Geological disaster deduction method and system based on digital twinborn simulation platform

Abstract

The invention provides a geological disaster deduction method and system based on a digital twinborn simulation platform, relates to the technical field of geological disaster deduction, and improves the accuracy and robustness of data fusion by adopting a dynamic weight fusion algorithm and combining information entropy and reliability score; constructing a digital twin model based on a multi-source data inversion and three-dimensional geological modeling technology, updating geological parameters in real time, and reflecting the state change of a geological body; a multi-level stepped early warning model is introduced, and the timeliness and foresight of early warning are enhanced by fusing a physical model and deep learning prediction; and in combination with the graded disaster deduction model, the precision is dynamically selected, and the risk thermodynamic diagram is generated, so that the spatial fineness and decision support capability of geological disaster risk identification are improved. Multi-dimensional organic collaboration is realized, the overall accuracy and response speed of the system are enhanced, and efficient and accurate disaster risk management requirements in a complex geological environment are met.

Description

Technical Field

[0001] This invention relates to the field of geological disaster simulation technology, specifically to a geological disaster simulation method and system based on a digital twin simulation platform. Background Technology

[0002] With the increasing frequency of geological disasters, real-time monitoring and accurate early warning have become crucial means to ensure regional safety. Existing geological disaster monitoring systems face challenges such as the difficulty of integrating multi-source heterogeneous data, low data utilization, and complex time-series dynamic changes. This results in monitoring information failing to fully realize its early warning function, particularly in mountainous areas prone to landslides and debris flows, where efficient and accurate risk identification and dynamic simulation technologies are lacking. Furthermore, traditional monitoring methods often rely on single data sources or static models, making it difficult to dynamically reflect the geological environment and conduct real-time risk assessments, thus limiting the response speed and accuracy of early warning systems.

[0003] In the prior art, CN115221704A discloses a geological hazard simulation method and system based on a digital twin simulation platform, relating to the field of geological hazards. The method includes: connecting a data acquisition device to obtain geological component information and ground building information, generating geological modeling data and building modeling data; inputting the geological modeling data and building modeling data into a digital twin simulation platform for modeling, obtaining regional modeling results; acquiring a video data set based on the video acquisition device; matching the regional modeling results with the video data set to optimize the modeling, obtaining optimized modeling results; using the optimized modeling results as a simulation model, and using a preset hazard simulation database as input variables to obtain a simulation result database for hazard identification. While it can use a digital twin simulation platform for geological modeling, it focuses on data acquisition, modeling, and video-assisted optimization based on the digital twin platform. The process is relatively traditional, lacking in-depth analysis and intelligent links such as multi-source data dynamic fusion, intelligent early warning models, and graded hazard simulation, resulting in insufficient risk identification and real-time response capabilities, making it difficult to meet the high-precision simulation requirements of complex and variable geological environments.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a geological disaster simulation method and system based on a digital twin simulation platform to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The geological disaster simulation method based on a digital twin simulation platform includes the following specific steps:

[0008] S1: Real-time acquisition of actual values ​​of multi-source monitoring data on geological disasters within the monitoring period, acquisition of predicted values ​​of multi-source monitoring data based on long short-term memory model, determination of fusion weight based on actual values ​​and predicted values, and weighting of actual values ​​based on fusion weight;

[0009] S2: Construct a geological twin model and use weighted multi-source monitoring data to update the geological twin model in real time;

[0010] S3: Construct a tiered early warning model that includes a physical base layer, an algorithm correction layer, and a decision fusion layer, and use the tiered early warning model to process the data of the geological twin model to generate geological disaster early warning levels for landslides;

[0011] S4: Construct a hierarchical disaster simulation model, and generate a dynamic risk heat map based on the geological disaster early warning level and the disaster simulation model, and output it in a visual format.

[0012] Preferably, the multi-source monitoring data includes, but is not limited to: satellite remote sensing displacement, GNSS displacement, rainfall, and rock mass deformation, and all multi-source monitoring data are processed using the minimax normalization method;

[0013] The dynamic weight fusion algorithm is constructed by combining entropy weighting and reliability correction factors, and its calculation formula is as follows:

[0014]

[0015] In the formula W i (t) represents the dynamic weight of the i-th type of multi-source monitoring data at time t, where t is the time variable within the monitoring period, E i (t) represents the entropy weight of the i-th type of multi-source monitoring data at time t, R i (t) represents the reliability score of the i-th type of multi-source monitoring data at time t, where a and β are both adjustment coefficients greater than 0, a∈[0.6,0.8], β∈[0.2,0.4], and a+β=1. The subscript i represents the index of the multi-source monitoring data, and n represents the number of types of multi-source monitoring data.

[0016] Preferably, the entropy weight is calculated as follows:

[0017]

[0018] In the formula H i (t) represents the information entropy of the i-th type of multi-source monitoring data at time t;

[0019] The reliability score is calculated as follows:

[0020]

[0021] In the formula Er i (t) represents the average prediction error of the i-th type of multi-source monitoring data within the time window [t-Δt,t], where Δt represents the window length, Er max denoted as the maximum permissible error, γ represents the attenuation coefficient, and γ∈[0.5,2].

[0022] Preferably, the construction logic of the geological twin model is as follows:

[0023] Rock mass integrity was inverted based on multi-source monitoring data, and a three-dimensional geological structure model was constructed using three-dimensional geological modeling technology;

[0024] Based on a three-dimensional geological structure model and multi-source monitoring data, a mapping relationship between multi-source monitoring data and geological parameters is established;

[0025] The geological parameter distribution field is generated using spatial interpolation methods, and the geological parameters are updated in real time.

[0026] Preferably, when using weighted multi-source monitoring data to update the geological twin model in real time, an integrated Kalman filter algorithm is used, and the expression for the update equation is:

[0027] X(t)=X(t-Δt)+K(t)·[Z(t)-H(X(t-Δt))]

[0028] In the formula, X(t) represents the geological parameter vector at time t, and Z(t) represents the weighted multi-source monitoring data vector at time t, Z(t)={x1(t)·W1(t),x2(t)·W2(t),…,x i (t)·W i (t),…,x n (t)·W n (t)},x i K(t) represents the measured value of the i-th type of multi-source monitoring data at time t, K(t) represents the Kalman gain matrix at time t, and H(·) represents the observation operator.

[0029] Preferably, the logic for using a tiered early warning model to process data from a geological twin model and generate geological disaster early warning levels is as follows:

[0030] The Bishop method and the central difference method were used respectively to process the geological parameters in the geological twin model to generate the slope stability coefficient and slope displacement rate.

[0031] The physical foundation layer determines the slope stability coefficient and slope displacement rate, and outputs the initial warning level for geological hazards based on different determination results. The determination expression is as follows:

[0032]

[0033] In the formula L b The initial warning level is indicated by FOS, the slope stability coefficient is indicated by V, and the slope displacement rate is indicated by V.

[0034] The algorithm correction layer determines the initial warning level. When L b When the value is ≥2, the LSTM network is activated, and the geological parameters of the geological twin model are used as input to output the predicted slope stability coefficient and calculate the change in stability coefficient.

[0035] The decision fusion layer uses a fuzzy rule engine to process the initial warning level and the stability coefficient correction, outputting the final warning level. The expression for the fuzzy rule is:

[0036]

[0037] In the formula L f The final warning level is indicated by ΔFOS, which represents the change in the stability coefficient. The final warning level is the geological disaster warning level for landslides.

[0038] Preferably, the LSTM network is trained using the dynamically adaptive cuckoo algorithm, and the expression for the step size adjustment strategy is:

[0039]

[0040] In the formula, D(k) represents the step size of the k-th iteration, D max Denotes the initial maximum step size, and D max ∈[0.3,0.6], sech(·) is a hyperbolic secant function, d represents the decay rate factor, k represents the index of the iteration number, and K represents the maximum number of iterations.

[0041] Preferably, the graded disaster simulation model includes a particle kinematics model and a discrete element refined model;

[0042] When the final warning level meets L f When the value is ≤3, a point mass kinematics model is used for disaster simulation;

[0043] When the final warning level meets L f When the value is >3, a refined discrete element model is used for disaster simulation.

[0044] Preferably, the risk value in the dynamic risk heatmap is calculated as follows:

[0045]

[0046] In the formula, R represents the risk value, and V max The maximum displacement rate is represented by λ1, λ2, and λ3, which are all weighting coefficients greater than 0, and λ1+λ2+λ3=1;

[0047] Risk values ​​are coupled into a three-dimensional geological structure model and identified by color.

[0048] A geological disaster simulation system based on a digital twin simulation platform, wherein the simulation system is used to execute the above-mentioned simulation method, specifically including:

[0049] The multi-source data acquisition module is used to collect multi-source heterogeneous geological disaster monitoring data in real time, perform time-series prediction on the collected data, and use a dynamic weight fusion algorithm to weight the predicted data.

[0050] The model building module is used to construct a three-dimensional digital geological twin model based on weighted multi-source monitoring data, and to dynamically update the model in real time by integrating the Kalman filter algorithm.

[0051] The tiered early warning module is used to generate geological disaster early warning levels based on data from the geological twin model, employing a tiered early warning model that combines physical calculations and deep learning algorithms.

[0052] The risk visualization module is used to select different levels of disaster simulation models based on the warning level to perform risk calculations, generate dynamic risk heat maps, and realize three-dimensional visualization of risks.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] This invention employs a dynamic weighted fusion algorithm, combining information entropy and reliability scoring, to achieve scientific weighted processing of multi-source monitoring data, improving the accuracy and robustness of data fusion. Based on multi-source data inversion and 3D geological modeling technology, a digital twin model is constructed, dynamically updating geological parameters in real time to accurately reflect changes in the geological body's state. A multi-level, tiered early warning model is introduced, integrating physical models and deep learning predictions to enhance the timeliness and foresight of early warnings. A graded disaster projection model is combined with dynamic selection of projection accuracy and generation of risk heat maps, improving the spatial precision and decision support capabilities of geological disaster risk identification. This invention achieves organic synergy between multi-source data fusion, digital twin modeling, intelligent early warning, and dynamic projection, enhancing the accuracy and response speed of the geological disaster monitoring and early warning system, and meeting the needs for efficient and accurate disaster risk management in complex geological environments. Attached Figure Description

[0055] Figure 1This is a schematic diagram of the overall method flow of the present invention;

[0056] Figure 2 This is a schematic diagram illustrating the changes in prediction error in this invention;

[0057] Figure 3 This is a schematic diagram illustrating the change in information entropy in this invention;

[0058] Figure 4 This is a schematic diagram illustrating the changes in reliability scores in this invention;

[0059] Figure 5 This is a schematic diagram illustrating the change of dynamic weights in this invention;

[0060] Figure 6 This is a schematic diagram of the module structure of the present invention. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0062] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0063] Example:

[0064] Please see Figures 1 to 5 The present invention provides a technical solution:

[0065] The geological disaster simulation method based on a digital twin simulation platform includes the following specific steps:

[0066] S1: Real-time collection of multi-source monitoring data on geological disasters, time-series prediction of the collection time based on historical values ​​of multi-source monitoring data, and weighted processing of the predicted multi-source monitoring data using a dynamic weighted fusion algorithm.

[0067] Here, using historical values ​​from multi-source monitoring data for time-series prediction is not simply to obtain the predicted data at the current acquisition time, but rather to compare the predicted data at the current time with the data collected by the monitoring equipment to quantify "the predictive capability of a certain type of monitoring equipment in the recent time period." This addresses the problem of "dynamic assessment of monitoring equipment reliability," enabling the system to proactively adjust its data fusion strategy before equipment completely fails, thereby improving the reliability of disaster early warning. For example, when GNSS (Global Satellite System) equipment exhibits continuous prediction deviations, its reliability is considered reduced, thus decreasing its weight in twin fusion.

[0068] Multi-source monitoring data include, but are not limited to: satellite remote sensing displacement, GNSS displacement, rainfall, and rock mass deformation. All multi-source monitoring data are processed using the minimax normalization method.

[0069] The dynamic weight fusion algorithm is constructed by combining the entropy weight method and the reliability correction factor. Its calculation formula is as follows:

[0070]

[0071] In the formula W i (t) represents the dynamic weight of the i-th type of multi-source monitoring data at time t during data acquisition, E i (t) represents the entropy weight of the i-th type of multi-source monitoring data at time t, R i (t) represents the reliability score of the i-th type of multi-source monitoring data at time t, where a and β are both adjustment coefficients greater than 0, a∈[0.6,0.8], β∈[0.2,0.4], and a+β=1. The subscript i represents the index of the multi-source monitoring data, and n represents the number of types of multi-source monitoring data.

[0072] Entropy weighting dynamically calculates weights based on the information entropy of the data itself, scientifically measuring the effectiveness of information provided by each data source. The higher the information entropy, the greater the variability and the richer the information from that data source, and the higher the weight. The reliability correction factor, on the other hand, combines a reliability score based on prediction error to dynamically reflect the accuracy and stability of each data source, avoiding assigning excessive weights to data with anomalies or noise.

[0073] The entropy weight is calculated as follows:

[0074]

[0075] In the formula H i (t) represents the information entropy of the i-th type of multi-source monitoring data at time t. The calculation logic is to collect the time series information of multi-source monitoring data, discretize the data series into finite intervals, calculate the probability of occurrence of each interval, and then express the information entropy as follows:

[0076]

[0077] Here p ∈ (t) represents the probability of the ∈-th interval occurring at time t, where ∈ represents the interval index.

[0078] The reliability score is calculated as follows:

[0079]

[0080] In the formula Er i (t) represents the average prediction error of the i-th type of multi-source monitoring data within the time window [t-Δt,t]. It can be obtained by statistically analyzing the difference between the predicted data and the actual collected data using the sliding window method. Δt represents the window length, Er max denoted as the maximum permissible error, γ represents the attenuation coefficient, and γ∈[0.5,2].

[0081] Since the definition and calculation method of the entropy weight method are existing technologies, they will not be elaborated upon here. The calculation formula for the reliability score shows that it uses an exponential function expression to convert the prediction error into a reliability score. That is, the larger the error, the larger the negative value of the exponent, and the closer the reliability score is to 0, indicating a lower reliability of the data source; conversely, the smaller the error, the smaller the negative value of the exponent, and the closer the reliability score is to 1, indicating a higher reliability of the data source. The attenuation coefficient is used to adjust the sensitivity of the error to the reliability score. The larger its value, the more sensitive the reliability score is to changes in error, suitable for scenarios with strict requirements; conversely, it is more lenient. Here, the entropy weight reflects the volatility and richness of the data, giving higher weights to data sources with a large amount of information. The reliability score quantifies the historical accuracy of the data source, avoiding the influence of abnormal data or noise, ensuring the scientific and reasonable allocation of fusion weights. The fusion of the two can achieve a dual assessment of information content and quality, ensuring data quality and credibility. It is particularly suitable for multi-source heterogeneous geological disaster monitoring data with large quality fluctuations, improving the accuracy and robustness of prediction and early warning.

[0082] In this embodiment, eight typical sets of data from multi-source monitoring data (1. satellite remote sensing displacement, 2. GNSS displacement, 3. rainfall, 4. rock mass deformation, 5. surface crack width, 6. groundwater level change, 7. surface temperature change, 8. soil moisture content) are studied to obtain their corresponding prediction errors, information entropy, entropy weights, reliability scores, and dynamic weight data. Noise is randomly added to the original data of two types of multi-source monitoring data, and the corresponding prediction errors, information entropy, entropy weights, reliability scores, and dynamic weight data are calculated again to verify the noise resistance of the dynamic weight data. Specific data are shown in the table below.

[0083] Table 1: Raw Data from Multi-Source Monitoring

[0084] Data Groups Prediction error Information entropy Entropy weight Reliability rating Dynamic weights 1 0.04 0.48 0.21 0.92 0.21 2 0.06 0.44 0.19 0.88 0.2 3 0.1 0.42 0.18 0.82 0.18 4 0.12 0.41 0.18 0.78 0.17 5 0.11 0.39 0.17 0.8 0.17 6 0.08 0.38 0.17 0.85 0.17 7 0.05 0.37 0.16 0.91 0.17 8 0.07 0.34 0.15 0.87 0.15

[0085] Table 2: Noise Data from Multi-Source Monitoring

[0086] Data Groups Prediction error Information entropy Entropy weight Reliability rating Dynamic weights 1 0.05 0.48 0.21 0.9 0.2 2 0.4 0.44 0.19 0.45 0.11 3 0.09 0.41 0.18 0.83 0.18 4 0.13 0.41 0.18 0.76 0.16 5 0.38 0.39 0.17 0.49 0.11 6 0.07 0.38 0.17 0.87 0.17 7 0.06 0.37 0.16 0.9 0.17 8 0.08 0.34 0.15 0.85 0.15

[0087] From the data in the table above and Figures 2-5 It can be seen that after noise processing, the prediction errors of GNSS displacement and surface crack width are significantly increased, which in reality is reflected as equipment or data abnormalities. Correspondingly, their dynamic weights decrease significantly. However, a small amount of disturbance is not enough to significantly affect the probability distribution, that is, it does not substantially change the probability distribution of the data itself. Therefore, the information entropy and entropy weight remain basically unchanged, which verifies that this dynamic weight algorithm has a good ability to suppress noisy data.

[0088] S2: Construct a geological twin model and use weighted multi-source monitoring data to update the geological twin model in real time.

[0089] The construction logic of the geological twin model is as follows:

[0090] Rock mass integrity was inverted based on multi-source monitoring data, and a three-dimensional geological structure model was constructed using three-dimensional geological modeling technology;

[0091] Based on a three-dimensional geological structure model and multi-source monitoring data, a mapping relationship between multi-source monitoring data and geological parameters is established;

[0092] The geological parameter distribution field is generated using spatial interpolation methods, and the geological parameters are updated in real time.

[0093] Specifically, monitoring data (such as GNSS displacement, remote sensing images, rainfall, and rock mass deformation) can be used to invert the degree of damage and integrity of the rock mass structure. Then, physical inversion algorithms or machine learning methods are used to map the monitoring data to rock mass integrity parameters (such as crack distribution and strength indicators). Next, based on geological exploration data, remote sensing data, and inversion results, a geological model including spatial structures such as strata, faults, and lithological distribution is built using 3D modeling software (such as GOCAD and Petrel). Then, statistical regression, neural networks, and other methods are used to establish the mapping relationship between data and parameters, corresponding changes in monitoring data to changes in geological parameters (such as mapping displacement to stress changes). Finally, based on a limited number of monitoring points, Kriging interpolation, inverse distance weighting (IDW), spline interpolation, and other methods are used to generate a continuous spatial distribution of geological parameters to form a three-dimensional continuous parameter field, which facilitates physical calculations and risk projection of the digital twin.

[0094] The geological twin model constructed by this method is based on the inversion of rock mass integrity from multi-source monitoring data and combined with three-dimensional geological modeling technology. It can construct a high-precision digital twin that reflects the actual underground rock mass structure and state, and has real spatial physical properties, thereby better dynamically expressing the state of the geological body.

[0095] When using weighted multi-source monitoring data to update the geological twin model in real time, an integrated Kalman filter algorithm is employed, and the expression for the update equation is:

[0096] X(t)=X(t-Δt)+K(t)·[Z(t)-H(X(t-Δt))]

[0097] In the formula, X(t) represents the geological parameter vector at time t, and Z(t) represents the weighted multi-source monitoring data vector at time t, Z(t)={x1(t)·W1(t),x2(t)·W2(t),…,x i (t)·W i (t),…,x n (t)·W n (t)},x i (t) represents the measured value of the i-th type of multi-source monitoring data at time t, K(t) represents the Kalman gain matrix at time t, and H(·) represents the observation operator. The observation operator refers to the measurement function in Kalman filtering, which maps the geological parameter vector in the state space to the observation space, that is, the correspondence between simulated observation values ​​and real observation data. It can be implemented using physical models, empirical formulas, neural networks, etc., and the specific form can be determined according to engineering needs or expert experience.

[0098] S3: Construct a tiered early warning model that includes a physical base layer, an algorithm correction layer, and a decision fusion layer. Use the tiered early warning model to process the data from the geological twin model and generate geological disaster early warning levels for landslides.

[0099] The logic for using a tiered early warning model to process data from a geological twin model and generate geological disaster early warning levels is as follows:

[0100] The Bishop method and the central difference method were used to process the geological parameters in the geological twin model to generate the slope stability coefficient and slope displacement rate. The larger the slope stability coefficient and the smaller the slope displacement rate, the more stable the geology.

[0101] The Bishop method for calculating slope stability coefficient and the central difference method for calculating slope displacement rate are both mature existing technologies, so their formulas, models and specific calculation methods will not be elaborated here.

[0102] The physical foundation layer determines the slope stability coefficient and slope displacement rate, and outputs the initial warning level for geological hazards based on different determination results. The determination expression is as follows:

[0103]

[0104] In the formula L b The initial warning level is indicated by FOS, the slope stability coefficient is indicated by V, and the slope displacement rate is indicated by V.

[0105] The algorithm correction layer determines the initial warning level. When L b When the value is ≥2, the LSTM network is activated, and the geological parameters of the geological twin model are used as input to output the predicted slope stability coefficient and calculate the change in stability coefficient.

[0106] The decision fusion layer uses a fuzzy rule engine to process the initial warning level and the stability coefficient correction, outputting the final warning level. The expression for the fuzzy rule is:

[0107]

[0108] In the formula L f The final warning level is indicated by ΔFOS, which represents the change in the stability coefficient, i.e., the difference between the predicted future stability coefficient and the current value, reflecting the trend change in slope stability.

[0109] As can be seen from the definition of warning levels, the higher the warning level, the more dangerous the situation. The initial warning level reflects three basic stages: In the first stage, the slope stability coefficient is relatively high, the slope displacement rate is relatively low, and the slope is relatively stable. In the second stage, the slope stability coefficient begins to decrease, and the slope displacement rate increases, indicating that the slope is beginning to show slight signs of instability, possibly with initial cracks or local deformation. In the third stage, the slope stability coefficient further decreases, and the slope displacement rate accelerates, indicating that the slope is significantly unstable, the risk of landslides is increased, the geological deformation is obvious, and landslide induction conditions exist, placing the slope in a critical state. Furthermore, if the prediction shows a significant decrease in the stability coefficient while the slope is in the critical state of the third stage, the slope stability is considered to be rapidly deteriorating. Even if the highest risk state has not yet been reached, a warning level of 4 can be issued in advance based on future trends to avoid delayed response. The time-series prediction here is to predict future trends. By combining the current state with future trends, the limitations of static judgments at a single moment are avoided, false alarms and false negatives are reduced, and the reliability of the system is improved.

[0110] The LSTM network is trained using the dynamically adaptive cuckoo algorithm, and the expression for the step size adjustment strategy is:

[0111]

[0112] In the formula, D(k) represents the step size of the k-th iteration, D max Denotes the initial maximum step size, and D max ∈[0.3,0.6], sech(·) is a hyperbolic secant function, d represents the decay rate factor, k represents the index of the iteration number, and K represents the maximum number of iterations.

[0113] S4: Construct a hierarchical disaster simulation model, and generate a dynamic risk heat map based on the geological disaster early warning level and the disaster simulation model, and output it in a visual format.

[0114] The graded disaster simulation model includes a particle kinematics model and a discrete element refined model;

[0115] When the final warning level meets L f When the value is ≤3, a point mass kinematics model is used for disaster simulation;

[0116] When the final warning level meets L f When the value is >3, a refined discrete element model is used for disaster simulation.

[0117] In most cases where the slope is at a low risk level, the particle kinematics model, which is simple to calculate and fast, is suitable for rapid and large-scale preliminary simulations, saving computational resources. In cases with a high risk level, the discrete element model is used to simulate the interaction between particles, which can more realistically reflect the rock mass failure, crack propagation and landslide process, thereby improving the realism and accuracy of disaster simulation.

[0118] The risk value in the dynamic risk heatmap is calculated as follows:

[0119]

[0120] In the formula, R represents the risk value, and V max λ1 represents the maximum displacement rate, λ2, and λ3 are all weighting coefficients greater than 0, and λ1+λ2+λ3=1.

[0121] In the formula, the slope stability coefficient reflects the mechanical stability state of the slope. The larger the value, the more stable the slope, and vice versa. The change in the stability coefficient reflects the dynamic trend of slope stability. A negative value indicates deterioration of stability. Therefore, the larger the sum of the two, the more stable the slope. Thus, in the risk value calculation, it is set as 1-(FOS+ΔFOS) to make the risk value positively correlated with the risk. The slope displacement rate is the direct dynamic manifestation of landslide occurrence. After normalization, it is easy to unify the dimensions with other indicators. The larger the slope displacement rate, the higher the risk. Finally, the comprehensive warning level directly reflects the degree of risk and is also proportional to the risk. Therefore, it is included in the risk value calculation to strengthen the early warning model.

[0122] Risk values ​​are coupled into a three-dimensional geological structure model and identified by color.

[0123] The risk values ​​in the dynamic risk heat map comprehensively consider slope stability, displacement rate, and final warning level, and use weighting coefficients to flexibly adjust the contribution of different indicators to the risk to adapt to different application needs. Combined with a three-dimensional geological structure model, the spatial distribution of risk is visualized with color, which facilitates intuitive understanding and decision-making.

[0124] Please see Figure 6 As shown, the present invention also provides a geological disaster simulation system based on a digital twin simulation platform. The simulation system is used to execute the above-described simulation method, specifically including:

[0125] The multi-source data acquisition module is used to collect multi-source heterogeneous geological disaster monitoring data in real time, perform time-series prediction on the collected data, and use a dynamic weight fusion algorithm to weight the predicted data.

[0126] The model building module is used to construct a three-dimensional digital geological twin model based on weighted multi-source monitoring data, and to dynamically update the model in real time by integrating the Kalman filter algorithm.

[0127] The tiered early warning module is used to generate geological disaster early warning levels based on data from the geological twin model, employing a tiered early warning model that combines physical calculations and deep learning algorithms.

[0128] The risk visualization module is used to select different levels of disaster simulation models based on the warning level to perform risk calculations, generate dynamic risk heat maps, and realize three-dimensional visualization of risks.

[0129] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0130] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0131] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0132] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A geological disaster simulation method based on a digital twin simulation platform, characterized in that, The specific steps include: S1: Real-time acquisition of actual values ​​of multi-source monitoring data on geological disasters within the monitoring period, acquisition of predicted values ​​of multi-source monitoring data based on long short-term memory model, determination of fusion weight based on actual values ​​and predicted values, and weighting of actual values ​​based on fusion weight; S2: Construct a geological twin model and use weighted multi-source monitoring data to update the geological twin model in real time; S3: Construct a tiered early warning model that includes a physical base layer, an algorithm correction layer, and a decision fusion layer, and use the tiered early warning model to process the data of the geological twin model to generate geological disaster early warning levels for landslides; S4: Construct a hierarchical disaster simulation model, and generate a dynamic risk heat map based on the geological disaster early warning level and the disaster simulation model, and output it in a visual format.

2. The geological disaster simulation method based on a digital twin simulation platform according to claim 1, characterized in that: The multi-source monitoring data includes, but is not limited to: satellite remote sensing displacement, GNSS displacement, rainfall, and rock mass deformation, and all multi-source monitoring data are processed using the minimax normalization method; The dynamic weight fusion algorithm is constructed by combining entropy weighting and reliability correction factors, and its calculation formula is as follows: In the formula W i (t) represents the dynamic weight of the i-th type of multi-source monitoring data at time t, where t is the time variable within the monitoring period, E i (t) represents the entropy weight of the i-th type of multi-source monitoring data at time t, R i (t) represents the reliability score of the i-th type of multi-source monitoring data at time t, where a and β are both adjustment coefficients greater than 0, a∈[0.6,0.8], β∈[0.2,0.4], and a+β=1. The subscript i represents the index of the multi-source monitoring data, and n represents the number of types of multi-source monitoring data.

3. The geological disaster simulation method based on a digital twin simulation platform according to claim 2, characterized in that: The entropy weight is calculated as follows: In the formula H i (t) represents the information entropy of the i-th type of multi-source monitoring data at time t; The reliability score is calculated as follows: In the formula Er i (t) represents the average prediction error of the i-th type of multi-source monitoring data within the time window [t-Δt,t], where Δt represents the window length, Er max denoted as the maximum permissible error, γ represents the attenuation coefficient, and γ∈[0.5,2].

4. The geological disaster simulation method based on a digital twin simulation platform according to claim 3, characterized in that: The construction logic of the geological twin model is as follows: Rock mass integrity was inverted based on multi-source monitoring data, and a three-dimensional geological structure model was constructed using three-dimensional geological modeling technology; Based on a three-dimensional geological structure model and multi-source monitoring data, a mapping relationship between multi-source monitoring data and geological parameters is established; The geological parameter distribution field is generated using spatial interpolation methods, and the geological parameters are updated in real time.

5. The geological disaster simulation method based on a digital twin simulation platform according to claim 4, characterized in that: When using weighted multi-source monitoring data to update the geological twin model in real time, an integrated Kalman filter algorithm is employed, and the expression for the update equation is: X(t)=X(t-Δt)+K(t)·[Z(t)-H(X(t-Δt))] In the formula, X(t) represents the geological parameter vector at time t, and Z(t) represents the weighted multi-source monitoring data vector at time t, Z(t)={x1(t)·W1(t),x2(t)·W2(t),…,x i (t)·W i (t),…,x n (t)·W n (t)},x i K(t) represents the measured value of the i-th type of multi-source monitoring data at time t, K(t) represents the Kalman gain matrix at time t, and H(·) represents the observation operator.

6. The geological disaster simulation method based on a digital twin simulation platform according to claim 5, characterized in that: The logic for using a tiered early warning model to process data from a geological twin model and generate geological disaster early warning levels is as follows: The Bishop method and the central difference method were used respectively to process the geological parameters in the geological twin model to generate the slope stability coefficient and slope displacement rate. The physical foundation layer determines the slope stability coefficient and slope displacement rate, and outputs the initial warning level for geological hazards based on different determination results. The determination expression is as follows: In the formula L b The initial warning level is indicated by FOS, the slope stability coefficient is indicated by V, and the slope displacement rate is indicated by V. The algorithm correction layer determines the initial warning level. When L b When the value is ≥2, the LSTM network is activated, and the geological parameters of the geological twin model are used as input to output the predicted slope stability coefficient and calculate the change in stability coefficient. The decision fusion layer uses a fuzzy rule engine to process the initial warning level and the stability coefficient correction, outputting the final warning level. The expression for the fuzzy rule is: In the formula L f The final warning level is indicated by ΔFOS, which represents the change in the stability coefficient. The final warning level is the geological disaster warning level for landslides.

7. The geological disaster simulation method based on a digital twin simulation platform according to claim 6, characterized in that: The LSTM network is trained using the dynamically adaptive cuckoo algorithm, and the expression for the step size adjustment strategy is as follows: In the formula, D(k) represents the step size of the k-th iteration, D max Denotes the initial maximum step size, and D max ∈[0.3,0.6], sech(·) is a hyperbolic secant function, d represents the decay rate factor, k represents the index of the iteration number, and K represents the maximum number of iterations.

8. The geological disaster simulation method based on a digital twin simulation platform according to claim 6, characterized in that: The hierarchical disaster simulation model includes a particle kinematics model and a discrete element refined model; When the final warning level meets L f When the value is ≤3, a point mass kinematics model is used for disaster simulation; When the final warning level meets L f When the value is >3, a refined discrete element model is used for disaster simulation.

9. The geological disaster simulation method based on a digital twin simulation platform according to claim 6, characterized in that: The risk value in the dynamic risk heatmap is calculated as follows: In the formula, R represents the risk value, and V max The maximum displacement rate is represented by λ1, λ2, and λ3, which are all weighting coefficients greater than 0, and λ1+λ2+λ3=1; Risk values ​​are coupled into a three-dimensional geological structure model and identified by color.

10. A geological disaster simulation system based on a digital twin simulation platform, characterized in that: The deduction system is used to execute the deduction method as described in any one of claims 1-9, specifically including: The multi-source data acquisition module is used to collect multi-source heterogeneous geological disaster monitoring data in real time, perform time-series prediction on the collected data, and use a dynamic weight fusion algorithm to weight the predicted data. The model building module is used to construct a three-dimensional digital geological twin model based on weighted multi-source monitoring data, and to dynamically update the model in real time by integrating the Kalman filter algorithm. The tiered early warning module is used to generate geological disaster early warning levels based on data from the geological twin model, employing a tiered early warning model that combines physical calculations and deep learning algorithms. The risk visualization module is used to select different levels of disaster simulation models based on the warning level to perform risk calculations, generate dynamic risk heat maps, and realize three-dimensional visualization of risks.

Images