal, vertical, and lateral displacements. The pore water pressure data records pressure change curves at different depths and locations. The analog signals acquired by various sensors are converted into digital signals by the data acquisition card and transmitted to the data storage server via Ethernet. This forms a multi-source heterogeneous raw dataset that includes mechanical parameters, displacement field data, pore water pressure data, and geological structure information. This dataset provides comprehensive and accurate basic data support for subsequent digital twin model construction, numerical simulation, and physical model testing, ensuring that the implementation of each subsequent step is based on real and reliable prototype data.

[0020] S2, based on digital twin technology, constructs a geometric mapping model of the slope prototype, obtains the surface topology of the prototype through three-dimensional laser scanning, and supplements the internal structural information by combining geological drilling data to establish a geometric digital representation that matches the prototype 1:1. Specifically, step S2 utilizes digital twin technology to construct a geometric mapping model of the slope prototype. First, a 3D laser scanner is used to comprehensively scan the surface of the slope prototype. The scanner's scanning accuracy is set to ±0.5mm, and the scanning resolution is 0.1mm. The scanning range covers the entire surface area of ​​the slope prototype, acquiring surface topological structure information including millions of point cloud data points. The density of the point cloud data is controlled at 50-80 points per square centimeter to ensure accurate reconstruction of the surface morphology of the slope prototype. Simultaneously, geological drilling is conducted. Drill holes are arranged in a grid pattern with a spacing of 5-8 meters, reaching a depth of 10 meters below the stable rock layer of the slope prototype. Core samples are obtained through drilling to analyze and determine the internal geological stratification, fault distribution, location and thickness of weak interlayers, and other internal structural information of the slope. The core sampling interval for each drill hole is 0.5 meters to ensure a comprehensive understanding of the internal geological structure details. Surface point cloud data acquired through 3D laser scanning and internal structural data obtained through geological drilling are imported into professional 3D modeling software. A geometric reconstruction algorithm is used for data fusion processing to eliminate data noise and redundant information, establishing a geometric digital representation model that perfectly matches the slope prototype in terms of geometric dimensions, surface morphology, and internal structure (1:1 scale). After model construction, the geometric accuracy of the model is verified by selecting no fewer than 100 surface feature points and 50 internal feature points for field measurement and comparison. This ensures that the positional error of the surface feature points does not exceed ±1mm, and the positional error of the internal feature points does not exceed ±2mm. This geometric mapping model provides an accurate geometric foundation for subsequent numerical simulations, ensuring that numerical calculations can accurately reflect the spatial structural characteristics of the slope prototype.

[0021] S3 uses a coupled finite element and discrete element algorithm to numerically simulate the mechanical behavior of slopes. By dividing unstructured meshes to characterize complex geological interfaces, constitutive relation parameters that conform to the actual prototype are set to dynamically replicate the slope deformation and failure process. Specifically, step S3 employs a coupled finite element method (FEM) and discrete element method (DEM) algorithm to conduct numerical simulations of the slope's mechanical behavior. First, based on the geological structural characteristics of the slope prototype, the geometric mapping model is divided into unstructured meshes. A hybrid tetrahedral and hexahedral mesh is used, with the mesh size dynamically adjusted according to the complexity of the geological interface. In complex geological interface areas such as faults and weak interlayers, the mesh size is controlled at 0.1-0.3 meters; in homogeneous rock and soil areas, the mesh size is set at 0.5-1.0 meters. The total number of meshes is controlled at 500,000-800,000 to ensure both accurate characterization of complex geological interfaces and computational efficiency. Subsequently, constitutive relation parameters consistent with the actual slope prototype are set. These parameters are determined based on indoor test results of the prototype rock and soil, combined with core sample test data obtained from geological drilling. Key mechanical parameters include cohesion, internal friction angle, elastic modulus, and Poisson's ratio. Simultaneously, considering the nonlinear deformation characteristics of the rock and soil, a nonlinear deformation coefficient is introduced to correct the constitutive relation. In the numerical simulation, boundary conditions and an initial stress field were set. The boundary conditions were set according to the actual constraints of the slope prototype: displacement constraints were used in the horizontal direction, and stress constraints were used in the vertical direction. The initial stress field was calculated and determined based on the self-weight stress distribution law of the slope prototype, and the initial pore water pressure was set according to measured data. The simulation calculation adopted a dynamic time step, which was dynamically adjusted according to the slope deformation rate. When the deformation rate was fast, the time step was set to 0.001-0.01 seconds; when the deformation rate was slow, the time step was set to 0.01-0.1 seconds. The finite element method and discrete element method calculation modules were coupled through a coupling algorithm to simulate the stress distribution, displacement evolution, and deformation failure process of the slope under natural conditions and external loads. The complete mechanical behavior of the slope from a stable state to instability failure was dynamically replicated, providing a reference for subsequent physical model tests.

[0022] S4. Build an indoor physical model test platform, determine the geometric similarity ratio, mechanical similarity ratio and time similarity ratio according to the similarity principle, select similar materials with the same mechanical properties as the prototype rock and soil, and pour them into an indoor slope model according to the design ratio. Specifically, step S4 focuses on the construction of the indoor physical model test platform. First, based on the principle of similarity, various similarity ratio parameters are determined. Considering the spatial size limitations of the indoor test platform (length × width × height = 20 × 10 × 15 meters) and the load range of the loading equipment (maximum loading force = 5000 kN), combined with the geometric dimensions and mechanical properties of the slope prototype, the geometric similarity ratio is determined to be 1:20-1:50, the mechanical similarity ratio (stress similarity ratio) to be 1:10-1:30, and the time similarity ratio to be 1:5-1:10. The conversion relationships between these similarity ratios satisfy the principle of similarity, ensuring that the physical model and the prototype maintain similarity in mechanical behavior. Subsequently, similar materials with mechanical properties consistent with the prototype soil and rock mass are selected. These similar materials consist of aggregates, binders, moisture content regulators, and additives. The aggregates used are quartz sand with a particle size of 0.075-5 mm. The binder is a mixture of cement, gypsum, and epoxy resin in a certain proportion. The moisture content regulator is distilled water, and the additive is montmorillonite to adjust cohesion and internal friction angle. Multiple mix design schemes were employed through orthogonal experimental design. Aggregate particle size, binder content, moisture content, and compaction degree were selected as experimental variables, with 4-5 levels for each variable, resulting in at least 30 mix design schemes. Standard specimens were prepared for each scheme, and mechanical properties such as compressive strength, shear strength, and elastic modulus were tested. The mix design with the highest degree of matching with the mechanical properties of the prototype soil and rock was selected. An indoor slope model was poured according to the design scale, using a layered pouring method. The thickness of each layer was controlled at 5-10 cm, and the pouring sequence strictly followed the geological stratification sequence of the slope prototype. After each layer was poured, a vibratory compaction device was used for compaction, with the compaction degree controlled at 95%-98% to ensure uniform density within the model and avoid segregation. After the model is poured, it is cured. The temperature of the curing environment is controlled at 20±2℃ and the relative humidity is maintained above 90%. The curing time is no less than 28 days to allow the similar materials to fully solidify and the mechanical properties to meet the design requirements. Finally, an indoor slope model similar to the slope prototype mechanics is formed, providing a physical carrier for subsequent synchronous tests.

[0023] S5, simultaneously starts numerical simulation and physical model test, and conducts real-time interactive feedback of mechanical parameters, displacement response and failure mode between the two through data transmission interface. The numerical model parameters are adjusted based on dynamic correction algorithm to keep the response characteristics of the two consistent. Specifically, step S5 enables the simultaneous execution and real-time interactive feedback of numerical simulation and physical model testing. First, the numerical simulation system and the indoor physical model testing system are started, with the synchronization start-up time deviation set to no more than ±0.1 seconds to ensure time consistency during the testing process. A real-time communication connection is established between the two through a data transmission interface. The data transmission interface uses an industrial Ethernet interface with TCP / IP protocol and a data transmission frequency set to 50Hz to ensure real-time transmission of mechanical parameters, displacement response, and failure mode data. In the physical model test, data such as strain, displacement, and pore water pressure are collected in real time through preset sensing elements. The numerical simulation system synchronously outputs the calculation results at the corresponding time. After the data is transmitted to the data processing center, a data comparison algorithm is used to compare the mechanical parameters (stress, strain), displacement response (horizontal displacement, vertical displacement), and failure mode characteristics of the two systems in real time, calculating the response error between them. The numerical model parameters are adjusted in real time based on a dynamic correction algorithm, with a correction cycle set to 0.5 seconds. During each correction, the constitutive parameters, boundary conditions, and initial stress field distribution in the numerical model are adjusted according to the magnitude and trend of the response error. The adjustment range is dynamically determined according to the error proportion; when the error is large, the adjustment range is controlled at 5%-10%, and when the error is small, the adjustment range is set at 1%-3%. Simultaneously, a displacement coordination and force transfer mechanism is established to ensure consistency between the numerical simulation and the physical experiment in terms of boundary conditions and interactions. When local deformation or failure occurs in the physical model, the numerical simulation system receives this information in real time and adjusts the calculation parameters of the corresponding region. Conversely, when the numerical simulation predicts a potential failure region, the physical experiment system focuses on monitoring the mechanical response of that region. Through continuous real-time interactive feedback and dynamic correction, the response characteristics of the numerical simulation and the physical experiment remain highly consistent, with the error controlled within 5%, fully leveraging the synergistic advantages of digital twin technology.

[0024] S6 integrates and analyzes the numerical simulation data and physical test data collected during the experiment, filters valid data through a data verification mechanism, and constructs a slope stability evaluation database to provide data support for slope engineering safety early warning.

[0025] Specifically, step S6 involves the fusion analysis and database construction of numerical simulation data and physical test data. First, the two types of data collected during the experiments are preprocessed. Numerical simulation data includes calculated results of stress, strain, displacement, pore water pressure, etc., with a data volume of 100-150GB. Physical test data includes measured data collected by various sensors, with a data volume of 50-80GB. During preprocessing, data cleaning algorithms are used to remove outlier and invalid data. Outlier identification is based on the 3σ criterion, i.e., data deviating from the mean by more than three standard deviations are removed. Simultaneously, missing data is supplemented using interpolation to ensure data integrity. Subsequently, a data verification mechanism is used to screen the validity of the preprocessed data. Verification indicators include data consistency, rationality, and reliability. Consistency verification is achieved by comparing the changing trends of numerical and physical data. Rationality verification is based on the theoretical range of the mechanical properties of soil and rock. Reliability verification is completed by comparing repeated experimental data, selecting datasets with a valid data ratio of no less than 90%. A multi-source data fusion algorithm was employed to fuse the effective data. During the fusion process, different weighting coefficients were assigned based on the data type and reliability. The weighting coefficients for numerical simulation data and physical test data were set to 0.4-0.6, ensuring that the fusion results reflected both the comprehensiveness of the numerical simulation and the realism of the physical tests. After fusion processing, a slope stability evaluation database was constructed. The database includes raw collected data, preprocessed data, effective data, fused data, and comparative analysis results of numerical simulation and physical tests. The database adopts a distributed storage architecture, supporting rapid data querying, retrieval, and updating. The data storage format is a standard structured format, facilitating subsequent access and analysis. This database includes data on the mechanical behavior, displacement response, and failure modes of slopes under different working conditions, providing rich and reliable data support for the establishment of slope engineering safety early warning models, ensuring the accuracy and timeliness of safety warnings.

[0026] Preferably, the constitutive relation parameters in S3 are determined by the following expression: ,in, For shear stress, For cohesion, For normal stress, It is the internal friction angle. The nonlinear deformation coefficient is... For axial normal strain, This represents shear strain.

[0027] Specifically, the constitutive relation parameter determination expression in step S3 is used to accurately describe the nonlinear relationship between shear stress and related influencing factors in slope soil and rock under stress, ensuring that the numerical simulation can truly reflect the mechanical response characteristics of the soil and rock. During implementation, the cohesion and internal friction angle of the prototype slope soil and rock are first obtained through indoor tests. A standard shear testing device is used, and at least three parallel tests are conducted on core samples collected at different depths. The average value of the test results is taken as the basic parameters. Axial normal strain and shear strain are acquired in real time using strain gauges at a frequency of 50Hz to ensure the capture of subtle strain changes. The nonlinear deformation coefficient is determined by fitting multiple sets of test data under different stress levels. The fitting process uses the least squares method, and adjustments are made based on the deviation feedback between the numerical simulation and physical tests, so that the coefficient values ​​can adapt to the deformation characteristics of soil and rock under different geological conditions. This model, based on the traditional Mohr-Coulomb theory, introduces first-order and second-order terms for axial normal strain and logarithmic terms for shear strain, fully considering the nonlinear deformation behavior of soil and rock under complex stress states. This avoids the shortcomings of traditional linear models in accurately characterizing the mechanical properties during large deformation stages. The constitutive parameters determined by this model enable more realistic stress calculations in numerical simulations, improving the accuracy of dynamically replicating the slope deformation and failure process. This lays a reliable mechanical parameter foundation for subsequent collaboration between numerical simulations and physical experiments.

[0028] Preferably, the proportion of similar materials in step S4 is calculated using the following expression: ,in, Let i be the mass percentage of the i-th component. The density of the prototype soil and rock mass. This represents the volume of the prototype rock and soil mass. Let be the mechanical property matching coefficient of the i-th component. Let be the density of the j-th similar material component. Let be the volume of the j-th similar material component, and n be the total number of similar material components.

[0029] Specifically, the calculation method for the similar material ratio in step S4 ensures, through quantitative calculation, that the mechanical properties of the indoor physical model are highly consistent with those of the prototype slope soil and rock, meeting the requirements of the similarity principle. During implementation, the density and volume of the prototype slope soil and rock are first measured. Density is measured using the ring cutter method, selecting no fewer than 10 measurement points, with each point being measured three times repeatedly. Volume is calculated using 3D laser scanning data combined with geological drilling results. The density of each component of the similar material is determined using the hydrometer bottle method, and the volume is determined based on the spatial dimensions of the indoor test platform and the model design scale. The mechanical property matching coefficient is set according to the degree of influence of each component on the overall mechanical properties of the similar material, and optimized through orthogonal experiments. The orthogonal experimental design includes 4 factors and 5 levels, conducting a total of 30 sets of experiments. The range of matching coefficient values ​​for each component is determined through range analysis and variance analysis. In the calculation process, the product of the density, volume, and matching coefficient of each component is calculated separately, and then summed to obtain the denominator. The numerator is the product of the density and volume of the prototype soil and rock and the corresponding component's matching coefficient; the ratio of the two is the mass percentage of that component. The similar material ratio calculated by this model can accurately control the amount of each component, so that the key mechanical indicators such as compressive strength, shear strength and elastic modulus of the similar materials match the prototype soil and rock mass by more than 90%. This avoids the large difference between the mechanical properties of the physical model and the prototype due to unreasonable ratio, and ensures the reliability and reference value of the indoor physical model test results.

[0030] Preferably, the dynamic correction algorithm in S5 is implemented based on the following expression: ,in, For the corrected first Numerical model parameters at time t. Let be the numerical model parameters at time k. For the first Numerical model parameters at time t. This is the proportional correction factor. This is the inertial correction factor. is the integral correction coefficient, E is the response error between numerical simulation and physical experiment, and t is the experiment time.

[0031] Specifically, the dynamic correction algorithm in step S5 is used to adjust the numerical model parameters in real time, ensuring that the response characteristics of the numerical simulation and the physical experiment are consistent, thus fully leveraging the synergistic advantages of digital twin technology. During implementation, initial values ​​for the proportional correction coefficient, inertial correction coefficient, and integral correction coefficient are first set. The initial range for the proportional correction coefficient is 0.1-0.3, the initial range for the inertial correction coefficient is 0.05-0.15, and the initial range for the integral correction coefficient is 0.01-0.05. These values ​​are then dynamically adjusted based on the response error during the experiment. The response error between the numerical simulation and the physical experiment is calculated by the difference between the mechanical parameters and displacement response data of both at the same time. The error calculation frequency is consistent with the data transmission frequency, set to 50Hz. The numerical model parameters at time k and time k-1 are retrieved in real time from the database of the numerical simulation system. These parameters include constitutive relation parameters, boundary condition parameters, and initial stress field parameters. The integral term is calculated using a numerical integration method. The integration interval is from the start of the experiment to the current time, and the integration step size is consistent with the time step size of the numerical simulation, set to 0.001-0.1 seconds. This model comprehensively considers the current parameter values, error gradient changes, historical parameter trends, and the cumulative effect of error gradients. Through a multi-dimensional correction mechanism, it dynamically optimizes the numerical model parameters, with a correction period set at 0.5 seconds to ensure timely and accurate parameter adjustments. This algorithm can control the response error between numerical simulation and physical experiment within 5%, effectively solving the problems of data disconnect and poor consistency in traditional methods, and improving the overall synergy and reliability of the experiment.

[0032] Preferably, the accuracy verification of the geometric mapping model in S2 is calculated using the following expression: ,in, M represents the geometric mapping accuracy error, and M represents the number of surface topological feature points. Let i be the coordinates of the i-th feature point on the model surface. Let N be the coordinates of the i-th feature point on the prototype surface, and N be the number of internal structural feature points. Let j be the coordinates of the j-th feature point inside the model. The coordinates of the j-th feature point inside the prototype.

[0033] Specifically, the accuracy verification method for the geometric mapping model in step S2 evaluates the geometric accuracy of the model through quantitative calculations, ensuring that the model can accurately replicate the surface topology and internal structure of the slope prototype. During implementation, surface topological feature points and internal structural feature points are first selected. For surface feature points, priority is given to key locations such as slope contour inflection points and abrupt slope changes, with no fewer than 100 points selected. For internal feature points, fault intersections and key points at weak interlayer interfaces are selected, with no fewer than 50 points selected. The coordinates of the feature points are obtained using high-precision measuring equipment. Surface feature points are measured using a 3D laser scanner with a measurement accuracy set to ±0.1mm, while internal feature points are measured using ground-penetrating radar combined with borehole sampling, with a measurement accuracy controlled within ±0.2mm. The coordinates of the surface and internal feature points are extracted from the completed geometric mapping model. The extraction process utilizes the coordinate query function of professional modeling software to ensure the accuracy of the extracted data. In the calculation process, the relative errors between the model coordinates and the prototype coordinates of each surface feature point and internal feature point are first calculated. Then, the squares of the relative errors of all surface feature points are summed and averaged. The same process is applied to internal feature points. Finally, the square root of the sum of the two averages is taken to obtain the geometric mapping accuracy error. This model comprehensively considers the errors of both surface and internal feature points, and can fully reflect the geometric accuracy of the model. The accuracy error threshold is set at ±0.5mm. When the calculation result exceeds the threshold, the model is corrected and adjusted by re-optimizing the point cloud data fusion algorithm and geometric reconstruction parameters until the error meets the requirements, providing an accurate geometric basis for subsequent numerical simulations.

[0034] Preferably, the data fusion analysis in S6 is implemented using the following expression: ),in, For the merged data, For data weighting coefficients, For numerical simulation data, For physical experimental data, The attenuation coefficient is... This is the phase adjustment coefficient.

[0035] Specifically, the data fusion analysis implementation model in step S6 improves the reliability and integrity of data through the organic fusion of multi-source data, providing high-quality data support for the construction of the slope stability evaluation database. During implementation, data weighting coefficients are first set. The initial weighting coefficients for both numerical simulation data and physical test data are 0.5, and are subsequently dynamically adjusted based on data reliability. Reliability is determined by the results of a data verification mechanism. When the proportion of effective data is higher than 95%, the weighting coefficient is increased to 0.6; when the proportion of effective data is lower than 85%, the weighting coefficient is decreased to 0.4. The attenuation coefficients are set according to the degree of influence of data deviation. λ1 ranges from 0.8 to 1.2, and λ2 ranges from 1.0 to 1.5, ensuring that the weight of the corresponding data decreases exponentially as the data deviation increases, reducing the impact of data with large deviations on the fusion results. The phase adjustment coefficient is set to 0.3-0.7 to adjust the contribution of the maximum value term to the fusion results, ensuring that the fusion results highlight the role of effective data while also considering the overall trend of the data. In the calculation, the products of numerical simulation data and the attenuation coefficient and the exponential term of the data deviation are calculated first, and the corresponding products of physical test data are calculated separately. The products of the maximum values ​​of both are then multiplied by the phase adjustment coefficient and the sinusoidal term of the data deviation. These three results are then added together to obtain the fused data. This model comprehensively considers the comprehensiveness of the numerical simulation data and the authenticity of the physical test data. It reduces the interference of outlier data through exponential attenuation and sinusoidal adjustment mechanisms, improving the stability and accuracy of the fused data. After standardization, the fused dataset is imported into a slope stability evaluation database, providing comprehensive and reliable data support for slope engineering safety early warning, ensuring the accuracy and timeliness of the warning results.

[0036] Preferred, such as Figure 2 As shown, S3 includes the following sub-steps: S31, obtaining spatial distribution information of faults and weak interlayers hidden structures inside the slope prototype through ground-penetrating radar detection, determining the differences in mechanical parameters of different geological layers by combining seismic wave test data, and establishing an initial numerical calculation model including complex geological structures; S32, using adaptive mesh refinement technology to refine the mesh of potential sliding surfaces and stress concentration areas of the slope, judging the rationality of mesh division through error estimation function, and readjusting the mesh density for areas that do not meet the calculation accuracy requirements; S33, based on the coupling interface of FLAC3D and PFC3D, performing collaborative calculation of the mechanical behavior of continuous and discrete media, setting the force transmission coefficient and displacement coordination conditions of the coupling interface to ensure the continuity of data interaction between different calculation modules; S34, introducing an adaptive time step adjustment mechanism to dynamically optimize the calculation time step according to the slope deformation rate. When the deformation rate exceeds the threshold, the time step is reduced; when the deformation rate is below the threshold, the time step is increased, improving calculation efficiency while ensuring calculation accuracy.

[0037] Specifically, step S3 ensures that the numerical simulation accurately replicates the slope's mechanical behavior through four sub-steps. In S31, ground-penetrating radar is used to detect hidden structures within the slope prototype. The detection frequency is set to 50-100MHz, and the detection depth covers the entire depth of the slope prototype. Combined with seismic wave test data, the seismic wave propagation velocity measurement accuracy is controlled within ±10m / s. Through data analysis, the spatial location, orientation, and thickness of faults and weak interlayers are determined. Simultaneously, the differences in mechanical parameters such as compressive strength and elastic modulus among different geological layers are clarified. Based on these data, an initial numerical calculation model including complex geological structures is established. In S32, adaptive mesh refinement technology is used. An initial mesh is first generated, and then the mesh generation error is calculated using an error estimation function. When the error exceeds 0.5%, the mesh is refined for potential sliding surfaces and stress concentration areas of the slope. The refined mesh size is 1 / 3 to 1 / 5 of the initial mesh size, ensuring that the mesh generation meets the calculation accuracy requirements. S33 relies on a specific coupling interface, setting the force transmission coefficient of the coupling interface to 0.95-0.99, and setting the displacement coordination condition according to the deformation continuity principle, to achieve collaborative calculation of the mechanical behavior of continuous and discrete media, ensuring uninterrupted and unbiased data interaction between different calculation modules. S34 introduces an adaptive time step adjustment mechanism, setting the deformation rate threshold to 0.01-0.05 mm / s. When the slope deformation rate exceeds this threshold, the time step is reduced to 1 / 2-1 / 3 of the original; when the deformation rate is below the threshold, the time step is increased to 1.5-2 times the original. While ensuring a calculation accuracy of no less than 95%, the calculation efficiency is greatly improved, enabling numerical simulation to efficiently and accurately reproduce the entire process of slope deformation and failure.

[0038] Preferred, such as Figure 3 As shown, S4 includes the following sub-steps: S41, determining the calibration influencing factors of similar materials through orthogonal experimental design, selecting aggregate particle size, binder content, moisture content and compaction degree as experimental variables, designing multiple mix proportion schemes at different levels, and screening out the basic mix proportion with the highest matching degree with the prototype soil and rock mass through mechanical performance testing; S42, deriving the conversion relationship between geometric similarity ratio, stress similarity ratio, strain similarity ratio and time similarity ratio based on the similarity principle, and determining each similarity ratio by combining the spatial size limitation of the indoor test platform and the load range of the loading equipment. The specific values ​​are as follows: S43, the indoor physical model is constructed using a layered casting method. Similar materials are laid in sequence according to the designed geological layering order. After each layer is laid, the compaction degree is controlled by a vibration compaction device to ensure uniform density inside the model and avoid segregation. S44, multiple sets of sensing elements are preset inside the physical model, including strain gauges, displacement gauges and pore water pressure sensors. The placement and number of sensors are determined according to the calibration stress area predicted by numerical simulation. The sensor leads are led out of the model through a sealed channel to avoid damage during the test.

[0039] Specifically, step S4 involves constructing an indoor physical model consistent with the mechanical properties of the prototype through four sub-steps. S41 employs orthogonal experimental design to determine key influencing factors of similar materials, selecting aggregate particle size, binder content, moisture content, and compaction degree as experimental variables. Each variable has 4-5 levels, resulting in 30-40 mix design schemes. Standard specimens are prepared for each scheme, and mechanical properties such as compressive strength, shear strength, and elastic modulus are tested with a precision controlled within ±0.1 MPa. By comparing the mechanical properties of the specimens with those of the prototype soil and rock, the basic mix design with the highest matching degree is selected. S32 derives the conversion relationships between various similarity ratios based on the principle of similarity. Considering the spatial size limitations of the indoor test platform (length × width × height) and the maximum load range of the loading equipment, the geometric similarity ratio is comprehensively determined to be 1:20-1:50, the stress similarity ratio to be 1:10-1:30, the strain similarity ratio to be 1:1, and the time similarity ratio to be 1:5-1:10, ensuring that each similarity ratio satisfies the mechanical similarity relationship. S43 employs a layered casting method to construct the indoor physical model. Similar materials are laid sequentially according to the designed geological stratification order, with each layer controlled to a thickness of 5-10 cm. After laying, a vibratory compaction device is used for compaction, with a compaction frequency set to 50-60 Hz to ensure uniform density within the model, with a density deviation not exceeding ±2%, thus avoiding segregation. S44, based on the key stress areas predicted by numerical simulation, pre-sets multiple sets of sensing elements within the physical model. The density of strain gauges, displacement gauges, and pore water pressure sensors is 3-5 per cubic meter. Sensor leads are led out of the model through pre-set sealed channels with an IP67 sealing rating to prevent damage to the sensors and leads from moisture, similar material particles, etc., during the experiment, ensuring stable acquisition of sensor data.

[0040] Preferred, such as Figure 4 As shown, step S5 includes the following sub-steps: S51, real-time acquisition of strain, displacement, and pore water pressure data from the physical model experiment using a data acquisition card, conversion of analog signals to digital signals via A / D conversion, and transmission to the data processing center via Ethernet; S52, establishment of a communication connection between the numerical simulation system and the physical experiment system based on the TCP / IP protocol, setting the data transmission frequency and data format to ensure the real-time performance and accuracy of data transmission between the two; S53, use of the Kalman filter algorithm to filter out noise data during transmission, and prediction of the optimal estimated value of the data by establishing state equations and observation equations, eliminating the impact of environmental interference and equipment errors on data quality; S54, comparison and analysis of the filtered physical experiment data and numerical simulation results, calculation of the deviation between the two, and adjustment of constitutive parameters, boundary conditions, and initial stress field distribution in the numerical model based on the deviation feedback signal to keep the numerical simulation results synchronized with the physical experiment response.

[0041] Specifically, step S5 achieves synchronous coordination between numerical simulation and physical experiment through four sub-steps. S51 utilizes a data acquisition card to collect strain, displacement, and pore water pressure data from the physical model experiment in real time. The sampling frequency of the acquisition card is set to 50-100Hz, with an acquisition accuracy of ±0.01mm. The analog signals are converted to digital signals via A / D conversion with a conversion accuracy of 16 bits, and then transmitted to the data processing center via Ethernet using the TCP / IP protocol, with a transmission delay controlled within 50ms. S52 establishes a communication connection between the numerical simulation system and the physical experiment system based on the TCP / IP protocol. The data transmission frequency is set to 50Hz, and the data format is standardized as a standard structured format. Field definitions and unit identifiers for data transmission are clearly defined. A data verification mechanism ensures the real-time performance and accuracy of the transmitted data, with a data transmission success rate of no less than 99.9%. S53 employs a Kalman filter algorithm to filter out noise data during transmission. By establishing state equations and observation equations, setting the filter gain coefficient, and predicting the optimal estimated value of the data, the algorithm effectively eliminates the impact of environmental temperature and humidity changes and equipment errors on data quality, improving the data signal-to-noise ratio to over 30dB. S54 performs point-by-point comparison and analysis between the filtered physical test data and the numerical simulation results, and calculates the deviation values ​​between the two in terms of mechanical parameters, displacement response, etc. When the deviation value exceeds 5%, the constitutive parameters, boundary conditions and initial stress field distribution in the numerical model are adjusted based on the deviation feedback signal. The adjustment range is dynamically determined according to the deviation ratio. After each adjustment, the numerical calculation is re-performed until the deviation value between the numerical simulation results and the physical test response is controlled within 5%, so as to achieve synchronous coordination between the two.

[0042] like Figure 5As shown, an indoor twin slope physical model test method is implemented through different units, including: a slope prototype multi-source data acquisition unit, composed of a 3D laser scanner, ground-penetrating radar, stress sensor, displacement sensor, and pore water pressure sensor, which comprehensively collects the geometric parameters and mechanical response data of the slope prototype through distributed deployment, and establishes a wired communication connection between the data output end and the data transmission unit; a slope digital twin modeling unit, equipped with 3D modeling software and a geological data processing module, which receives the raw data transmitted by the data acquisition unit, constructs a digital mapping model of the slope prototype through geometric reconstruction algorithm and geological information fusion technology, and outputs the modeling results to the numerical simulation unit; and a slope numerical simulation calculation unit, which integrates a finite element-discrete element coupled calculation module and an adaptive solver, performs slope mechanical behavior simulation calculations based on the digital twin model, and transmits the data through a wired communication connection between the data output end and the data transmission unit; The system enables real-time data exchange with the physical model testing unit via a data interaction interface. The indoor physical model testing unit, including a similar material preparation device, model casting mold, loading system, and sensing module, constructs an indoor slope model based on similarity principles and conducts mechanical tests. Test data, after preprocessing, is transmitted to the data fusion and analysis unit. The data fusion and analysis unit, equipped with a multi-source data verification module, error analysis module, and database storage module, receives numerical simulation data and physical test data. After processing through a data fusion algorithm, it forms a standardized dataset, which is then output to the results application unit. The results application unit, equipped with a slope stability evaluation model and data visualization module, receives the fused and analyzed dataset, generates a slope safety assessment report, and displays it in graphical form, providing technical support for engineering decisions. All units establish data transmission links via industrial Ethernet for data interaction and collaborative work.

[0043] An indoor twin slope physical model test method is proposed. It uses a multi-source sensor array to achieve comprehensive acquisition of slope prototype data, combines digital twin technology to create a geometric mapping model that accurately matches the prototype, and uses coupling algorithms and adaptive optimization technology to improve the dynamic replication capability of numerical simulation. At the same time, it optimizes the material ratio and structural design of the indoor physical model based on the similarity principle. Through the coordinated operation of six all-round functional units, a closed-loop technology chain is formed from data acquisition, modeling and simulation to experimental verification and result application, so as to achieve a comprehensive and high-precision restoration of slope mechanical behavior and deformation and failure process.

[0044] This method establishes a real-time interactive feedback mechanism between numerical simulation and physical experiment, dynamically adjusts the parameters of both through data transmission interfaces and dynamic correction technology, and significantly improves data consistency and reliability by combining multi-source data fusion and verification mechanisms, fully leveraging the synergistic advantages of digital twin technology. Addressing the issues of model adaptability and accuracy deficiencies, it improves the characterization of complex geological structures through detection technology, optimizes the numerical model and similar material ratios using grid densification and orthogonal experiments, precisely deploys sensing elements in key areas, and employs noise filtering and error correction techniques to effectively reduce the deviation between experimental results and the actual prototype. This completely overcomes the limitations of traditional methods in complex geological conditions, providing stronger technical support for slope engineering safety early warning.

[0045] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0046] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for indoor twin slope physical model test, characterized in that, Includes the following steps: S1 uses a multi-dimensional sensor array to collect the mechanical parameters, displacement field distribution and pore water pressure changes of the slope prototype in real time, forming a raw dataset including multi-source heterogeneous data. S2, based on digital twin technology, constructs a geometric mapping model of the slope prototype, obtains the surface topology of the prototype through three-dimensional laser scanning, and supplements the internal structural information by combining geological drilling data to establish a geometric digital representation that matches the prototype 1:

1. S3 uses a coupled finite element and discrete element algorithm to numerically simulate the mechanical behavior of slopes. By dividing unstructured meshes to characterize complex geological interfaces, constitutive relation parameters that conform to the actual prototype are set to dynamically replicate the slope deformation and failure process. S4. Build an indoor physical model test platform, determine the geometric similarity ratio, mechanical similarity ratio and time similarity ratio according to the similarity principle, select similar materials with the same mechanical properties as the prototype rock and soil, and pour them into an indoor slope model according to the design ratio. S5, simultaneously starts numerical simulation and physical model test, and conducts real-time interactive feedback of mechanical parameters, displacement response and failure mode between the two through data transmission interface. The numerical model parameters are adjusted based on dynamic correction algorithm to keep the response characteristics of the two consistent. S6 integrates and analyzes the numerical simulation data and physical test data collected during the experiment, filters valid data through a data verification mechanism, and constructs a slope stability evaluation database to provide data support for slope engineering safety early warning.

2. The indoor twin slope physical model test method according to claim 1, characterized in that, The constitutive relation parameters in S3 are determined by the following expression: ， wherein, is the shear stress, is the cohesion, is the normal stress, is the internal friction angle, is the non-linear deformation coefficient, is the axial normal strain, is the shear strain.

3. The indoor twin slope physical model test method according to claim 1, characterized in that, The proportion of similar materials in S4 is calculated using the following expression: ， wherein, is the mass proportion of the i-th component, is the prototype rock-soil density, is the prototype rock-soil volume, is the mechanical property matching coefficient of the i-th component, is the j-th similar material component density, is the j-th similar material component volume, and n is the total number of similar material components.

4. The indoor twin slope physical model test method according to claim 1, characterized in that, The dynamic correction algorithm in S5 is implemented based on the following expression: ， wherein, is the numerical model parameter at the kth time instant after correction, is the numerical model parameter at the kth time instant after correction, is the numerical model parameter at the kth time instant, is the numerical model parameter at the kth time instant, is the numerical model parameter at the kth time instant, is the proportional correction coefficient, is the integral correction coefficient, is the integral correction coefficient, E is the response error of numerical simulation and physical test, and t is the test time.

5. The indoor twin slope physical model test method according to claim 1, characterized in that, The accuracy verification of the geometric mapping model in S2 is calculated using the following expression: ， wherein, is the geometric mapping precision error, M is the number of surface topological feature points, is the i-th feature point coordinate of the model surface, is the i-th feature point coordinate of the prototype surface, N is the number of internal structure feature points, is the j-th feature point coordinate of the model internal structure, is the j-th feature point coordinate of the prototype internal structure.

6. The indoor twin slope physical model test method according to claim 1, characterized in that, The data fusion analysis in the S6 is realized by the following expression: , wherein, is the fused data, is a data weight coefficient, is the numerical simulation data, is the physical test data, is an attenuation coefficient, is a phase adjustment coefficient.

7. The indoor twin slope physical model test method according to claim 1, characterized in that, S3 includes the following steps: S31: Spatial distribution information of faults and weak interlayers hidden structures inside the slope prototype is obtained by ground-penetrating radar detection. Combined with seismic wave test data, the differences in mechanical parameters of different geological layers are determined, and an initial numerical calculation model including complex geological structures is established. S32 employs adaptive mesh refinement technology to refine the mesh for potential sliding surfaces and stress concentration areas of slopes. It uses an error estimation function to determine the rationality of mesh division and readjusts the mesh density for areas that do not meet the calculation accuracy requirements. S33, based on the coupling interface between FLAC3D and PFC3D, performs collaborative calculations of the mechanical behavior of continuous and discrete media, sets the force transmission coefficient and displacement coordination conditions of the coupling interface, and ensures the continuity of data interaction between different calculation modules. S34 introduces an adaptive time step adjustment mechanism, which dynamically optimizes the calculation time step based on the slope deformation rate. When the deformation rate exceeds the threshold, the time step is reduced, and when the deformation rate is below the threshold, the time step is increased, thereby improving calculation efficiency while ensuring calculation accuracy.

8. The indoor twin slope physical model test method according to claim 1, characterized in that, S4 includes the following steps: S41, through orthogonal experimental design, the calibration influencing factors of similar materials were determined. Aggregate particle size, binder content, moisture content and compaction degree were selected as experimental variables. Multiple mix proportion schemes at different levels were designed. The basic mix proportion with the highest matching degree with the prototype soil and rock mass was screened through mechanical performance testing. S42. Based on the principle of similarity, the conversion relationship between geometric similarity ratio, stress similarity ratio, strain similarity ratio and time similarity ratio is derived. Combined with the spatial size limitations of the indoor test platform and the load range of the loading equipment, the specific values ​​of each similarity ratio are determined. S43 uses a layered casting method to construct an indoor physical model. Similar materials are laid in sequence according to the designed geological layering order. After each layer is laid, the compaction degree is controlled by a vibratory compaction device to ensure uniform density inside the model and avoid segregation. S44 has multiple sets of sensing elements pre-set inside the physical model, including strain gauges, displacement gauges and pore water pressure sensors. The placement and number of sensors are determined based on the calibrated stress area predicted by numerical simulation. The sensor leads are led out of the model through a sealed channel to avoid damage during the experiment.

9. The indoor twin slope physical model test method according to claim 1, characterized in that, S5 includes the following steps: S51 collects strain, displacement and pore water pressure data in physical model tests in real time through a data acquisition card, converts analog signals into digital signals through A / D conversion, and transmits them to the data processing center via Ethernet. S52 establishes a communication connection between the numerical simulation system and the physical experiment system based on the TCP / IP protocol, sets the data transmission frequency and data format, and ensures the real-time performance and accuracy of data transmission between the two. S53 uses the Kalman filter algorithm to filter out noise data during transmission. By establishing state equations and observation equations, it predicts the optimal estimate of the data and eliminates the impact of environmental interference and equipment errors on data quality. S54 compares and analyzes the filtered physical test data with the numerical simulation results, calculates the deviation between the two, and adjusts the constitutive parameters, boundary conditions and initial stress field distribution in the numerical model based on the deviation feedback signal, so that the numerical simulation results and the physical test response remain synchronized.

10. A method for indoor physical model testing of twin slopes according to any one of claims 1-9, characterized in that, This method is implemented through different units, including: The multi-source data acquisition unit for the slope prototype consists of a 3D laser scanner, ground-penetrating radar, stress sensor, displacement sensor and pore water pressure sensor. It comprehensively acquires the geometric parameters and mechanical response data of the slope prototype through distributed deployment. The data output end establishes a wired communication connection with the data transmission unit. The slope digital twin modeling unit is equipped with 3D modeling software and a geological data processing module. It receives the raw data transmitted by the data acquisition unit, constructs a digital mapping model of the slope prototype through geometric reconstruction algorithm and geological information fusion technology, and outputs the modeling results to the numerical simulation unit. The slope numerical simulation calculation unit integrates a finite element-discrete element coupled calculation module and an adaptive solver. It performs slope mechanical behavior simulation calculations based on a digital twin model. During the calculation process, it communicates with the physical model test unit in real time through a data interaction interface. The indoor physical model test unit includes a similar material preparation device, a model casting mold, a loading system and a sensing and detection module. It constructs an indoor slope model according to the similarity principle and conducts mechanical tests. The test data is preprocessed and then transmitted to the data fusion and analysis unit. The data fusion and analysis unit is equipped with a multi-source data verification module, an error analysis module, and a database storage module. It receives numerical simulation data and physical experiment data, processes them through a data fusion algorithm to form a standardized dataset, and outputs it to the result application unit. The results application unit is equipped with a slope stability evaluation model and a data visualization module. It receives the dataset after fusion analysis, generates a slope safety assessment report and displays it in the form of charts, providing technical support for engineering decision-making. Each unit establishes a data transmission link through industrial Ethernet to conduct data interaction and collaborative work.

Images