Method, device and medium for analyzing time-space evolution characteristics of landslide body based on parameter inversion

Abstract

This application proposes an analytical method, equipment, and medium for analyzing the spatiotemporal evolution characteristics of landslides based on parameter inversion. The method includes: collecting surface deformation monitoring data and internal deformation monitoring data of the landslide body, fusing the data to obtain multi-source monitoring data; establishing a three-dimensional numerical simulation model based on the geological information of the landslide body, inputting the generated mechanical parameter combination samples into the three-dimensional numerical simulation model to obtain a response dataset; constructing a surrogate model, training the surrogate model using the response dataset, and establishing a nonlinear mapping relationship between the mechanical parameter combination and landslide displacement; using the multi-source monitoring data as the input of the objective function, performing parameter inversion using an optimization algorithm combined with the nonlinear mapping relationship to obtain the optimal mechanical parameter combination to characterize the spatiotemporal evolution characteristics of the landslide body. This method can accurately obtain the spatiotemporal evolution characteristics of landslide bodies based on multi-source monitoring data, improve the accuracy of parameter inversion, and enhance the agreement between simulation results and measured data.

Description

Technical Field

[0001] This application relates to the field of landslide geological hazard monitoring and early warning technology, and in particular to an analysis method, equipment and medium for the spatiotemporal evolution characteristics of landslide bodies based on parameter inversion. Background Technology

[0002] Currently, landslide geological hazards pose a significant safety risk to hydropower projects, transportation facilities, and urban construction. Monitoring, early warning, and prevention technologies for landslides have always been core research topics in engineering geology and geotechnical mechanics. With the expansion of hydropower station reservoir areas and new energy power plant construction, the spatiotemporal evolution of large landslides is becoming increasingly complex due to multiple factors, including human disturbance, reservoir impoundment, and water level fluctuations.

[0003] However, the landslide identification and static treatment schemes in related technologies directly use single monitoring data or empirical analog parameters for stability evaluation, without effectively integrating the nonlinear mapping relationship between multi-source monitoring data and soil mechanical parameters. This may lead to inaccurate characterization of landslide deformation features and delayed acquisition of mechanical parameters, thereby affecting the dynamic prevention and control and treatment decisions for the safety and stability of landslides throughout their entire life cycle. Summary of the Invention

[0004] This application aims to at least partially address one of the technical problems in the related art.

[0005] Therefore, the first objective of this application is to propose an analytical method for the spatiotemporal evolution characteristics of landslides based on parameter inversion.

[0006] The second objective of this application is to propose an electronic device.

[0007] The third objective of this application is to provide a computer-readable storage medium.

[0008] To achieve the above objectives, the first aspect of this application is to propose an analytical method for the spatiotemporal evolution characteristics of landslides based on parameter inversion, comprising the following steps:

[0009] Surface deformation monitoring data and internal deformation monitoring data of the landslide body are collected, and the surface deformation monitoring data and internal deformation monitoring data are fused to obtain multi-source monitoring data; A three-dimensional numerical simulation model is established based on the geological information of the landslide body, and multiple sets of mechanical parameter combination samples are generated. Each set of mechanical parameter combination samples is input into the three-dimensional numerical simulation model for deformation simulation to obtain the response dataset between the mechanical parameter combination and the landslide displacement. A proxy model is constructed, and the proxy model is trained using the response dataset to establish a nonlinear mapping relationship between the combination of mechanical parameters and the landslide displacement; Using the multi-source monitoring data as input to the objective function, the optimization algorithm is combined with the nonlinear mapping relationship to perform parameter inversion and search for the optimal combination of mechanical parameters to characterize the spatiotemporal evolution characteristics of the landslide body.

[0010] To achieve the above objectives, a second aspect of this application also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the analysis method for the spatiotemporal evolution characteristics of landslide bodies based on parameter inversion as described in any one of the first aspects above.

[0011] To achieve the above objectives, the third aspect of this application also proposes a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the analysis method for the spatiotemporal evolution characteristics of landslide bodies based on parameter inversion as described in any one of the first aspects.

[0012] The technical solutions provided by the embodiments of this application bring at least the following beneficial effects: This application, combining multi-source monitoring data, can accurately reproduce the deformation characteristics of landslides through a rapid inversion method of soil and rock mechanical parameters, demonstrating the advantages of constructing a multi-dimensional monitoring system and providing prior information for the dynamic prevention and control of landslides. This application can improve the accuracy of landslide spatiotemporal evolution characteristic analysis, effectively reproduce landslide deformation characteristics, and provide a reliable basis for the dynamic prevention and control of landslides. Therefore, this application can accurately obtain the spatiotemporal evolution characteristics of landslides based on multi-source monitoring data, improve parameter inversion accuracy, and enhance the consistency between simulation results and measured data.

[0013] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0014] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart illustrating an analysis method for the spatiotemporal evolution characteristics of landslides based on parameter inversion, as proposed in an embodiment of this application; Figure 2 This is a schematic diagram illustrating the principle of an analysis method for the spatiotemporal evolution characteristics of landslides based on parameter inversion, as proposed in an embodiment of this application. Figure 3 This is a schematic diagram of a deformed time series curve proposed in an embodiment of this application; Figure 4(a) is a schematic diagram of the parametric inversion simulation results using appearance data alone, as proposed in an embodiment of this application. Figure 4(b) is a schematic diagram of the parametric inversion simulation results using only internal observation data according to an embodiment of this application; Figure 4(c) is a schematic diagram of the parameter inversion simulation results of multi-source monitoring data proposed in the embodiment of this application. Detailed Implementation

[0015] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0016] The following description, with reference to the accompanying drawings, describes an analytical method for analyzing the spatiotemporal evolution characteristics of landslides based on parameter inversion, as proposed in an embodiment of this application.

[0017] Example 1 Figure 1 This is a flowchart illustrating an analysis method for the spatiotemporal evolution characteristics of landslides based on parameter inversion, as proposed in an embodiment of this application. Figure 1 As shown, the method includes the following steps: Step S101: Collect surface deformation monitoring data and internal deformation monitoring data of the landslide body, and fuse the surface deformation monitoring data and internal deformation monitoring data to obtain multi-source monitoring data.

[0018] Specifically, this step involves collecting surface deformation monitoring data and internal deformation monitoring data of the landslide body, and then fusing these two types of data to form a multi-source monitoring dataset that comprehensively reflects the spatiotemporal evolution characteristics of the landslide body. Surface deformation monitoring data characterizes the displacement field distribution on the landslide surface and its temporal variation, while internal deformation monitoring data reveals the deformation state of different strata of rock and soil within the deeper layers of the landslide body and the development of the sliding surface. By fusing these two types of data, the limitations of single monitoring methods in terms of spatial coverage or detection depth can be overcome, thereby constructing a multi-dimensional, three-dimensional monitoring information system covering both the surface and interior of the landslide body, as well as macroscopic and microscopic aspects. This fusion process includes unifying the spatiotemporal references of data from different monitoring sources, standardizing data formats, and identifying and removing outliers to ensure the consistency and reliability of the multi-source monitoring data in subsequent analyses.

[0019] As one possible implementation method, surface deformation monitoring data can be obtained through time-series synthetic aperture radar interferometry, while internal deformation monitoring data can be obtained through borehole inclinometers. The two types of data are then integrated according to a unified spatiotemporal coordinate system to form a multi-source monitoring dataset that includes surface displacement time series and deep displacement profiles.

[0020] Therefore, by collecting and integrating surface deformation monitoring data and internal deformation monitoring data, this step can comprehensively obtain the spatial deformation distribution characteristics and temporal evolution trend of the landslide body, providing an accurate and complete data foundation for subsequent parameter inversion analysis based on multi-source monitoring data, thereby improving the reliability and accuracy of the analysis of the spatiotemporal evolution characteristics of the landslide body.

[0021] Step S102: Establish a three-dimensional numerical simulation model based on the geological information of the landslide body, and generate multiple sets of mechanical parameter combination samples. Input each set of mechanical parameter combination samples into the three-dimensional numerical simulation model for deformation simulation to obtain the response dataset between the mechanical parameter combination and the landslide displacement.

[0022] Specifically, this step aims to construct a numerical foundation that can reflect the quantitative relationship between the deformation response of a landslide and its mechanical parameters. First, based on geological information such as the topography, stratigraphic structure, and distribution of slip zones or weak interlayers of the landslide, a three-dimensional numerical simulation model is established. This model includes settings for mesh generation, material partitioning, boundary conditions, and calculation time histories to simulate the deformation process of the landslide under the influence of its own weight, water level fluctuations, and creep effects.

[0023] Furthermore, multiple sets of mechanical parameter combinations are generated, covering the value space of the mechanical parameters to be inverted, including but not limited to elastic modulus, cohesion, internal friction angle, and creep model parameters. The samples can be generated using orthogonal experimental design, random sampling, or a combination of both to ensure global homogeneity and local diversity within the parameter space.

[0024] Subsequently, the samples of each set of mechanical parameter combinations are sequentially input into the three-dimensional numerical simulation model. While maintaining consistency in model geometry, mesh, boundary conditions, and load conditions, long-term deformation forward simulation calculations are performed to obtain the corresponding landslide displacement field. Based on the spatial locations of external and internal monitoring points, surface and deep displacements at corresponding nodes or interpolation locations are extracted from the displacement field. This establishes a one-to-one correspondence between each set of mechanical parameter combinations and the corresponding displacement response. The records are then filtered, normalized, and formatted to ultimately generate a response dataset for subsequent surrogate model training.

[0025] As one possible implementation method, 500 sets of parameter combination samples can be generated based on a combination of orthogonal experimental design and random sampling. Each set of samples is then input into a three-dimensional numerical simulation model to extract displacement data from 7 external InSAR monitoring points and 5 borehole inclination monitoring points at different depths, forming a response dataset.

[0026] Therefore, this step, by constructing a three-dimensional numerical simulation model and generating a response dataset, provides a reliable training data foundation for establishing the nonlinear mapping relationship between mechanical parameters and landslide displacement, thereby ensuring the accuracy and efficiency of the parameter inversion process and enabling the inversion results to effectively reflect the actual deformation characteristics of the landslide body.

[0027] Step S103: Construct a surrogate model, train the surrogate model using the response dataset, and establish a nonlinear mapping relationship between the combination of mechanical parameters and the landslide displacement.

[0028] Specifically, after obtaining the response dataset between the combination of mechanical parameters and the landslide displacement, a surrogate model is constructed that can characterize the complex mapping relationship contained in the response dataset. The surrogate model is then trained using the response dataset to establish a nonlinear mapping relationship between the combination of mechanical parameters and the landslide displacement.

[0029] The surrogate model is essentially a computational entity obtained through a data-driven learning process, capable of approximating the original high-fidelity numerical simulation model. Its core function is to receive a set of mechanical parameter combinations as input and quickly and accurately output the corresponding landslide displacement response. The construction and training process of the surrogate model aims to learn and generalize its inherent laws from known input-output sample pairs, enabling the model to efficiently predict landslide displacements caused by any given combination of mechanical parameters without repeatedly calling computationally expensive three-dimensional numerical simulation models. The establishment of this nonlinear mapping relationship provides a computational foundation for subsequently rapidly evaluating the deformation response corresponding to different parameter combinations in a vast parameter space, thus transforming the iterative optimization process, which originally relied on numerous forward simulations, into a rapid invocation of the surrogate model.

[0030] As one possible implementation, the surrogate model can be implemented using a multilayer feedforward neural network based on the backpropagation algorithm. The neural network includes an input layer, a hidden layer, and an output layer. The input layer nodes correspond to the mechanical parameters to be inverted, and the output layer nodes correspond to the displacements of the monitoring points. The network is trained by using parameter combination samples from the response dataset as input and the corresponding displacement results as the expected output. The backpropagation algorithm is then used to adjust the network weights, ultimately enabling the trained neural network to accurately map the nonlinear relationship between the combination of mechanical parameters and the landslide displacement.

[0031] Therefore, this step, by constructing and training a surrogate model to establish a nonlinear mapping relationship between the combination of mechanical parameters and landslide displacement, significantly reduces the consumption of computational resources in the subsequent parameter inversion process, realizes rapid prediction of landslide displacement response, and thus greatly improves the efficiency of global search and optimization in complex parameter space, providing reliable computational support for accurately and efficiently obtaining the optimal combination of mechanical parameters characterizing the spatiotemporal evolution of landslide bodies.

[0032] Step S104: Using multi-source monitoring data as input to the objective function, the optimization algorithm combined with nonlinear mapping relationship is used to perform parameter inversion and search for the optimal combination of mechanical parameters to characterize the spatiotemporal evolution characteristics of the landslide body.

[0033] Specifically, this step aims to construct an objective function with the core objective of minimizing the difference between measured multi-source monitoring data and predicted data, and then use an intelligent optimization algorithm to efficiently search in the parameter space to determine a set of mechanical parameters that best reflects the current actual deformation state of the landslide.

[0034] The objective function quantifies the deviation between the displacement response predicted by the surrogate model based on a set of candidate mechanical parameters and the measured displacement obtained from multi-source monitoring data. Its form can be the norm or weighted norm of the error between the predicted and measured values, where the weights can be set according to the reliability or importance of data from different monitoring points or different monitoring types. The optimization algorithm iteratively searches for the global optimum in a multi-dimensional parameter space defined by the mechanical parameters to be inverted (such as elastic modulus, cohesion, internal friction angle, and creep model parameters), based on the objective function value.

[0035] During the search process, the optimization algorithm calls a pre-trained surrogate model to quickly evaluate the displacement response corresponding to each candidate parameter combination, thus avoiding the direct use of computationally expensive 3D numerical simulation models and significantly improving inversion efficiency. The candidate parameter combinations are iteratively updated until the objective function meets the preset convergence condition or reaches the maximum number of iterations. Finally, the parameter combination with the minimum objective function value is determined as the optimal mechanical parameter combination.

[0036] As one possible implementation method, a particle swarm optimization algorithm based on population clustering and gradient adaptation can be used. This algorithm initializes the particle swarm, calculates particle fitness, iteratively updates particle positions, and combines population clustering to maintain diversity and adaptively adjusts inertia weights and acceleration coefficients to balance global and local search capabilities. Finally, it determines the optimal combination of parameters, and the parameters in this combination serve as key physical parameters characterizing the spatiotemporal evolution of the landslide body.

[0037] Therefore, this step, by combining multi-source monitoring data with surrogate models and optimization algorithms, achieves efficient and accurate inversion of the mechanical parameters of landslides. Its beneficial effects are as follows: on the one hand, it can make full use of multi-source monitoring information from the surface and interior, so that the inversion results can more comprehensively reflect the true deformation state of the landslide and improve the accuracy and reliability of parameter identification; on the other hand, by using surrogate models to replace high-cost three-dimensional numerical simulations, the computation time of parameter inversion is greatly reduced, providing an efficient technical means for the rapid acquisition and dynamic analysis of the spatiotemporal evolution characteristics of landslides.

[0038] The following describes in detail the specific implementation process of each step in the method for analyzing the spatiotemporal evolution characteristics of landslides based on parameter inversion of this application, using several specific embodiments.

[0039] Example 2 Based on the above embodiments, this embodiment provides a detailed description of the specific implementation of step S101, "collecting surface deformation monitoring data and internal deformation monitoring data of the landslide body, and fusing the surface deformation monitoring data and the internal deformation monitoring data to obtain multi-source monitoring data".

[0040] In this embodiment, step S101 specifically includes acquiring surface deformation monitoring data, acquiring internal deformation monitoring data, and fusing multi-source monitoring data. First, regarding the acquisition of surface deformation monitoring data, ascending and descending orbit SAR images of the landslide body are obtained using time-series InSAR technology. For example, 73 ascending and descending orbit SAR images collected by a satellite can be obtained from a space agency's science center to overcome the limitations of single orbit SAR data and SAR imaging geometry.

[0041] Subsequently, the orbital error of the satellite data was corrected using precise orbital data provided by a space agency, and the terrain phase was eliminated and geocoded using a 30-meter resolution space radar altimeter digital elevation model, thereby obtaining landslide surface deformation monitoring data, which includes deformation rate distribution maps and deformation time series curves of monitoring points.

[0042] Furthermore, regarding the acquisition of internal deformation monitoring data, displacement data at different depths within the landslide body are obtained using borehole inclinometers. For example, multiple boreholes ranging from 51.64 meters to 80.49 meters in depth are drilled into the landslide body, and inclinometers are installed inside the boreholes to monitor displacement changes at different depths within the landslide body, thereby obtaining internal deformation monitoring data.

[0043] Finally, the surface deformation monitoring data and internal deformation monitoring data obtained above were spatiotemporally aligned and fused to obtain multi-source monitoring data. This multi-source monitoring data comprehensively reflects the deformation characteristics of the landslide body from the surface to its depths, providing a comprehensive data foundation for subsequent parameter inversion analysis.

[0044] Therefore, through the above specific implementation methods, this application can comprehensively utilize InSAR technology and borehole inclinometers to obtain deformation data of the landslide surface and interior, and form multi-source monitoring data through spatiotemporal alignment and fusion, thereby providing more comprehensive and accurate input information for subsequent parameter inversion analysis, and improving the reliability and accuracy of landslide spatiotemporal evolution characteristic analysis.

[0045] Example 3 Based on the above embodiments, this embodiment provides a detailed description of the specific implementation of step S102, which involves "establishing a three-dimensional numerical simulation model based on the geological information of the landslide body, generating multiple sets of mechanical parameter combination samples, inputting each set of mechanical parameter combination samples into the three-dimensional numerical simulation model for deformation simulation, and obtaining a response dataset between the mechanical parameter combination and the landslide displacement".

[0046] In this embodiment, firstly, a constitutive model of the rock mass is selected based on the geological information of the landslide, and a three-dimensional finite element numerical simulation model is established. Specifically, based on geological exploration data, topography, stratigraphic structure, and the distribution of slip zones or weak interlayers, the constitutive relationships of each soil and rock layer in the landslide are determined, for example, using a Mohr-Coulomb model or a creep model. Based on this, a three-dimensional finite element model is established, including mesh generation, material partitioning, boundary conditions, and calculation time history settings, to reflect the deformation process of the landslide under the combined effects of its own weight, water level fluctuations, and creep.

[0047] Then, a combination of orthogonal experimental design and random sampling is used to generate multiple sets of mechanical parameter combination samples. In one possible implementation, the process includes: determining the parameters to be inverted and their value ranges, including elastic modulus, cohesion, internal friction angle, and creep model parameters; determining the upper and lower limits of each parameter based on experimental values ​​or engineering analogies, combined with the long-term deformation characteristics of the landslide body, to form parameter value intervals; dividing each parameter into several levels and selecting corresponding orthogonal tables to ensure that the combinations of parameters at different levels are evenly distributed, constructing a basic sample covering the main parameter space; using this basic sample as a framework, random sampling and perturbation are performed within the value intervals corresponding to each parameter to generate continuously distributed parameter values, thereby obtaining, for example, 500 sets of parameter combination samples, so that the samples have both global uniformity and local randomness and diversity.

[0048] Finally, the samples of each set of mechanical parameter combinations are sequentially input into the three-dimensional numerical simulation model for long-term deformation forward simulation. Surface and deep displacements at corresponding external and internal monitoring points are extracted to generate a response dataset. Specifically, for each set of parameter combinations, the elastic modulus, cohesion, internal friction angle, and creep model parameters are written into the three-dimensional numerical simulation model according to the soil layer or material zoning number. Calculations are performed while maintaining consistency in model geometry, mesh, boundary conditions, and load conditions to obtain the corresponding landslide displacement field. Subsequently, based on the spatial locations of the external InSAR monitoring points and the internal borehole inclination monitoring points, the surface and deep displacements at corresponding nodes or interpolation locations are extracted. Each set of parameter combinations is used as input, and the corresponding extracted surface and deep displacements are used as outputs to establish a one-to-one data record. Samples with non-convergent calculations or abnormal displacements are removed. The retained samples are numbered, normalized, and formatted to generate a "Mechanical Parameter Combination - Landslide Displacement" response dataset for surrogate model training.

[0049] Therefore, this embodiment generates parameter combination samples by combining orthogonal experimental design with random sampling, which not only ensures the global uniform coverage of the samples in the parameter space, but also enhances the local randomness and diversity, thereby improving the representativeness of the response dataset and the accuracy of surrogate model training, and providing a reliable data foundation for subsequent parameter inversion.

[0050] Example 4 Based on the above embodiments, this embodiment provides a detailed description of the specific implementation of step S103, which involves "constructing a surrogate model, training the surrogate model using the response dataset, and establishing a nonlinear mapping relationship between the combination of mechanical parameters and the landslide displacement".

[0051] In this embodiment, the surrogate model constructed in step S103 specifically employs a multilayer feedforward neural network based on the error backpropagation algorithm. This network structure includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer corresponds to the number of mechanical parameters to be inverted. Specifically, the input layer nodes receive the elastic modulus, cohesion, internal friction angle, and creep model parameters of different soil and rock layers; these parameters constitute each set of mechanical parameter combinations. The number of nodes in the output layer corresponds to the number of displacements at the selected external and internal monitoring points, and is used to output the displacement response of the landslide body under the corresponding parameter combinations. The external monitoring points correspond to InSAR monitoring points in the surface deformation monitoring data, and the internal monitoring points correspond to borehole inclination monitoring points in the internal deformation monitoring data. The hidden layer is located between the input and output layers and is used to characterize the complex nonlinear relationship between the mechanical parameters and the landslide displacement.

[0052] When training this multilayer feedforward neural network using the response dataset, each combination of mechanical parameters in the response dataset is first used as the input to the neural network, and the landslide surface displacement and deep displacement calculated by the corresponding three-dimensional numerical simulation model are used as the expected output of the neural network. Before training, the input parameters and output displacements are normalized to reduce the impact of differences in dimensions and orders of magnitude between different physical quantities on the training results. During training, parameter combination samples are passed from the input layer to the hidden layer and the output layer to obtain the displacement results predicted by the surrogate model, which are then compared with the three-dimensional numerical simulation displacement results in the response dataset to form the prediction error.

[0053] Subsequently, an error backpropagation algorithm is used to propagate the prediction error layer by layer from the output layer to the hidden and input layers. The network weights and biases are continuously adjusted based on the error gradient, allowing the surrogate model's output displacement to gradually approximate the 3D numerical simulation result. Once the training error meets the preset accuracy requirement or reaches the maximum number of iterations, the trained neural network surrogate model is obtained. This surrogate model can quickly predict the corresponding landslide displacement response based on the input combination of geotechnical mechanics parameters without repeatedly calling the 3D numerical simulation model, thus establishing a nonlinear mapping relationship between "parameter combination—landslide displacement".

[0054] Therefore, by using the multilayer feedforward neural network based on the error backpropagation algorithm as a surrogate model, this embodiment can efficiently and accurately establish the nonlinear mapping relationship between mechanical parameters and landslide displacement, providing a rapid computational basis for subsequent parameter inversion using intelligent optimization algorithms, and significantly improving the overall efficiency of parameter inversion.

[0055] Example 5 Based on the above embodiments, this embodiment provides a detailed description of the specific implementation of step S104, which involves "using the multi-source monitoring data as input to the objective function, employing an optimization algorithm in conjunction with the nonlinear mapping relationship to perform parameter inversion, and searching for the optimal combination of mechanical parameters to characterize the spatiotemporal evolution characteristics of the landslide body".

[0056] In this embodiment, step S104, which utilizes an optimization algorithm combined with a nonlinear mapping relationship to perform parameter inversion and search for the optimal combination of mechanical parameters, is specifically implemented as follows: First, the combination of parameters of the soil and rock mass to be inverted is used as particle positions. Each particle corresponds to a set of possible elastic modulus, cohesion, internal friction angle, and creep model parameters, and an initial particle swarm is randomly generated within the range of each parameter value. The generation of this initial particle swarm is based on the parameter value range determined in the aforementioned steps, ensuring that the particles are uniformly distributed within the feasible region.

[0057] Subsequently, for each particle, the nonlinear mapping relationship between the mechanical parameter combination and landslide displacement established by the trained surrogate model is invoked to quickly predict the displacement response of the corresponding external and internal monitoring points. Specifically, the parameter combination of the particle is input into the surrogate model, and the surrogate model outputs the predicted displacement value, which includes the surface displacement of external monitoring points (such as InSAR monitoring points) and the deep displacement of internal monitoring points (such as borehole inclination monitoring points). Next, the predicted displacement is compared with the measured displacement in the multi-source monitoring data, and the objective function is constructed using the L2 or weighted L2 of the error between the two. In this embodiment, the measured displacement data specifically includes the displacement data of 7 external InSAR monitoring points and 5 borehole inclination monitoring points at different depths. The smaller the objective function value, the closer the parameter combination is to the actual deformation state of the landslide.

[0058] Finally, the population is clustered based on particle fitness to maintain population diversity. Simultaneously, the inertia weights and acceleration coefficients of the particles are adaptively adjusted based on error reduction, achieving a dynamic balance between global and local searches. Each particle updates its velocity and position based on its individual optimal position, subgroup optimal position, and global optimal position, and its fitness is repeatedly calculated. When the objective function satisfies the convergence condition or reaches the maximum number of iterations, the position of the particle with the minimum objective function value is determined as the optimal combination of mechanical parameters. This optimal combination of mechanical parameters includes the elastic modulus, cohesion, internal friction angle, and creep model parameters for different soil and rock layers, used to characterize the spatiotemporal evolution characteristics of the landslide body.

[0059] Therefore, this specific implementation method introduces a particle swarm optimization algorithm based on population clustering and gradient adaptation, and combines it with a surrogate model to quickly predict displacement response. This enables efficient utilization of multi-source monitoring data and accurate inversion of mechanical parameters, significantly improving the efficiency and accuracy of searching for the optimal parameter combination, thereby more reliably obtaining the spatiotemporal evolution characteristics of the landslide body.

[0060] In this embodiment, the optimal mechanical parameter combination searched by the parameter inversion in step S104 through the optimization algorithm combined with the nonlinear mapping relationship specifically includes the elastic modulus, cohesion, internal friction angle and creep model parameters of different soil and rock layers, wherein the creep model parameters further include the first creep model parameter A and the second creep model parameter n.

[0061] Specifically, during particle swarm initialization, the position of each particle represents a set of soil and rock mechanical parameters to be inverted. This set encompasses the elastic modulus, cohesion, internal friction angle, and creep model parameters A and n for each soil and rock layer of the landslide. For example, for a landslide, its soil and rock layers can be divided into multiple strata such as Qdel1, Qal, Qdel2, STR, and WEAK. The aforementioned mechanical parameters of each stratum are used as variables to be inverted, collectively forming a high-dimensional parameter vector. When calculating particle fitness, for each particle's corresponding parameter combination, a trained surrogate model is first invoked. This surrogate model, built on a neural network, can quickly predict the displacement response of the landslide at external and internal monitoring points based on the input mechanical parameter combination.

[0062] Subsequently, the displacement response predicted by the surrogate model is compared with the measured displacement from the multi-source monitoring data obtained in step S101, and an objective function is constructed using the L2 or weighted L2 of the error between the two. The smaller the objective function value, the closer the combination of mechanical parameters is to the actual deformation state of the landslide. During the iterative update of particle positions, the particle swarm optimization algorithm clusters the population based on particle fitness and adaptively adjusts the inertia weight and acceleration coefficient of the particles to achieve dynamic balance between global and local searches. Each particle updates its velocity and position based on its individual optimal position, subgroup optimal position, and global optimal position, and repeatedly calculates its fitness. When the objective function meets the convergence condition or reaches the maximum number of iterations, the particle position with the smallest objective function value is determined as the optimal combination of mechanical parameters. The optimal combination of mechanical parameters is the key physical parameter characterizing the spatiotemporal evolution of the landslide. Its specific values ​​include the elastic modulus, cohesion, internal friction angle of each soil and rock layer, as well as creep model parameter A and creep model parameter n. These parameters can accurately reflect the deformation response of the landslide under different soil and rock mechanical properties, providing accurate input basis for subsequent input into the three-dimensional numerical simulation model to obtain the surface displacement, deep displacement and long-term deformation evolution process of the landslide.

[0063] Therefore, by specifically defining the optimal combination of mechanical parameters as the elastic modulus, cohesion, internal friction angle, and creep model parameters A and n of different soil and rock layers, this embodiment enables the parameter inversion results to comprehensively reflect the mechanical properties and creep behavior of each soil and rock layer of the landslide body, thereby more accurately depicting the spatiotemporal evolution of the landslide body and improving the physical significance and engineering applicability of the inversion results.

[0064] Example 6 This embodiment describes in detail the complete implementation process of the landslide spatiotemporal evolution characteristic analysis method based on parameter inversion of this application. This method uses a specific landslide as an application demonstration object, constructs a multi-dimensional monitoring system, integrates multi-source monitoring data, establishes a surrogate model, and uses an optimization algorithm for parameter inversion, ultimately obtaining the key physical parameters characterizing the spatiotemporal evolution characteristics of the landslide. The principle of the landslide spatiotemporal evolution characteristic analysis method of this application embodiment is as follows: Figure 2 As shown.

[0065] The first step involved the acquisition and fusion of multi-source monitoring data. The landslide body is approximately 450m to 800m wide (north-south) and 1300m long (east-west), with a total volume of about 20 million cubic meters. Locally, the overburden layer at the reservoir bank exhibits bank collapse and deformation. This embodiment constructs a multi-dimensional, integrated "space-ground" monitoring system for both internal and external observations. External observation utilizes time-series InSAR technology, while internal observation employs borehole inclinometers. In InSAR monitoring, 73 ascending and descending SAR images acquired from satellites were obtained from the Space Agency Science Centre to overcome the limitations of single-orbit SAR data and SAR imaging geometry. Subsequently, precise orbital data provided by the European Space Agency was used to correct orbital errors in the satellite data. A 30-meter resolution spaceborne radar altimeter digital elevation model was used to eliminate topographic phase and perform geocoding, thereby obtaining a deformation rate distribution map of the landslide surface deformation and other data. Figure 3 The deformation time series curves of the monitoring points are shown. Figure 3 The horizontal axis represents time, and the vertical axis represents deformation values. In the internal monitoring, 13 boreholes ranging in depth from 51.64 meters to 80.49 meters were drilled. A borehole inclinometer was used to monitor displacement data at different depths within the landslide body. The monitoring data spanned from September 2019 to May 2020. The surface deformation monitoring data and the internal deformation monitoring data were spatiotemporally aligned and fused to obtain multi-source monitoring data for subsequent parameter inversion.

[0066] The second step involves rapid parameter inversion analysis based on multi-source data. This process includes two main parts: constructing a surrogate model to replace the finite element simulation and searching for the optimal combination of mechanical parameters using optimization algorithms. First, a constitutive model of the rock mass is selected based on the geological information of the Cheyiping landslide obtained from geological exploration, and a three-dimensional finite element numerical simulation model is established. Then, several sets of parameter combination samples are generated using a combination of orthogonal experimental design and random sampling. Specifically, based on geological exploration data, experimental results, and engineering analogy data, the initial mechanical parameters of each soil and rock layer of the landslide are determined, and the elastic modulus, cohesion, internal friction angle, and creep model parameters are selected as the parameters to be inverted. Based on experimental values ​​or engineering analogy values, and combined with the long-term deformation characteristics of the landslide, the upper and lower limits of each parameter are determined, forming the parameter value range. Next, according to the number of parameters to be inverted and the value range of each parameter, each parameter is divided into several levels, and a corresponding orthogonal array is selected to ensure that the combinations of parameters at different levels are evenly distributed. Each row in the orthogonal array corresponds to a set of parameter combinations, and each column corresponds to a parameter to be inverted, thus obtaining a basic sample covering the main parameter space. Subsequently, using the basic samples obtained from the orthogonal experimental design as a framework, random sampling and perturbation are performed within the value ranges or horizontal intervals corresponding to each parameter to generate continuously distributed parameter values. This ensures that the samples possess both global uniformity and local randomness and diversity. In this embodiment, 500 sets of parameter combination samples are preferably generated. These parameter combination samples are then sequentially input into the three-dimensional numerical simulation model of the landslide body for long-term deformation forward simulation. This three-dimensional numerical simulation model is established based on the landslide's topography, stratigraphic structure, distribution of slip zones or weak interlayers, and geological exploration data. It also completes mesh generation, material partitioning, boundary conditions, and calculation time history settings to reflect the deformation process of the landslide body under the combined effects of its own weight, water level fluctuations, and creep. For each set of parameter combination samples, according to the soil layer or material partition number, the elastic modulus, cohesion, internal friction angle, and creep model parameters are written into the three-dimensional numerical simulation model. Calculations are performed while maintaining consistency in model geometry, mesh, boundary conditions, and load conditions to obtain the corresponding landslide displacement field. Subsequently, based on the spatial locations of the external InSAR monitoring points and the internal borehole inclination monitoring points, the surface and deep displacements of the corresponding nodes or interpolation locations are extracted. Each set of parameter combinations is used as input, and the corresponding extracted surface and deep displacements are used as outputs to establish a one-to-one data record. Samples with non-convergent calculations or abnormal displacements are removed, and the retained samples are numbered, normalized, and formatted to generate a "mechanical parameter combination—landslide displacement" response dataset for training the surrogate model.

[0067] The third step is to construct a surrogate model. This embodiment uses a multilayer feedforward neural network based on the error backpropagation algorithm as the surrogate model to construct a nonlinear mapping relationship between "mechanical parameter combinations and landslide displacement." This multilayer feedforward neural network includes an input layer, a hidden layer, and an output layer. The input layer nodes correspond to the mechanical parameters to be inverted, used to input the elastic modulus, cohesion, internal friction angle, and creep model parameters of different soil and rock layers. The output layer nodes correspond to the displacements of selected external and internal monitoring points, used to output the displacement response of the landslide body under the corresponding parameter combinations. The hidden layer is used to characterize the complex nonlinear relationship between mechanical parameters and landslide displacement. During training, each set of mechanical parameter combinations in the response dataset is used as the neural network input, and the landslide surface displacement and deep displacement calculated by the corresponding three-dimensional numerical simulation model are used as the expected output of the neural network. Before training, the input parameters and output displacements are normalized to reduce the impact of differences in dimensions and orders of magnitude between different physical quantities on the training results. The parameter combination samples are passed from the input layer to the hidden and output layers to obtain the displacement results predicted by the surrogate model, which are then compared with the three-dimensional numerical simulation displacement results in the response dataset to form the prediction error. Subsequently, an error backpropagation algorithm is used to propagate the prediction error from the output layer back to the hidden and input layers. The network weights and biases are continuously adjusted based on the error gradient, allowing the surrogate model's output displacement to gradually approximate the 3D numerical simulation result. Once the training error meets the preset accuracy requirement or reaches the maximum number of iterations, the trained neural network surrogate model is obtained. This surrogate model can quickly predict the corresponding landslide displacement response based on the input combination of geotechnical mechanics parameters without repeatedly calling the 3D numerical simulation model.

[0068] The fourth step employs a particle swarm optimization (PSO) algorithm based on population clustering and gradient adaptation to obtain the optimal combination of mechanical parameters. Internal and external multi-source monitoring data are input into the algorithm as the basis for constructing the objective function. The specific implementation process includes: initializing the particle swarm by using the combined parameters of the soil and rock mass to be inverted as particle positions. Each particle corresponds to a set of possible elastic modulus, cohesion, internal friction angle, and creep model parameters, and the initial particle swarm is randomly generated within the range of each parameter value. The particle fitness is calculated. For each particle, the trained nonlinear mapping relationship between "parameter combination" and "landslide displacement" is invoked to quickly predict the displacement response of the external and internal monitoring points corresponding to that parameter combination. The predicted displacement is then compared with the measured displacement. The objective function is constructed using the L2 or weighted L2 of the error between the two. The smaller the objective function value, the closer the parameter combination is to the actual deformation state of the landslide. In this embodiment, the measured displacement data includes displacement data from 7 external InSAR monitoring points and 5 borehole inclination monitoring points at different depths. The particle positions are iteratively updated, and the population is clustered based on particle fitness to maintain diversity. Simultaneously, the inertia weights and acceleration coefficients of the particles are adaptively adjusted based on error reduction, achieving a dynamic balance between global and local searches. Each particle updates its velocity and position based on its individual optimal position, subgroup optimal position, and global optimal position, and fitness is repeatedly calculated. When the objective function meets the convergence condition or reaches the maximum number of iterations, the particle position with the minimum objective function value is determined as the optimal combination of mechanical parameters. The parameters determined through inversion include the elastic modulus, cohesion, internal friction angle, and creep model parameters for different soil and rock layers. The creep model parameters include creep model parameter A and creep model parameter n.

[0069] Taking this landslide as an example, the inverted values ​​of various rock mass parameters obtained from the inversion process are in good agreement with the experimental values ​​of rock mass parameters determined based on rock tests and engineering analogies. For example, for strata Qdel1 and Qal, the experimental value of elastic modulus is 4.92 GPa, and the inverted value is 4.27 GPa; the experimental value of cohesion is 20.00 kPa, and the inverted value is 18.53 kPa; the experimental value of internal friction angle is 25.20°, and the inverted value is 21.12°; the experimental value of creep model parameter A is 2.19 × 10¹ 7 The inversion value is 2.47 × 10¹ 7 The experimental value of the creep model parameter n is 8.57, and the inversion value is 8.95. For formation Qdel2, the experimental value of the elastic modulus is 9.84 GPa, and the inversion value is 8.49 GPa; the experimental value of the cohesion is 28.00 kPa, and the inversion value is 23.28 kPa; the experimental value of the internal friction angle is 24.20°, and the inversion value is 22.63°; the experimental value of the creep model parameter A is 9.80 × 10¹ 7 The inversion value is 9.92 × 10¹ 7The experimental value of the creep model parameter n is 8.02, and the inversion value is 7.48. For the formation STR, the experimental value of the elastic modulus is 11.07 GPa, and the inversion value is 12.81 GPa; the experimental value of the cohesion is 150.00 kPa, and the inversion value is 132.56 kPa; the experimental value of the internal friction angle is 33.00°, and the inversion value is 28.43°; the experimental value of the creep model parameter A is 1.62 × 10¹ 7 The inversion value is 1.29 × 10¹ 7 The experimental value of the creep model parameter n is 8.88, and the inversion value is 6.84. For the WEAK formation, the experimental value of the elastic modulus is 4.43 GPa, and the inversion value is 4.19 GPa; the experimental value of the cohesion is 65.00 kPa, and the inversion value is 59.14 kPa; the experimental value of the internal friction angle is 40.00°, and the inversion value is 37.71°; the experimental value of the creep model parameter A is 1.27 × 10¹ 7 The inversion value is 1.14 × 10¹ 7 The experimental value of the creep model parameter n is 8.71, and the inversion value is 7.26. The above inversion parameters can be used as key physical parameters to characterize the spatiotemporal evolution of the landslide body, and can be further input into a three-dimensional numerical simulation model to obtain the surface displacement, deep displacement, and long-term deformation evolution process of the landslide body.

[0070] To verify the accuracy of parameter inversion based on internal and external monitoring data, this embodiment compares the simulation results of parameter inversion using only external InSAR data, only internal borehole data, and multi-source monitoring data from both internal and external sources. Comparison of Figures 4(a), 4(b), and 4(c) shows that the inversion calculation results using internal and external observation point data better match the actual landslide deformation characteristics. Furthermore, the forward simulation results of parameter inversion are compared with the results from global navigation satellite system measurement points and deep borehole monitoring. The results show that the simulation results calculated from multi-source monitoring data inversion agree well with the measured data, further verifying the accuracy of the inversion results based on multi-source monitoring data.

[0071] In summary, this embodiment acquires multi-source monitoring data by constructing a "sky-ground" multi-dimensional monitoring system, generates parameter combination samples by combining orthogonal experimental design and random sampling, establishes a nonlinear mapping relationship between mechanical parameters and landslide displacement using a neural network surrogate model, and performs parameter inversion using a particle swarm optimization algorithm based on population clustering and gradient adaptation. This approach can accurately obtain the spatiotemporal evolution characteristics of the landslide body, providing reliable prior information for subsequent dynamic feedback and optimization of landslide engineering treatment schemes.

[0072] To implement the above embodiments, this application also proposes an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the analysis method for the spatiotemporal evolution characteristics of landslide bodies based on parameter inversion as described in any of the first aspect embodiments above.

[0073] To implement the above embodiments, this application also proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the analysis method for the spatiotemporal evolution characteristics of landslide bodies based on parameter inversion as described in any one of the first aspect embodiments above.

[0074] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0075] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0076] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0077] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0078] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0079] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0080] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0081] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for analyzing the spatiotemporal evolution characteristics of landslides based on parameter inversion, characterized in that, Includes the following steps: Surface deformation monitoring data and internal deformation monitoring data of the landslide body are collected, and the surface deformation monitoring data and internal deformation monitoring data are fused to obtain multi-source monitoring data; A three-dimensional numerical simulation model is established based on the geological information of the landslide body, and multiple sets of mechanical parameter combination samples are generated. Each set of mechanical parameter combination samples is input into the three-dimensional numerical simulation model for deformation simulation to obtain the response dataset between the mechanical parameter combination and the landslide displacement. A proxy model is constructed, and the proxy model is trained using the response dataset to establish a nonlinear mapping relationship between the combination of mechanical parameters and the landslide displacement; Using the multi-source monitoring data as input to the objective function, the optimization algorithm is combined with the nonlinear mapping relationship to perform parameter inversion and search for the optimal combination of mechanical parameters to characterize the spatiotemporal evolution characteristics of the landslide body.

2. The method according to claim 1, characterized in that, The process of collecting surface deformation monitoring data and internal deformation monitoring data of the landslide body, and fusing the surface deformation monitoring data and the internal deformation monitoring data, includes: The landslide body's ascending and descending orbit SAR images were obtained using time-series InSAR technology. The orbital errors of the SAR images were corrected using precise orbital data. The terrain phase was eliminated and geocoding was performed using a digital elevation model to obtain the surface deformation monitoring data. Displacement data at different depths inside the landslide body were obtained using a borehole inclinometer to obtain internal deformation monitoring data. The surface deformation monitoring data and the internal deformation monitoring data are spatiotemporally aligned and fused to obtain the multi-source monitoring data.

3. The method according to claim 1, characterized in that, The process of establishing a three-dimensional numerical simulation model based on the geological information of the landslide body, generating multiple sets of mechanical parameter combination samples, and inputting each set of mechanical parameter combination samples into the three-dimensional numerical simulation model for deformation simulation includes: Based on the geological information of the landslide, a rock mass constitutive model was selected and a three-dimensional finite element numerical simulation model was established. Multiple sets of mechanical parameter combinations were generated by combining orthogonal experimental design and random sampling. The mechanical parameter combination samples of each group are sequentially input into the three-dimensional finite element numerical simulation model for long-term deformation forward simulation. The surface displacement and deep displacement of the corresponding appearance monitoring point and internal monitoring point are extracted to generate the response dataset.

4. The method according to claim 3, characterized in that, The method of generating multiple sets of mechanical parameter combinations using a combination of orthogonal experimental design and random sampling includes: Determine the parameters to be inverted and the value range of each parameter, wherein the parameters to be inverted include elastic modulus, cohesion, internal friction angle and creep model parameters; Based on the number of parameters to be inverted and the range of their values, each parameter to be inverted is divided into several level values ​​and a corresponding orthogonal array is selected to construct a basic sample covering the main parameter space. Using the basic sample as a framework, random sampling and perturbation are performed within the value range corresponding to each parameter to be inverted to generate continuously distributed parameter values, thus obtaining a preset number of parameter combination samples.

5. The method according to claim 1, characterized in that, The surrogate model is a multilayer feedforward neural network based on the error backpropagation algorithm. The multilayer feedforward neural network includes an input layer, a hidden layer, and an output layer. The nodes of the input layer correspond to the mechanical parameters to be inverted, and the nodes of the output layer correspond to the displacements of the selected external and internal monitoring points.

6. The method according to claim 5, characterized in that, The step of training the proxy model using the response dataset includes: Each set of mechanical parameters in the response dataset is used as the input of the multilayer feedforward neural network. The landslide surface displacement and deep displacement calculated by the corresponding three-dimensional numerical simulation model are used as the expected output, and the input parameters and output displacement are normalized. The prediction error is propagated backward from the output layer to the hidden layer and the input layer through the backpropagation algorithm. The network weights and biases are adjusted according to the error gradient. When the training error meets the preset accuracy requirement or reaches the maximum number of iterations, the trained neural network surrogate model is obtained.

7. The method according to claim 1, characterized in that, The step of using an optimization algorithm combined with the nonlinear mapping relationship to perform parameter inversion and search for the optimal combination of mechanical parameters includes: The combination of parameters of the soil and rock mass to be inverted is used as the particle position, and an initial particle swarm is randomly generated within the range of each parameter value. For each particle, the nonlinear mapping relationship is called to predict the displacement response of the external and internal monitoring points corresponding to the parameter combination, and the predicted displacement is compared with the measured displacement in the multi-source monitoring data. The objective function is constructed using the L2 or weighted L2 of the error between the two. The population is clustered based on particle fitness, and the inertial weights and acceleration coefficients of the particles are adaptively adjusted according to the error reduction, and the particle positions are updated iteratively. When the objective function satisfies the convergence condition or reaches the maximum number of iterations, the position of the particle with the minimum objective function value is determined as the optimal combination of mechanical parameters.

8. The method according to claim 1, characterized in that, The measured displacement data in the multi-source monitoring data includes displacement data from multiple external InSAR monitoring points and multiple borehole inclination monitoring points at different depths. The optimal combination of mechanical parameters includes the elastic modulus, cohesion, internal friction angle and creep model parameters of different soil and rock layers. The creep model parameters include the first creep model parameter A and the second creep model parameter n.

9. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the analysis method for the spatiotemporal evolution characteristics of landslide bodies based on parameter inversion as described in any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the analysis method for the spatiotemporal evolution characteristics of landslide bodies based on parameter inversion as described in any one of claims 1-8.

Images