Landslide mass dynamic simulation monitoring and early warning method based on multi-source sensing fusion

Abstract

The invention relates to the technical field of geological disaster monitoring, and discloses a landslide dynamic simulation monitoring and early warning method based on multi-source sensing fusion, and the method comprises the steps: collecting multi-source data of a landslide body through a plurality of heterogeneous sensors, and enabling the multi-source data to be in space-time alignment after preprocessing; building a multi-parameter fusion model fusing a displacement field, a mechanical field and an environment field based on the preprocessed data, enabling the multi-parameter fusion model to output a deformation rate and a stability coefficient, and dynamically adjusting the weights of the displacement field, the mechanical field and the environment field according to a landslide evolution stage; predicting a future deformation trend of the landslide mass in combination with a geological structure and historical data, comparing a stability coefficient with a dynamic safety threshold to judge a risk level, and generating early warning information; and dynamically correcting a reference weight coefficient in the model based on the deviation between the monitoring data and the model output, so that the model is adaptively optimized. The problems of single monitoring dimension and static model solidification in the prior art are solved, and accurate and adaptive monitoring and early warning of the risk state of the landslide mass are realized.

Description

TECHNICAL FIELD

[0001] The application relates to a landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion, and belongs to the technical field of geological disaster monitoring and early warning. BACKGROUND

[0002] Landslide is a common and highly destructive geological disaster in China. With the intensification of climate change and the expansion of engineering construction to complex geological areas (such as high-altitude cold regions and mountainous areas along highways / railways), the suddenness, concealment and chain generation of landslide disasters have significantly increased, posing a serious threat to people's life and property safety and major engineering operations.

[0003] The traditional technology in the field of landslide monitoring and early warning currently adopts a single and static implementation method in each link: in the data collection stage, a single type of sensor is often relied on for monitoring, such as only deploying GNSS ground displacement sensors to obtain coordinate changes of the landslide surface or only using rain gauges to record rainfall. Although a few solutions combine two types of sensors, they do not form a multi-dimensional parameter collaborative collection system. In the data processing stage, the processing of collected data mainly focuses on basic denoising (such as simple filtering), and lacks spatiotemporal alignment processing of multi-source heterogeneous data (displacement, stress, and environmental parameters), resulting in different sensor data being out of sync in time and not unified in spatial reference, making it difficult to effectively fuse the data. In the risk assessment and early warning stage, a fixed weight model is often used to calculate the stability coefficient (such as only using displacement rate as the core indicator), a static safety threshold is used to divide the risk level, and once the model parameters (such as weight coefficients and safety margins) are set, they cannot be adjusted, which cannot adapt to the dynamic changes of the landslide evolution stage. SUMMARY

[0004] The application provides a landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion, which can realize deep fusion of multi-source information, dynamic adaptation of model parameters, and accurate reflection of the risk state of the entire life cycle of the landslide. The method comprises the following steps: S1, real-time collection of multi-source data of the landslide by multiple heterogeneous sensors, and preprocessing of the multi-source data to unify them to the same spatiotemporal reference; S2, based on the preprocessed data, a multi-parameter fusion model for dynamic simulation of the landslide is constructed by integrating the displacement field, mechanical field and environmental field of the landslide; The parameter fusion model outputs the deformation rate V(t) and the stability coefficient K(t) of the landslide, and the multi-parameter fusion model dynamically calculates the displacement weight ω d , the mechanical weight ω m , and the environmental weight ωe , wherein the displacement weight ω d Mechanical weights ω m Environmental weight ω e It is calculated based on the corresponding benchmark weight coefficients and evolution stages; S3, based on the deformation rate output in S2 V(t) and stability coefficient K(t) By combining the geological structure parameters and historical deformation data of the landslide, a dynamic simulation algorithm is used to predict the future deformation trend of the landslide. S4, the real-time stability coefficient output from S2 K(t) and the future stability system predicted in S3 The data is compared with a preset stability threshold, and the risk level of the landslide is output based on the comparison results. S5. Generate corresponding early warning information based on the risk level and transmit it to the early warning platform. At the same time, dynamically correct the benchmark weight coefficients in the multi-parameter fusion model based on the deviation between the current monitoring data and the model output.

[0005] Preferably, in S1, the multi-source monitoring data includes displacement parameters of the landslide surface, displacement or strain data of the landslide body, external inducing factors of the landslide body, and data on changes in the internal mechanical state of the landslide body. The external triggering factors include rainfall and groundwater level in the landslide area.

[0006] Preferably, in S1, the preprocessing of multi-source data includes denoising and spatiotemporal alignment. The spatiotemporal alignment process achieves time synchronization of data from different sensors through timestamp calibration, and spatial synchronization by locating the measurement points of different sensors to a unified spatial grid through spatial coordinate mapping, so as to ensure the comparability of data in both time and space dimensions.

[0007] Preferably, in S3, the predicted future deformation trend includes at least one of a short-term trend and a long-term trend.

[0008] Preferably, in S4, the stability threshold includes a first stability threshold. Kc Second stability threshold Kc-ΔK ,in, ΔK This is a dynamic safety margin; The specific process for determining the risk level of a landslide based on comparison results includes: Dangdang K(t) ≥ Kc At that time, the risk level was determined to be stable; when Kc - ΔK ≤ K(t) < KcAt that time, the risk level was determined to be potentially unstable; when K(t) < Kc - ΔK At that time, the risk level was determined to be unstable.

[0009] Preferably, the dynamic safety margin ΔK Calculated using the following formula: ; in, μ Based on the basic safety margin factor, V thre This is the preset displacement rate threshold.

[0010] Preferably, step S4 further includes a verification step; the verification step is used to confirm the predicted deformation trend and / or calculate the stability coefficient. K(t) The simulation results do not exceed the preset allowable error range.

[0011] Preferably, in S5, the specific process of dynamically correcting the weight coefficients in the multi-parameter fusion model includes: Calculate the measured value of the stability coefficient at the current moment. K(t)' With critical stability coefficient Kc deviation value ΔK' ; According to the deviation value ΔK' The baseline weighting coefficients corresponding to displacement, mechanics, and environment are corrected.

[0012] Preferably, the correction rule for the benchmark weighting coefficient is as follows: The baseline weighting coefficient of displacement varies with ΔK' As it increases, the mechanical reference weight coefficient increases accordingly. ΔK' Increase and decrease, the baseline weighting coefficient of the environment and ΔK' It is positively correlated with current environmental factors.

[0013] Preferably, the evolution stages of the landslide are based on the normalized displacement rate. The process is divided into three stages: initial deformation, accelerated deformation, and critical instability. Different functional relationships are used to calculate the displacement weights at each stage. ω d Mechanical weights ω m and environmental weight ω e .

[0014] The beneficial effects that this application can produce include: By collaboratively collecting data from multiple sources of sensors, covering the surface, depth, external environment, and internal mechanics of the landslide body, the collected multi-source data includes key parameters such as displacement, strain, rainfall, groundwater level, and stress, avoiding the loss of crucial information due to single sensors. This provides comprehensive data support for subsequent multi-field fusion analysis, laying the foundation for a full-chain monitoring system from cause to response to characterization, and reducing early warning bias caused by incomplete data. It effectively avoids missed or false alarms due to missing key parameters.

[0015] In the data preprocessing stage, high-frequency noise is removed by denoising to improve the quality of the original data and ensure the reliability of the input model. Multi-source data is synchronized in time and space by timestamp calibration and spatial coordinate mapping to solve the problems of asynchronous time and inconsistent spatial reference of heterogeneous sensor data, thus ensuring data comparability and fusion effectiveness.

[0016] By integrating displacement field, mechanical field, and environmental field to construct a unified model, a multi-parameter fusion model is built, transforming multi-dimensional data into quantified deformation rate. V (t) and stability coefficient K (t) This enables precise quantification of risk status; Furthermore, the multi-parameter fusion model is constructed by dividing the evolution stages (e.g., initial, acceleration, and critical instability) according to the normalized displacement rate. A dynamic weight calculation mechanism based on the evolution stages is introduced, enabling the model to adapt to different stages of landslide evolution, which significantly improves the response accuracy and early warning timeliness of risk-sensitive features.

[0017] This application combines geological structural parameters and historical data to predict the short-term and long-term deformation trends of landslides using algorithms such as finite element method and time series analysis. It can predict the evolution direction of landslides in advance, and the simulation results error is controlled within 5%. This provides a forward-looking basis for risk level determination, extends the early warning preparation time, and improves the timeliness of early warning.

[0018] Introducing dynamic safety margin ΔK Constructing a two-level stability threshold （ Right now Kc, Kc-ΔK To avoid the shortcomings of fixed thresholds in adapting to the dynamic evolution of landslides, an adaptive learning algorithm based on bias feedback was developed, which enabled real-time optimization of stability thresholds and model weight coefficients. This significantly reduced the false alarm rate and false negative rate in long-term applications, and improved the reliability and accuracy of the system.

[0019] In this application, the three-tier risk classification of stable, potentially unstable, and unstable clearly defines the risk level boundaries, and the verification steps ensure the reliability of the judgment results, making the early warning information more targeted.

[0020] Based on the deviation between monitoring data and model output ΔK'The model parameters are optimized in real time by dynamically adjusting the weight coefficients of displacement, mechanics, and environmental references.

[0021] In this application, the modified rules are aligned with the evolution of risk (displacement weight increases with the increase of deviation, and environmental weight is positively correlated with environmental factors), so that the model can continuously adapt to the actual state of landslides and reduce the false alarm rate and false negative rate in long-term applications.

[0022] In this application, sensor deployment, threshold parameters (such as...) V thre , σ max The method selects reference industry standards (GB / T38509-2020, etc.) and historical data to adapt to complex geological environments such as high-altitude cold regions and mountainous highway / railway lines. This ensures the engineering applicability and promotional value of the method, solving the problem of landslide monitoring and early warning in special and complex environments. Attached Figure Description

[0023] Figure 1 A flowchart illustrating the overall process of the landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion provided in this application embodiment; Figure 2 This is a schematic diagram of the dynamic weight adjustment mechanism of the multi-parameter fusion model in the embodiments of this application; Figure 3 This is a schematic diagram of the landslide risk level determination logic in the embodiments of this application; Figure 4 The stability coefficient in the embodiments of this application K(t) A schematic diagram showing the relationship between the evolution process and the early warning threshold. Detailed Implementation

[0024] This invention discloses a landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion, the specific process of which is as follows: Figure 1 As shown, the method includes the following steps: S1. Real-time multi-source data of the landslide body is collected using various heterogeneous sensors, and the multi-source data is preprocessed to unify it to the same spatiotemporal reference, specifically as follows: S11, the implementation process of multi-source data acquisition is as follows: Multiple heterogeneous sensors are deployed in the landslide monitoring area to synchronously collect multi-source data of the landslide body at a set sampling frequency. These heterogeneous sensors include, but are not limited to: surface displacement monitoring sensors (such as GNSS receivers), deep deformation monitoring sensors (such as fixed inclinometers), environmental parameter monitoring sensors (such as rain gauges, groundwater level gauges, and thermometers), and mechanical parameter monitoring sensors (such as earth pressure cells). The multi-source data includes: surface displacement monitoring sensors acquiring three-dimensional coordinate change data (i.e., displacement field) of the landslide surface at a fixed sampling frequency; deep deformation monitoring sensors recording displacement or strain data (i.e., displacement field / mechanical field) at various depths within the landslide with high precision; and environmental parameter sensors synchronously recording rainfall R(t) (unit: mm / h), temperature T(t) (unit: ℃), and groundwater level. W(t) (Unit: m) and other external inducing factors (i.e., environmental field), as well as mechanical parameter sensors through strain-stress relationships (such as... ,in For stress, For elastic modulus, (To quantify the changes in the internal mechanical state of the landslide body for strain)

[0025] S12, The preprocessing procedure for multi-source data is as follows: In this application, multi-source data preprocessing includes denoising the collected multi-source data and spatiotemporal alignment of the denoised data.

[0026] The denoising process for multi-source data is implemented using either wavelet transform or moving average algorithms to remove high-frequency noise interference from the data.

[0027] After denoising, the data is synchronized across different sensors through timestamp calibration, and its spatiotemporal alignment is achieved by mapping the measurement points of different sensors to a unified spatial grid. Specifically, in the time dimension, interpolation algorithms are used to unify the time series of all sensor data to the same timestamp, achieving time synchronization; in the spatial dimension, a pre-established three-dimensional geological model of the landslide (e.g., using an XYZ coordinate system) is used to precisely map the coordinates of the measurement points of each sensor to a unified spatial grid, achieving spatial synchronization. This step ensures that all subsequent analyses are based on high-quality fused data from the same spatiotemporal reference.

[0028] S2. Based on the preprocessed data, a multi-parameter fusion model for dynamic simulation of the landslide is constructed by integrating the displacement field, mechanical field, and environmental field of the landslide body. Specifically: The inputs to the multi-parameter fusion model include: Displacement field: , , ,in, , , For the surface or interior of a landslide in time The amount of three-dimensional displacement change; Mechanical field: , , where σ(t) and ε(t) represent the stress and strain data within the landslide body, respectively; Environmental field: , , , Rainfall over time The cumulative value, Temperature over time The instantaneous value, For groundwater level in time The depth value; The output of the multi-parameter fusion model is: deformation rate. V(t) and stability coefficient K(t) Two key risk quantification indicators. One is the deformation rate. V(t) Calculated using the following formula:

[0029] Stability coefficient Based on the theory of limit equilibrium, the following formula is used for calculation:

[0030] in, c' For effective cohesion (unit: Pa or N / m) 2 ); A The sliding surface area of ​​the landslide body; N Normal force of the sliding surface (N = W cos θ) ; Pore ​​water pressure (Pa), For the effective friction angle, W Total weight of the landslide (W = pgV) θ is the inclination angle of the sliding surface. This coefficient comprehensively reflects the ratio of the resisting force to the sliding force of the landslide body; For the anti-sliding force of the landslide body, The sliding force of the landslide is determined by incorporating weighting coefficients of different parameters, such as displacement weight. Mechanical weights Environmental weight The sensitivity of the model to landslide risk is dynamically adjusted. If we use volume With density Indicates total weight ,but Substituting the stability coefficient into the above equation The calculation formula is:

[0031] in, Represents the total resistance to sliding of the landslide mass; The cohesive force of the landslide sliding surface. The product of the two is the area of ​​the sliding surface. It is the anti-slip force generated by the bonding effect of the sliding surfaces; The density of the landslide soil and rock. It is the acceleration due to gravity. The volume of the landslide is the product of the three factors. That is, the total weight of the landslide body ; The inclination angle of the sliding surface. It is the component of the total weight of the landslide body perpendicular to the sliding surface (i.e., the normal force of the sliding surface). ); For sliding face gap water pressure ， It is the total counteracting force generated by the pore water pressure, therefore The actual effective normal force; The effective friction angle of the sliding surface. It is the anti-slip force generated by the friction of the sliding surface.

[0032] The total sliding force of the landslide body, i.e., the total weight of the landslide body. The component of the force along the tilt direction of the sliding surface directly drives the landslide to slide. The larger the value, the higher the risk of landslide.

[0033] Overall ratio That is, the stability coefficient, the magnitude of which directly reflects the stability of the landslide. The larger the value, the stronger the anti-sliding force relative to the sliding force, and the more stable the landslide; conversely, the smaller the value, the worse the landslide stability, and the greater the risk of sliding.

[0034] In this application, a dynamic weighting mechanism is introduced to improve the model's representation accuracy at different landslide evolution stages. First, based on the normalized displacement rate... ΔV / V thre The different stages of landslide evolution are divided into three phases: initial deformation, accelerated deformation, and critical instability. Specifically: when When this occurs, it is the initial deformation stage; when At 5 o'clock, it is the accelerated deformation stage; when When this happens, it is the critical instability stage.

[0035] in, V thre The preset displacement rate threshold is set at 0.5 mm / d, which is determined by referring to the suggestions on the division of deformation stages of landslides in cold regions in the "Code for Design of Landslide Prevention Engineering" (GB / T38509-2020) and combining the statistical analysis of historical monitoring data of landslides around high-altitude cold region tunnel projects.

[0036] Displacement weights at each stage ω d Mechanical weights ω m and environmental weight ω e Calculated dynamically in the following ways: ω d =α F d ; ω m =β F m ; ω e = γ F e ; in, α , β , γ These are displacement, mechanical, and environmental parameters determined by fitting historical data. The benchmark weighting coefficient; F d , F m , F e为 Adjustment functions related to evolutionary stages, F d = , F m = ，F e = .

[0037] At each evolution stage, displacement weights were calculated using linear, power functions with an exponent of 1.5, and power functions with an exponent of 2, respectively. ω d Mechanical weights ω m and environmental weightω e This results in a sharp increase in the response to displacement and key environmental factors (such as heavy rainfall) when instability is imminent. The specific calculation process is as follows: In the initial deformation stage ( The formulas for calculating the weighting coefficients are as follows: , , ; During the accelerated deformation stage ( The formulas for calculating the weighting coefficients are as follows: , , ; In the critical instability stage ( The formulas for calculating the weighting coefficients are as follows: , , ; in, Displacement weights ω d Mechanical weights ω m and environmental weight ω e The empirical correction factor was determined by fitting historical data based on the geological conditions of the landslide. For displacement rate threshold, The maximum stress threshold, The maximum rainfall threshold, This refers to the maximum groundwater level threshold. The threshold values ​​for classifying landslide evolution stages in this application are as follows: , , Maximum threshold parameter ( , , All criteria were referenced from industry standards and authoritative research findings. Specifically, the threshold for classifying landslide evolution stages was determined by referencing the recommendations for classifying deformation stages of landslides in cold regions in the "Code for Design of Landslide Prevention Engineering" (GB / T38509-2020), combined with statistical analysis of historical monitoring data of landslides surrounding high-altitude cold-region tunnel projects (such as data from a landslide monitoring project in a tunnel on the Qinghai-Tibet Plateau from 2018 to 2023). The displacement rate threshold was also determined accordingly. The value is taken as the typical displacement rate (0.5 mm / d) at the initial stage of accelerated deformation in this type of landslide; maximum threshold parameter The mechanical parameters of frozen soil in cold regions are determined by referring to the limits of mechanical parameters of frozen soil in the Code for Geotechnical Investigation (GB50021-2001, 2009 edition) and combining the maximum compressive strength of the soil and rock layers of the landslide body in the geological investigation report of the tunnel engineering area. The maximum daily rainfall recorded by meteorological stations in the monitoring area over the past 30 years (e.g., the XX meteorological station near the project, 1994-2023) was taken. The highest historical groundwater level value revealed by drilling during the engineering survey phase was used. In this application, the displacement weight correction factor Mechanical weight correction coefficient and environmental weight correction factor The specific criteria for determining the initial values ​​are as follows: Combining geological conditions such as landslide lithology, structural surface development, and water content, historical landslide monitoring data from the same region or similar geological background are selected. The initial values ​​are determined using a multiple linear regression fitting method, with the fitting formula being: Displacement weight correction factor :

[0038] Mechanical weight correction factor :

[0039] Environmental weighting adjustment factor :

[0040] in, This represents the number of historical data samples. For the first The actual stability coefficient of the landslide body in the historical data set (obtained through on-site measurement). This is the average of the actual stability coefficients across all historical data. , , , respectively The calculated values ​​of displacement weight, mechanical weight, and environmental weight in the historical data set (derived from the corresponding stage weight formula). 、 、 These are the average values ​​calculated for the corresponding weights.

[0041] By constructing a unified model, multi-dimensional monitoring information is transformed into quantifiable landslide deformation rates. With stability coefficient Simultaneously, the displacement weights are dynamically adjusted according to different stages of landslide evolution. Mechanical weights Environmental weight The calculation method enables the model output to accurately reflect the risk sensitivity characteristics of landslides at different stages, providing scientific and unified quantitative indicators for subsequent analysis.

[0042] S3. Dynamic simulation and trend prediction.

[0043] The real-time output of S2 V(t) and K(t) Combined with the geological structure parameters of the landslide body (such as sliding...) face angle θ Cohesion c' (etc.) and historical deformation data (such as the maximum displacement in the past 30 days) , , ), input into a physics-based or data-driven dynamic simulation algorithm (if applicable) In finite element simulation and time series prediction models, landslides are predicted to occur in specific future periods (such as the short term). Deformation trends (72 hours, long-term 30 days). This is achieved by comparing the predicted deformation trends with subsequent 72-hour (72-hour, 30-day) deformation trends. Compare the measured deformation data within the time frame to confirm that the simulation result error does not exceed 5%; if the error exceeds the range, repeat the prediction process of this step.

[0044] S4. Risk assessment and risk level determination.

[0045] The real-time output of S2 K(t) and / or the future predicted by S3 K(t) The value is compared with a preset stability threshold. The risk level is then determined based on the comparison result. In one embodiment of this application, two threshold levels are set: a critical stability coefficient and a threshold value. Kc and warning threshold Kc -ΔK Among them, dynamic safety margin ΔK The calculation formula is: ; μ The basic safety margin factor; V thre This is the displacement rate threshold. ΔK It increases with the deformation rate V(t), making the warning threshold more stringent when the landslide accelerates.

[0046] The risk level determination rules are as follows (see Figure 3 ): when At that time, the risk level is Level 1, which is considered stable; when At that time, the risk level was Level 2 and potentially unstable; when At that time, the risk level was level three, which means unstable. This is a safety margin that can be dynamically adjusted. μ If we take 0.1, then The formula for calculation is: ; Among them, the dynamically adjusted safety margin As a key threshold difference for classifying landslide risk levels and distinguishing between Level II "potentially unstable" and Level III "unstable," its core function is to adjust the safety threshold for stability assessment based on the real-time deformation state of the landslide, making the risk level determination more closely reflect the actual landslide evolution. ΔK The value will change dynamically with the landslide deformation rate to avoid misjudgment caused by a fixed safety margin, so that the warning threshold is more stringent when the landslide accelerates. in, μ A safety margin of 0.1 is taken as the basic safety margin coefficient, which is determined with reference to the recommended range of safety margin for landslide stability assessment in the "Technical Standard for Geological Disaster Monitoring Engineering" (GB / T33584-2017). It is a benchmark value for dynamic adjustment. (Unit: mm / d) refers to the current actual deformation rate of the landslide body, reflecting the current deformation activity of the landslide, and is calculated through a multi-parameter fusion model; This refers to the displacement rate threshold, which is a typical displacement rate value taken during the initial stage of accelerated deformation in this type of landslide, such as... Figure 4 As shown, Figure 4 Reaction stability coefficient Evolution over time, and with critical values Warning threshold The contrast relationship, in Figure 4 The blue curve represents the result of the simulation. The evolution process (gradually decreasing with noise) is represented by the green dashed line, which indicates the critical stability coefficient. K C =0.8, red dashed line: warning threshold , representing the potential sliding risk boundary.

[0047] S5. Output of early warning information and dynamic correction of multi-parameter fusion model.

[0048] Based on the risk level determined by S4, corresponding early warning information is automatically generated, specifically as follows: when K(t) ≥ Kc If the landslide is stable, the risk level is Level I, indicating that the landslide is stable and no warning is needed; the output is a stable information. when Kc - ΔK ≤ K(t) < KcIf the landslide occurs at this time, the risk level is Level II, which indicates that the landslide body has potential instability. At this time, a warning should be issued and landslide risk information should be output. When K(t) < Kc-ΔK If the landslide occurs at a certain time, the risk level is Level III. At this point, the landslide body is unstable, and an emergency warning should be issued. The warning information should be output and transmitted in real time to the warning platform via a communication module (such as 4G / 5G, satellite communication, etc.). At the same time, the adaptive correction process of the multi-parameter fusion model should be initiated. 1) Based on newly collected monitoring data, such as the current time of , , Deviation from simulation results .

[0049] 2) Based on the deviation value The displacement reference weight coefficient α, mechanical reference weight coefficient β, and environmental reference weight coefficient γ in the dynamically corrected multi-parameter fusion model are obtained as the corrected weight coefficients. , , .

[0050] in: ; ; By dynamically adjusting the displacement reference weight coefficient α Mechanical reference weighting coefficient β and environmental benchmark weighting coefficient γ This aims to improve the model's adaptability to landslide evolution and reduce false alarm rates.

[0051] The adaptive learning algorithm used in the dynamic correction step is based on dynamically optimizing the weight coefficients according to the deviation between real-time monitoring data and simulation results, ensuring that the model continuously adapts as the landslide evolves. The specific process is as follows: At the current moment Based on the measured data, the three-dimensional displacement increment of the landslide body at that moment was extracted. Internal stress Rainfall and groundwater level Substituting these measured parameters into the multi-parameter fusion model, the actual stability coefficient at the current moment is calculated. Subsequently, by comparing with the preset stability threshold... By comparison, the deviation value is obtained. This deviation value directly reflects the difference between the model simulation results and the actual landslide condition. Based on this, the current time step is introduced. (i.e., the time interval between two consecutive monitoring sessions, such as 1 hour or 1 day) Construct a correction factor λ=ΔK / (Kc Δtn) ,in Used for standardization deviation This correlates the time dimension of the correction, avoiding over-correction caused by short-term deviation fluctuations, so that the correction factor can reflect both the magnitude of the deviation and the monitoring frequency.

[0052] Based on the above correction factors For the initial empirical correction coefficient Targeted adjustments were made to obtain the corrected weighting coefficients: for displacement weighting correction coefficients ,use ,in This is the ratio of the current deformation rate to the rate threshold, ensuring that when the model's simulation of displacement parameters deviates significantly, [the following occurs]. The product of this ratio amplifies the correction magnitude, increasing the model's sensitivity to displacement changes; for the mechanical weight correction coefficient ,use , This is the ratio of the current stress to the maximum stress threshold. The negative sign helps prevent excessive weight bias when the mechanical parameter simulation deviation is too large, maintaining the stability of the model's assessment of the mechanical state; for the environmental weight correction coefficient... ,use By superimposing the ratios of rainfall, groundwater level, and their respective maximum thresholds, the algorithm enhances the correction effect of environmental causal simulation bias. The core value of this adaptive algorithm lies in its ability to automatically optimize weights as monitoring data accumulates without human intervention. This allows the model's sensitivity to displacement, mechanical, and environmental parameters to continuously match the actual evolution stage of the landslide, fundamentally reducing early warning bias caused by fixed parameter weights and improving long-term early warning accuracy.

[0053] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion, characterized in that, The method includes the following steps: S1 collects multi-source data of the landslide body in real time through a variety of heterogeneous sensors, and preprocesses the multi-source data to unify it to the same spatiotemporal reference. S2, based on the preprocessed data, constructs a multi-parameter fusion model for dynamic simulation of landslides by integrating the displacement field, mechanical field and environmental field of the landslide body; The parameter fusion model outputs the deformation rate of the landslide. V(t) and stability coefficient K(t) Furthermore, the multi-parameter fusion model dynamically calculates displacement weights based on the evolution stages of the landslide. ω d Mechanical weights ω m Environmental weight ω e , wherein the displacement weight ω d Mechanical weights ω m Environmental weight ω e It is calculated based on the corresponding benchmark weight coefficients and evolution stages; S3, based on the deformation rate output in S2 V(t) and stability coefficient K(t) By combining the geological structure parameters and historical deformation data of the landslide, a dynamic simulation algorithm is used to predict the future deformation trend of the landslide. S4, the real-time stability coefficient output from S2 K(t) and the future stability system predicted in S3 The data is compared with a preset stability threshold, and the risk level of the landslide is output based on the comparison results. S5. Generate corresponding early warning information based on the risk level and transmit it to the early warning platform. At the same time, dynamically correct the benchmark weight coefficients in the multi-parameter fusion model based on the deviation between the current monitoring data and the model output.

2. The landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion according to claim 1, characterized in that, In S1, the multi-source monitoring data includes displacement parameters of the landslide surface, displacement or strain data inside the landslide body, external inducing factors of the landslide body, and data on changes in the internal mechanical state of the landslide body. The external triggering factors include rainfall and groundwater level in the landslide area.

3. The landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion according to claim 1, characterized in that, In S1, the preprocessing of multi-source data includes noise reduction and spatiotemporal alignment. The spatiotemporal alignment process achieves time synchronization of data from different sensors through timestamp calibration, and spatial synchronization by locating the measurement points of different sensors to a unified spatial grid through spatial coordinate mapping, so as to ensure the comparability of data in both time and space dimensions.

4. The landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion according to claim 1, characterized in that, In S3, the predicted future deformation trend includes at least one of the short-term trend and the long-term trend.

5. The landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion according to claim 1, characterized in that, In S4, the stability threshold includes a first stability threshold. Kc Second stability threshold Kc-ΔK ,in, ΔK This is a dynamic safety margin; The specific process for determining the risk level of a landslide based on comparison results includes: when K(t) ≥ Kc At that time, the risk level was determined to be stable; when Kc - ΔK ≤ K(t) < Kc At that time, the risk level was determined to be potentially unstable; when K(t) < Kc - ΔK At that time, the risk level was determined to be unstable.

6. The landslide dynamic simulation monitoring and prediction system based on multi-source sensor fusion as described in claim 1 The alarm method is characterized by, The dynamic safety margin ΔK Calculated using the following formula: ； in, μ Based on the basic safety margin factor, V thre This is the preset displacement rate threshold.

7. The method according to claim 1, characterized in that, Step S4 further includes a verification step; the verification step is used to confirm the predicted deformation trend and / or calculate the stability coefficient. K(t) The simulation results do not exceed the preset allowable error range.

8. The landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion according to claim 1, characterized in that, In S5, the specific process of dynamically correcting the weight coefficients in the multi-parameter fusion model includes: Calculate the measured value of the stability coefficient at the current moment. K(t)´ With critical stability coefficient Kc deviation value ΔK´ ; According to the deviation value ΔK´ The baseline weighting coefficients corresponding to displacement, mechanics, and environment are corrected.

9. The landslide dynamic simulation monitoring and early warning method based on multi-source sensor fusion according to claim 8, characterized in that, The correction rule for the benchmark weighting coefficient is as follows: The baseline weighting coefficient of displacement varies with ΔK´ As it increases, the mechanical reference weight coefficient increases accordingly. ΔK´ Increase and decrease, the baseline weighting coefficient of the environment and ΔK´ It is positively correlated with current environmental factors.

10. The method according to claim 1, characterized in that, The evolution stages of the landslide are based on the normalized displacement rate. The process is divided into three stages: initial deformation, accelerated deformation, and critical instability. Different functional relationships are used to calculate the displacement weights at each stage. ω d Mechanical weights ω m and environmental weight ω e .

Images