Multi-sensor fusion intelligent slope stability analysis system and edge computing method

Abstract

The present invention relates to the technical field of slope stability data identification, and in particular provides a multi-sensor fusion intelligent slope stability analysis system and edge computing method. The system includes an environmental perception subsystem that uses a soil moisture sensor to monitor changes in soil moisture content, a displacement sensor to obtain small slope movement data, and a pressure sensor to sense changes in ground pressure. At the same time, abnormal data is pre-screened to form a fused data set. The data analysis subsystem converts the abnormal data in the fused data set into a quantitative indicator of slope stability, and infers the potential sliding risk of the slope by analyzing the correlation between soil moisture content, small movement data, or ground pressure. The decision support subsystem proposes specific reinforcement countermeasures based on the analysis results of the potential sliding risk. The present invention comprehensively infers the potential sliding risk of the slope by analyzing the correlation between soil moisture content, small movement data, and ground pressure.

Description

Technical Field

[0001] The present invention relates to the technical field of slope stability data identification, and in particular to a multi-sensor fusion slope stability intelligent analysis system and an edge computing method. Background Art

[0002] Slope stability analysis is a crucial task in engineering geology, involving a comprehensive assessment of multiple factors, including slope deformation, stress, and hydrogeology. With the advancement of computer technology and sensor technology, researchers are beginning to explore intelligent approaches to slope stability analysis. Traditional analysis methods rely on limited sensor data and expert experience, resulting in information insufficiency and low efficiency. Therefore, it is crucial to develop a sensor-fusion-based intelligent slope stability analysis technology that can integrate multi-source sensor data to improve analysis accuracy and real-time performance.

[0003] Prior art 1, Chinese patent application number: 202210302009.2, discloses a multi-sensor data fusion method based on a piecewise function. The method comprises the following steps: acquiring multiple sensor data and statistically outputting target distribution; analyzing the stability of the multiple sensor data and outputting a transition flag; fitting the target distribution based on a piecewise function and outputting a weight curve; and fusing the multiple sensor data based on the weight curve to obtain fused data. While this method offers simpler steps and lower computing power requirements compared to existing methods, and also adjusts the weight curve by analyzing the stability of the multiple sensor data, making the fusion of multiple sensor data more accurate and reliable, it lacks dynamic coupling analysis of multiple physical fields. This technique achieves data fusion by adjusting the weight curve using a piecewise function, but fails to consider the dynamic interactions between parameters such as soil moisture, displacement, and pressure in slope stability analysis, and thus fails to reveal the mechanism by which multi-physical coupling influences slip risk. Limitations of static weight allocation: Weight adjustment based on stability analysis still relies on a static model, making it difficult to adapt to the spatiotemporal variations of sensor data during the dynamic evolution of the slope, resulting in a deviation between the fusion results and the actual risk.

[0004] The second prior art, Chinese patent application number 202210248874.3, discloses a situation assessment method that uses a probabilistic reasoning model to fuse multi-source heterogeneous information. In the first step, within a military situation data fusion system, a strategy of enabling a specific sensor is employed to gather more information. This strategy may also expose the sensor's location to the enemy, prompting them to take action to effectively change the situation. Secondly, the uncertainty in the fusion reasoning process reflects the dynamic accumulation and propagation of uncertainty in multi-source information. At each fusion step, the uncertainty factors of multi-source information must be integrated, and as the reasoning proceeds, an uncertain conclusion is ultimately reached. While this method addresses the impact of various uncertainties in the fusion system and effectively improves its stability, its application in certain scenarios is limited. Designed for military situation assessment, this technology requires the active exposure of sensor locations to obtain information, making it unsuitable for slope scenarios requiring long-term, covert monitoring. Furthermore, it lacks real-time dynamic feedback: its uncertainty reasoning mechanism is not embedded in the dynamic closed loop between real-time sensor data and reinforcement measures, making it impossible to adjust early warning strategies in real time based on changing slope conditions.

[0005] Prior art three, Chinese patent application number 202210504092.1, discloses a multi-sensor asynchronous information fusion method and system. This method extracts filter kernel information from the collected raw data of multiple sensors, uses the filter kernel information to calculate the number of clusters k, then analyzes and calculates the number of clusters k based on the number of clusters k. The calculated center data information of each cluster is weighted and combined to solve the problem before information fusion is performed to complete the multi-sensor asynchronous information fusion. Although cluster analysis based on the filter kernel information of the raw data of multiple sensors can effectively avoid sensor information asynchrony or packet loss, the analysis and calculation based on the number of clusters K to obtain the center data information of each cluster ensures the stability of information fusion, reduces the difficulty of fusion, and solves the problem of difficult real-time multi-sensor fusion in practical applications. However, it ignores physical correlation: this technology processes asynchronous data through clustering but does not incorporate the actual physical correlation of parameters such as soil moisture, displacement, and pressure (for example, increased moisture leads to decreased soil strength), resulting in a disconnect between the fused data and the slope instability mechanism. Failure to pre-screen abnormal data: Directly clustering and fusing the raw data may introduce sensor noise or outlier interference, affecting the reliability of the final risk assessment.

[0006] Currently, existing technologies 1, 2, and 3 have problems such as disconnection between data fusion and physical mechanisms, significant early warning lag, and lack of decision support. Therefore, the present invention provides a multi-sensor fusion intelligent slope stability analysis system and edge computing method. Summary of the Invention

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] In one aspect of the present invention, a multi-sensor fusion intelligent slope stability analysis system is provided, comprising:

[0009] The environmental perception subsystem uses soil moisture sensors to monitor changes in soil moisture content, displacement sensors to acquire small slope movement data, and pressure sensors to sense changes in ground pressure. It also pre-screens abnormal data to form a fused data set.

[0010] The data analysis subsystem is used to convert abnormal data in the fused dataset into quantitative indicators of slope stability. By analyzing the correlation between soil moisture content, micro-movement data, or ground pressure, it can infer the potential slide risk of the slope.

[0011] The decision support subsystem is used to propose specific reinforcement measures based on the analysis results of potential sliding risks.

[0012] In an optional implementation, the environment perception subsystem includes:

[0013] The multi-dimensional perception module is used to obtain a three-dimensional model of the slope to be detected and determine the layout coordinates of soil moisture sensors, displacement sensors, and pressure sensors based on historical slope landslide risks. This forms a slope data detection network consisting of soil moisture sensors, displacement sensors, and pressure sensors.

[0014] The anomaly pre-screening module is used to initialize the slope data detection network and compare the fluctuation trend of slope data within the same time window. It sets a dynamic threshold based on the physical coupling law of moisture-displacement-pressure. It then obtains abnormal soil moisture content, micro-displacement data, and ground pressure data in the slope data detection network based on the preset fluctuation trend standard and dynamic threshold.

[0015] The data fusion module is used to receive abnormal soil moisture content, micro-movement data and ground pressure data, and use the layout coordinates of soil moisture sensors, displacement sensors and pressure sensors as tags to form a fused data set containing multi-dimensional associations of tags.

[0016] In an optional implementation, the multi-dimensional perception module includes:

[0017] The 3D model construction submodule is used to obtain the surface geometric data of the slope to be tested. Through point cloud processing and surface reconstruction, a 3D model is generated, which includes the elevation, slope, and curvature geometric characteristics of the terrain. It also uses historical landslide data including the location of the slip surface and the extent of the damage to demarcate potential risk areas, forming a digital slope model with geological risk markers.

[0018] The risk weight calculation submodule is used to calculate the risk weights of different slope areas to be tested based on the digital slope model, taking into account the location of the slip surface and the spatial distribution of the damage range, the damage depth and the triggering factors;

[0019] The sensor coordinate optimization submodule is used to determine the mechanical coupling relationship between soil moisture, displacement, and pressure, and set the layout conditions of the corresponding sensors; through iterative optimization, the spatial matching coordinates of the three types of sensors are determined.

[0020] In an optional embodiment, the abnormality pre-screening module includes:

[0021] The multi-source data spatiotemporal alignment submodule is used to unify the spatiotemporal benchmarks of three types of data: soil moisture, displacement, and ground pressure, based on the constructed slope data monitoring network;

[0022] The coupled feature extraction submodule is used to establish a correlation analysis framework for three types of parameters: moisture-displacement correlation, displacement-pressure correlation, and moisture-pressure correlation, based on the physical properties of the rock and soil mass of the slope to be tested;

[0023] The dynamic threshold generation submodule uses a sliding time window mechanism to construct adaptive discrimination criteria for baseline thresholds, trend corrections, and spatial corrections; implements a three-level discrimination mechanism for single-point anomalies, local anomalies, and system anomalies; and performs structured processing on confirmed anomaly data by adding spatiotemporal tags, adding confidence levels, and recording the evolution process.

[0024] In an optional embodiment, the abnormality pre-screening module includes:

[0025] The time dimension of the multi-source data spatiotemporal alignment submodule uses the data acquisition timestamp as the benchmark to establish a millisecond-level synchronization mechanism; the spatial dimension uses the sensor layout coordinate information to construct a three-dimensional spatial reference system; the data dimension converts raw data of different dimensions into standard monitoring indicators;

[0026] The coupled feature extraction submodule uses moisture-displacement correlation to analyze soil deformation response caused by seepage; displacement-pressure correlation to monitor stress redistribution caused by soil sliding; and moisture-pressure correlation to evaluate the effect of pore water pressure on effective stress.

[0027] The baseline threshold of the dynamic threshold generation submodule establishes the initial judgment standard based on the historical stable period data; the trend correction adjusts the threshold sensitivity according to the recent data change rate; the spatial correction is set differently according to the risk weight of the sensor location; the single point anomaly is that the data of a single sensor exceeds the dynamic threshold range; the local anomaly is that the adjacent sensor group has a coordinated anomaly. System anomaly: the three types of parameters present an abnormal combination that is consistent with the damage mechanism.

[0028] In an optional implementation, the coupled feature extraction submodule includes:

[0029] The parameter association model construction unit is used to gradually establish the association framework of three types of parameters based on the physical characteristics of the slope rock and soil. The coupled feature extraction submodule adopts a progressive analysis strategy;

[0030] A correlation logic confirmation unit is used to set the analysis logic of humidity-displacement correlation, displacement-pressure correlation and humidity-pressure correlation;

[0031] The anomaly coupling verification unit is used to cross-validate the results of the three types of correlation analysis; if the anomaly signal is valid in multiple correlation models, it is marked as a high-confidence anomaly.

[0032] In an optional embodiment, the data analysis subsystem includes:

[0033] A logic startup module is used to receive abnormal data, start the corresponding analysis logic of humidity-displacement correlation, displacement-pressure correlation and humidity-pressure correlation, and filter abnormal data according to the corresponding input content;

[0034] Coupling analysis module, used for multi-analysis coupling of humidity, displacement, and pressure and risk quantification;

[0035] The collaborative verification module is used for multi-parameter collaborative verification and risk classification, checking whether the three types of parameter anomalies meet the typical combination pattern of slope instability; and dividing the risk level according to the number of abnormal parameters, coupling strength and spatial distribution range.

[0036] In an optional implementation, the collaborative verification module has the following local low risks: single parameter anomaly; local high risk dual parameter coupling anomaly; and systemic risk: all three types of parameters are abnormal and consistent with the damage mechanism.

[0037] In an optional implementation, the decision support subsystem includes:

[0038] The risk feature analysis and classification module is used to receive risk assessment results from the data analysis subsystem, including risk type labeling, risk level classification, and abnormal characteristics of key parameters, and automatically construct a risk feature matrix;

[0039] Targeted measures matching and optimization module, used to initiate graded response strategies based on the risk feature matrix;

[0040] The spatial adaptation and strength calibration module is used to adjust the coverage density of engineering measures according to the spatial distribution density of abnormal parameters, determine the design strength level of the structure based on the quantitative value of the risk level, consider the temporal characteristics of parameter anomalies, and plan the priority of measure implementation.

[0041] Another aspect of the present invention provides a multi-sensor fusion slope stability edge computing method of the multi-sensor fusion slope stability intelligent analysis system, the multi-sensor fusion slope stability edge computing method comprising the following steps:

[0042] The edge computing method of slope stability based on multi-sensor fusion includes the following steps:

[0043] Deploy micro-computing units at the edge nodes of sensor networks to achieve in-situ feature extraction, dynamic data compression, cross-modal alignment, and lightweight fusion;

[0044] Build a three-level analysis architecture with rapid response at the edge layer, correlation analysis at the fog layer, and deep verification at the cloud layer;

[0045] Establish a dynamically updated edge intelligence model that includes cloud model slimming, incremental learning, cross-node knowledge sharing, and working condition adaptation.

[0046] The present invention comprehensively monitors and pre-processes data. Through the environmental perception subsystem, it monitors soil moisture, slope displacement, and ground pressure, and pre-screens abnormal data, ensuring the accuracy and reliability of the data and providing real and effective input information for potential sliding risk analysis. In-depth analysis and risk assessment: The data analysis subsystem converts the pre-processed data into quantitative indicators of slope stability. By analyzing the correlation between soil moisture content, micro-movement data, and ground pressure, it infers the potential sliding risk of the slope based on these factors, providing a scientific assessment basis for slope safety. Timely response and decision support: The decision support subsystem proposes specific reinforcement measures based on the analysis results, so that corresponding preventive measures can be taken for different risk levels, thereby reducing or even avoiding the occurrence of slope sliding accidents. The present invention realizes comprehensive and continuous monitoring and real-time assessment of slope stability, providing scientific and accurate decision support for slope engineering management, thereby effectively preventing and reducing the risk of slope disasters and protecting people's lives and property. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0048] Figure 1 This is a block diagram of the multi-sensor fusion slope stability intelligent analysis system provided in Example 1 of the present invention;

[0049] Figure 2 This is a block diagram of the environment perception subsystem provided in Example 2 of the present invention;

[0050] Figure 3 This is a block diagram of the data analysis subsystem provided in Example 6 of the present invention;

[0051] Figure 4 This is a block diagram of the decision support subsystem provided in Example 7 of the present invention;

[0052] Figure 5 This is a flow chart of the edge computing method for slope stability based on multi-sensor fusion provided in Example 8 of the present invention;

[0053] Figure 6 A block diagram of the electronic device provided by the present invention;

[0054] Figure 7 A block diagram of a computer-readable storage medium provided by the present invention. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present invention will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0056] In the following, the terms "first," "second," etc., are used for descriptive convenience only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature identified with "first," "second," etc., may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0057] In the present invention, unless otherwise clearly specified and limited, the term "connection" should be understood in a broad sense. For example, "connection" can be a fixed mechanical connection, a detachable mechanical connection, or an integrated one; or, "connection" can be a direct connection or an indirect connection through an intermediate medium. In addition, unless otherwise clearly specified and limited, the term "coupling" should be understood in a broad sense. For example, "coupling" can be a direct electrical connection, such as physical contact and electrical conduction between two components, or it can be understood as the electrical connection between different components in a circuit structure through a physical line that can transmit electrical signals, such as printed circuit board (PCB) copper foil or wire, so as to transmit electrical signals; or, "coupling" can be an indirect electrical connection between two components through an intermediate medium; or, "coupling" can be an electrical connection between two components in an airless / non-contact manner, such as electrical connection between two components using capacitive coupling to transmit electrical signals.

[0058] In an embodiment of the present invention, directional terms such as "up", "down", "left" and "right" may be defined including but not limited to the orientation relative to the schematic placement of the components in the drawings. It should be understood that these directional terms may be relative concepts, which are used for relative description and clarification, and may change accordingly according to changes in the orientation of the components in the drawings.

[0059] Example 1:

[0060] like Figure 1 As shown, an embodiment of the present invention provides a multi-sensor fusion intelligent slope stability analysis system, comprising:

[0061] The environmental perception subsystem uses soil moisture sensors to monitor changes in soil moisture content, displacement sensors to acquire small slope movement data, and pressure sensors to sense changes in ground pressure. It also pre-screens abnormal data to form a fused data set.

[0062] The data analysis subsystem is used to convert abnormal data in the fused dataset into quantitative indicators of slope stability. By analyzing the correlation between soil moisture content, micro-movement data, or ground pressure, it can infer the potential slide risk of the slope.

[0063] The decision support subsystem is used to propose specific reinforcement measures based on the analysis results of potential sliding risks.

[0064] In the above embodiment, comprehensive monitoring and data preprocessing, through the environmental perception subsystem to monitor soil moisture, slope displacement and ground pressure, and pre-screen abnormal data, ensure the accuracy and reliability of the data, and provide real and effective input information for potential sliding risk analysis. In-depth analysis and risk assessment: The data analysis subsystem converts the preprocessed data into quantitative indicators of slope stability. By analyzing the correlation between soil moisture content, micro-movement data and ground pressure, these factors are combined to infer the potential sliding risk of the slope, which provides a scientific assessment basis for slope safety. Timely response and decision support: The decision support subsystem proposes specific reinforcement response measures based on the analysis results, so that corresponding preventive measures can be taken for different risk levels, thereby reducing or even avoiding the occurrence of slope sliding accidents.

[0065] In summary, this embodiment realizes comprehensive, continuous monitoring and real-time assessment of slope stability, provides scientific and accurate decision-making support for slope engineering management, and thus effectively prevents and reduces the risk of slope disasters, protecting people's lives and property.

[0066] Example 2:

[0067] like Figure 2As shown, based on Example 1, the environment perception subsystem provided by this embodiment of the present invention includes:

[0068] The multi-dimensional perception module is used to obtain a three-dimensional model of the slope to be detected and determine the layout coordinates of soil moisture sensors, displacement sensors, and pressure sensors based on historical slope landslide risks. This forms a slope data detection network consisting of soil moisture sensors, displacement sensors, and pressure sensors.

[0069] The anomaly pre-screening module is used to initialize the slope data detection network and compare the fluctuation trend of slope data within the same time window. It sets a dynamic threshold based on the physical coupling law of moisture-displacement-pressure. It then obtains abnormal soil moisture content, micro-displacement data, and ground pressure data in the slope data detection network based on the preset fluctuation trend standard and dynamic threshold.

[0070] The data fusion module is used to receive abnormal soil moisture content, micro-movement data and ground pressure data, and use the layout coordinates of soil moisture sensors, displacement sensors and pressure sensors as tags to form a fused data set containing multi-dimensional associations of tags.

[0071] In the above embodiment, the environmental perception subsystem of this embodiment realizes the intelligent monitoring and early warning of slope risks as a whole through the collaborative work of multiple modules. The multi-dimensional perception module realizes the precise spatial positioning of sensor deployment based on historical landslide data and three-dimensional terrain modeling; through the topological deployment of three types of sensors, namely soil moisture, displacement, and pressure, a three-dimensional data acquisition matrix covering the key mechanical nodes of the slope is formed, providing a spatial basic data source. The abnormal pre-screening module establishes a dynamic coupling analysis model of moisture-displacement-pressure, and realizes the joint verification of cross-sensor data through the comparison of time series data fluctuation trends and adaptive threshold adjustment. The screening mechanism based on the principles of geotechnical mechanics significantly improves the reliability of single parameter anomaly detection. The data fusion module reconstructs discrete abnormal data into a multidimensional data set with spatial topological relationships through sensor coordinate marking; the fusion method retains the spatial distribution characteristics of abnormal events, and provides a standardized input with both time series characteristics and spatial correlation structure for subsequent landslide prediction models.

[0072] In summary, this embodiment forms a closed technical loop of "spatial deployment-dynamic detection-correlation fusion," achieving a complete processing chain from raw data acquisition to risk feature extraction, providing a standardized and structured data foundation for slope stability analysis. The cascade design of each module ensures the physical consistency of monitoring data from acquisition to processing, avoiding the information fragmentation problem common in traditional single-point monitoring.

[0073] Example 3:

[0074] Based on Example 2, the multi-dimensional perception module provided by the embodiment of the present invention includes:

[0075] The 3D model construction submodule is used to obtain the surface geometric data of the slope to be tested. Through point cloud processing and surface reconstruction, a 3D model is generated, which includes geometric features such as terrain elevation, slope, and curvature. It also uses historical landslide data including the location of the slip surface and the extent of damage to demarcate potential risk areas, forming a digital slope model with geological risk markers.

[0076] The risk weight calculation submodule is used to calculate the risk weights of different slope areas to be tested based on the digital slope model, taking into account the location of the slip surface and the spatial distribution of the damage range, the damage depth and the triggering factors;

[0077] The sensor coordinate optimization submodule is used to determine the mechanical coupling relationship between soil moisture, displacement, and pressure, and set the corresponding sensor layout conditions. Through iterative optimization, the spatial matching coordinates of the three types of sensors are determined.

[0078] Humidity sensors should be placed preferentially in areas prone to seepage (e.g., inside the slope, near potential slip surfaces);

[0079] Displacement sensors are distributed along the potential sliding direction (e.g., the main axis of slope inclination) to capture the accumulation of small displacements;

[0080] Pressure sensors are concentrated in load-sensitive areas (such as slope toe and near supporting structures) to monitor stress concentration phenomena.

[0081] In the above-mentioned embodiment, the multi-dimensional perception module of this embodiment, through the synergistic effect of its submodules, constructs a complete slope monitoring sensor network optimization system. From data acquisition to final deployment, it achieves: The spatial risk quantification system is constructed. The three-dimensional model construction submodule integrates raw terrain data with historical disaster data to create a digital model that includes geometric features and risk markers. The risk weight calculation submodule builds on this foundation and spatially quantifies slope risk, converting qualitative risk assessments into computable spatial weight distributions. The sensor coordinate optimization submodule, based on geomechanical principles, establishes sensor placement constraints for humidity, displacement, and pressure. Through a parametric iterative approach, it achieves optimal spatial matching of the three sensor types, ensuring that the monitoring network can simultaneously capture the key coupling parameters of slope instability. An adaptive monitoring architecture is implemented. Through a closed-loop process from model construction to coordinate optimization, the system develops a monitoring solution that can be dynamically adjusted according to actual geological conditions. This system considers historical disaster patterns and incorporates real-time mechanical coupling relationships, enabling the sensor network to adapt to different slope characteristics.

[0082] In summary, this embodiment achieves an upgrade from empirical deployment to model-driven deployment of slope monitoring through the technical chain of geometric modeling-risk assessment-sensor optimization. The data transmission and logical connection of each submodule ensure that the final monitoring network has both reasonable spatial coverage and correlation of physical parameters, providing an optimized data collection basis for slope stability analysis.

[0083] Example 4:

[0084] Based on Example 2, the abnormality pre-screening module provided in this embodiment of the present invention includes:

[0085] The multi-source data spatiotemporal alignment submodule is used to unify the spatiotemporal benchmarks of three types of data: soil moisture, displacement, and ground pressure, based on the constructed slope data monitoring network;

[0086] Among them, the time dimension: based on the data collection timestamp, a millisecond-level synchronization mechanism is established;

[0087] Spatial dimension: Use sensor coordinate information to build a three-dimensional spatial reference system;

[0088] Data dimension: converting raw data of different dimensions into standard monitoring indicators;

[0089] The coupled feature extraction submodule is used to establish a correlation analysis framework for three types of parameters: moisture-displacement correlation, displacement-pressure correlation, and moisture-pressure correlation, based on the physical properties of the rock and soil mass of the slope to be tested;

[0090] Among them, humidity-displacement correlation: analyzes the soil deformation response caused by seepage; displacement-pressure correlation: monitors the stress redistribution caused by soil sliding; humidity-pressure correlation: evaluates the impact of pore water pressure on effective stress;

[0091] The dynamic threshold generation submodule uses a sliding time window mechanism to construct adaptive discrimination criteria for baseline thresholds, trend corrections, and spatial corrections; implements a three-level discrimination mechanism for single-point anomalies, local anomalies, and system anomalies; and performs structured processing on confirmed anomaly data by adding spatiotemporal tags, adding confidence levels, and recording the evolution process.

[0092] Among them, the benchmark threshold: establishes the initial judgment standard based on the historical stable period data; trend correction: adjusts the threshold sensitivity according to the recent data change rate; spatial correction: differentiates the risk weight according to the location of the sensor; single point anomaly: a single sensor data exceeds the dynamic threshold range; local anomaly: collaborative anomalies occur in adjacent sensor groups; system anomaly: the three types of parameters present an abnormal combination that is consistent with the damage mechanism.

[0093] In the aforementioned embodiments, this embodiment implements intelligent anomaly identification and analysis of slope monitoring data. The specific significance is as follows: Through the multi-source data spatiotemporal alignment submodule, the spatiotemporal benchmarks of three heterogeneous data types—soil moisture, displacement, and ground pressure—are unified. This resolves analysis challenges caused by asynchronous acquisition times, dispersed spatial locations, and dimensional differences in monitoring data, providing standardized data input for correlation analysis. The coupled feature extraction submodule establishes a three-dimensional moisture-displacement-pressure correlation analysis framework that overcomes the limitations of traditional single-parameter monitoring, capturing the mechanical interactions between geotechnical parameters and more accurately reflecting the induction mechanisms and evolution processes of slope instability. A sliding time window mechanism enables adaptive updating of the discrimination criteria. Its three-level anomaly discrimination mechanism (single-point / local / systematic anomalies) distinguishes between random errors, local hazards, and precursors to systemic instability, significantly improving the accuracy of early warnings. A structured processing approach that adds spatiotemporal tags, confidence assessments, and evolution records to confirmed anomalies preserves the complete context of the anomaly data while providing a quantitative basis for subsequent risk assessment. It realizes the automated processing of slope monitoring from raw data to anomaly identification, and provides analysis results with spatiotemporal dimensions and multi-parameter correlation characteristics for engineering early warning.

[0094] Example 5:

[0095] Based on Example 4, the coupling feature extraction submodule provided in this embodiment of the present invention includes:

[0096] The parameter association model construction unit is used to gradually establish the association framework of three types of parameters based on the physical characteristics of the slope rock and soil. The coupled feature extraction submodule adopts a progressive analysis strategy;

[0097] A correlation logic confirmation unit is used to set the analysis logic of humidity-displacement correlation, displacement-pressure correlation and humidity-pressure correlation;

[0098] The humidity-displacement correlation inputs are standardized soil moisture (humidity) and displacement data. The analysis logic is that changes in soil moisture (such as rainfall seepage) affect the mechanical properties of the soil, leading to deformation. By analyzing historical stable period data, a typical relationship between humidity changes and displacement responses is established. When abnormal humidity fluctuations are output, the possible displacement development trend is predicted and the correlation strength is quantified.

[0099] The displacement-pressure correlation inputs displacement data and ground pressure data (aligned in time and space). The output of the moisture-displacement correlation (e.g., whether anomalies in a certain area are caused by seepage) relies on previous results. The analysis logic states that soil sliding can lead to internal stress redistribution. Combined with displacement trends, the pressure sensor data is analyzed to determine whether it exhibits a response consistent with mechanical laws (e.g., pressure decreases in areas with increased displacement and increases in adjacent areas). The output determines whether displacement anomalies are accompanied by pressure changes consistent with mechanical expectations, thereby determining the risk of sliding.

[0100] The humidity-pressure correlation inputs humidity and pressure data and relies on the results of the previous humidity-displacement and displacement-pressure analyses (e.g., whether increased humidity first triggers displacement and then indirectly affects pressure). The analysis logic indicates that increased soil moisture content increases pore water pressure and reduces effective soil stress. By analyzing the direct relationship between humidity and pressure, the influence of displacement can be eliminated (e.g., in certain areas, increased humidity without significant displacement changes, but abnormal pressure fluctuations may indicate potential liquefaction or softening risks). The output independently assesses the direct impact of pore water pressure on slope stability, addressing the blind spots of the previous two correlations.

[0101] Anomaly coupling verification unit, used to cross-validate the results of the three types of correlation analysis; if an abnormal signal is established in multiple correlation models, it is marked as a high-confidence anomaly;

[0102] Single correlation anomaly verification checks whether the three types of correlations, humidity-displacement, displacement-pressure, and humidity-pressure, are independent (e.g., whether an increase in humidity is accompanied by an increase in displacement, and whether the pressure change is as expected).

[0103] Multi-correlation collaborative verification: If all three types of parameters show abnormalities, determine whether their combination is consistent with the slope failure mechanism (e.g., humidity increase → displacement acceleration → pressure drop, consistent with the slip trend);

[0104] Confidence assessment calculates the comprehensive anomaly confidence based on the correlation strength (e.g., humidity-displacement correlation coefficient) and the degree of anomaly (e.g., the amplitude exceeding the threshold);

[0105] Output: Coupling anomaly marking: If the anomaly signal is established in multiple correlation models, it will be marked as high-confidence anomaly type classification: distinguish between seepage-dominated type (significant humidity-displacement anomaly), sliding-dominated type (significant displacement-pressure anomaly) or composite type (all three parameters are abnormal); Risk level assessment: Combined with the spatial correction weight of the dynamic threshold generation submodule, output local or systemic risk level.

[0106] In the aforementioned embodiment, the coupled feature extraction submodule of this embodiment achieves multi-parameter coupled analysis and anomaly verification of slope monitoring data through the collaborative operation of multiple units. The specific significance is as follows: Through the collaboration of the parameter association model construction unit and the association relationship logic confirmation unit, a systematic mechanical correlation framework for the three parameters of moisture, displacement, and pressure is established, transforming the previously isolated monitoring indicators into an analytical system with a geomechanical basis. A progressive strategy of "single-parameter relationship establishment → two-parameter coupled analysis → three-parameter coordinated verification" ensures that each correlation is verified by physical mechanisms, avoiding one-sided, data-driven conclusions. The anomaly coupling verification unit, through a three-level mechanism consisting of single-correlation verification, multi-correlation coordinated verification, and confidence assessment, upgrades traditional single-point anomaly detection to composite anomaly identification based on mechanical consistency, significantly reducing false alarm rates. The output anomaly type classification (seepage, sliding, and composite) and risk level assessment provide differentiated basis for response decisions, achieving an upgrade from "anomaly alarm" to "mechanistic diagnosis." It has enabled slope monitoring to develop from a single parameter threshold alarm to an intelligent diagnosis system based on a multi-parameter coupling mechanism, providing analytical results supported by mechanical theory for engineering early warning.

[0107] Example 6:

[0108] like Figure 3 As shown, based on Example 1, the data analysis subsystem provided by this embodiment of the present invention includes:

[0109] A logic startup module is used to receive abnormal data, start the corresponding analysis logic of humidity-displacement correlation, displacement-pressure correlation and humidity-pressure correlation, and filter abnormal data according to the corresponding input content;

[0110] Coupling analysis module, used for multi-analysis coupling of humidity, displacement, and pressure and risk quantification;

[0111] Humidity-displacement coupling, seepage deformation response analysis

[0112] Input: Abnormal soil moisture and displacement data; Analysis logic: Increased soil moisture softens the rock mass and reduces its shear strength. By comparing typical patterns of moisture changes and displacement responses (such as lag time and deformation rate) in historical data, it is determined whether the current moisture anomaly is likely to cause significant displacement. Output: If the moisture anomaly is accompanied by accelerated displacement, it is marked as a "seepage-deformation" coupling risk, and the impact is quantified.

[0113] The displacement-pressure coupled sliding stress transfer analysis uses inputs of abnormal displacement and pressure data (relying on the results of a previous moisture-displacement analysis). The analysis logic states that when soil slides, areas with increased displacement will experience a decrease in pressure due to soil loosening, while adjacent areas may experience an increase in pressure due to stress transfer. Combined with the moisture-displacement analysis results (e.g., whether the displacement is caused by seepage), the pressure change is verified to be consistent with the mechanical response caused by sliding. Output: If the displacement and pressure anomalies show spatial coordination (e.g., increased displacement at the toe of the slope and decreased pressure), the risk is flagged as "sliding-stress redistribution."

[0114] Analysis of the impact of humidity-pressure coupling on pore water pressure involves inputting abnormal humidity and pressure data (combining the results of the previous two analyses). The analysis logic eliminates displacement interference and directly analyzes the relationship between humidity and pressure. For example, if humidity suddenly increases but displacement remains unchanged, a simultaneous abnormal pressure fluctuation could indicate soil softening or liquefaction risk caused by a sudden increase in pore water pressure. The output independently identifies anomalies directly related to humidity and pressure, addressing risk scenarios not covered in the previous two phases.

[0115] The collaborative verification module is used for multi-parameter collaborative verification and risk classification. It checks whether the three types of parameter anomalies meet the typical combination pattern of slope instability; and classifies the risk level according to the number of abnormal parameters, coupling strength and spatial distribution range;

[0116] Local low risk: single parameter abnormality (such as only humidity increase); local high risk: dual parameter coupling abnormality (such as humidity-displacement synchronous abnormality); systemic risk: all three types of parameters are abnormal and consistent with the damage mechanism.

[0117] In the above-mentioned embodiment, the data analysis subsystem of this embodiment constructs a progressive, multi-parameter fusion slope stability analysis system through the collaborative operation of a logic activation module, a coupling analysis module, and a collaborative verification module. The logic activation module automatically triggers the corresponding correlation analysis logic based on the type of abnormal data (humidity, displacement, or pressure), ensuring that the data input meets the requirements of subsequent analysis, avoiding interference from invalid data, and improving the targetedness and efficiency of the analysis. Moisture-displacement coupling identifies the impact of seepage on slope deformation and determines whether displacement acceleration is caused by soil softening. Displacement-pressure coupling combines moisture-displacement analysis results to verify whether sliding triggers stress redistribution, improving the reliability of sliding risk identification. Moisture-pressure coupling independently analyzes the direct impact of pore water pressure, addressing softening or liquefaction risks that may be missed by the previous two correlations. Through these three types of coupled analysis, the main causes of slope instability (seepage, sliding, and pore water pressure anomalies) are systematically covered. The moisture, displacement, and pressure anomalies are checked to see if they meet the typical slope instability combination pattern (e.g., increased humidity → accelerated displacement → sudden pressure drop). Based on the number, coupling intensity and spatial distribution of abnormal parameters, local low risk (single parameter anomaly), local high risk (two parameter anomaly) and systemic risk (three parameter anomaly) are divided to make risk assessment more hierarchical and operational.

[0118] In summary, this embodiment establishes a logically coherent analysis process from abnormal data screening to coupled analysis, and then to collaborative verification and risk grading. The final output risk level (low / high / systemic) can directly support the formulation of reinforcement measures, achieving a seamless transition from monitoring to decision-making. Through a modular and progressive analysis strategy, multi-source sensor data is converted into risk indicators with a mechanical basis, and slope stability assessment is upgraded from a single-parameter alarm to a multi-parameter coupled intelligent diagnosis, thereby improving the accuracy of early warning and the scientific nature of decision-making.

[0119] Example 7:

[0120] like Figure 4 As shown, based on Example 1, the decision support subsystem provided by this embodiment of the present invention includes:

[0121] The risk feature analysis and classification module receives risk assessment results from the data analysis subsystem, including risk type labeling (seepage-dominated / sliding-dominated / combined), risk level classification (local low risk / local high risk / systemic risk), and key parameter anomaly characteristics (such as humidity surge amplitude, displacement acceleration, and pressure imbalance range). It automatically constructs a risk feature matrix. For seepage-dominated risks, the module focuses on the spatial correspondence between humidity anomaly areas and displacement responses. For sliding-dominated risks, the module focuses on analyzing the spatial distribution pattern of displacement-pressure coupling anomalies. For combined risks, the module comprehensively evaluates the coordinated anomaly characteristics of the three types of parameters.

[0122] Among them, the risk characteristic matrix R is expressed as:

[0123] ;

[0124] Where, Indicates the seepage-dominated risk marker (value 0 or 1); Indicates the sliding dominant risk flag (value 0 or 1); Indicates a composite risk flag (value 0 or 1); Indicates a local low risk level (quantitative value 0.1~0.3); Indicates a local high risk level (quantitative value 0.4~0.7); Indicates the level of systemic risk (quantitative value 0.8~1.0); Indicates the sudden increase in humidity (unit: percentage change in volume moisture content; Indicates displacement acceleration (unit: mm / h²); Represents the range of pressure imbalance (unit: m²); the risk characteristic matrix is ​​constructed based on multi-source data fusion theory and space-time coupling analysis model.

[0125] Through the tensor product operation ( ) Separate the independent influencing factors of risk type, level, and parameters to avoid decision ambiguity caused by feature coupling; normalize the raw sensor data (such as humidity, displacement, and pressure) into dimensionless risk level values ​​(L / H / S) and physical dimension parameters (ΔH / A / P) to achieve unified representation of multi-dimensional data; the combination of marker variables (α / β / γ) and quantitative values ​​can describe the risk evolution path (for example, when α=1 and ΔH>30%, the seepage risk upgrade is triggered). Specific functions: Structured decision input: Provide a standardized parameter interface for matching reinforcement measures; predict the critical point where the risk evolves from local to systemic through the time-series update of matrix elements; combine coverage density (spatial distribution), design strength (level quantification), and implementation priority (time-series characteristics) to generate a three-dimensional optimization solution; through orthogonal feature space and dynamic weight allocation mechanism, abstract complex slope risks into computable and interventionable mathematical objects, which is the key hub for intelligent analysis systems from data perception to engineering decision-making.

[0126] Targeted measures matching and optimization module, used to initiate graded response strategies based on the risk feature matrix;

[0127] (1) Seepage-dominated risk response:

[0128] Local drainage measures: Arrange a light drainage network in areas with significant humidity anomalies, using a radial drainage ditch design;

[0129] Deep water diversion scheme: When humidity anomalies are accompanied by deep displacement signals, add vertical drainage wells;

[0130] Anti-seepage reinforcement: Implement surface anti-seepage treatment in areas with persistent high humidity, combined with grass planting to stabilize the soil;

[0131] (2) Sliding-dominated risk response:

[0132] Local reinforcement: Arrange micro anti-slide pile arrays in displacement-pressure anomaly coupling areas;

[0133] Stress transfer design: according to the pressure redistribution characteristics, set up the guide retaining wall;

[0134] Slope foot protection: Step-by-step reinforcement is used for slope foot areas where displacement accumulation is significant;

[0135] (3) Response to complex risks:

[0136] Combined treatment: Drainage system and anti-slip structure are arranged in coordination;

[0137] Layered treatment: three-layer protection system of surface anti-seepage + middle layer water diversion + deep anchoring;

[0138] Dynamic adjustment: Set up an adjustable support structure to adapt to subsequent deformation development;

[0139] The spatial adaptation and strength calibration module is used to adjust the coverage density of engineering measures according to the spatial distribution density of abnormal parameters, determine the design strength level of the structure based on the quantitative value of the risk level, consider the temporal characteristics of parameter anomalies, and plan the priority of measure implementation.

[0140] In the above-mentioned embodiment, this embodiment achieves accurate identification and classification of slope risks through a risk feature analysis and classification module. Based on the risk assessment results provided by the data analysis subsystem (including risk type, level, and abnormal characteristics of key parameters), a multi-dimensional risk feature matrix is ​​constructed: for seepage-dominated risks, a moisture-displacement spatial correlation model is established; for sliding-dominated risks, a displacement-pressure coupling distribution map is constructed; for composite risks, a multi-parameter collaborative assessment framework is formed. Raw monitoring data is converted into characteristic indicators with engineering significance, providing structured input for decision-making. Based on the risk feature matrix, the targeted measure matching and optimization module automatically generates a corresponding engineering solution library for different risk types (seepage / sliding / combined); dynamically adjusts the intensity of measures based on the risk level (low / high / systemic); and optimizes the plan layout of engineering measures based on the spatial distribution characteristics of abnormal parameters. The output preliminary plan has spatial targeting and intensity differentiation characteristics. The spatial adaptation and intensity calibration module optimizes the preliminary plan in multiple dimensions, dynamically adjusting the coverage density of engineering measures based on the concentration of abnormal parameters. It also determines the design parameter thresholds of structures based on risk quantification and prioritizes the implementation of measures based on the evolution of parameter anomalies. It transforms complex monitoring data into actionable engineering features, establishes a mapping between risk features and engineering measures, and achieves precise adaptation across three dimensions: space, intensity, and time.

[0141] Example 8:

[0142] like Figure 5 As shown, based on Examples 1 to 7, the edge computing method for slope stability based on multi-sensor fusion provided by the embodiment of the present invention includes the following steps:

[0143] S100: Deploys micro-computing units at the edge nodes of sensor networks to achieve in-situ feature extraction, dynamic data compression, cross-modal alignment, and lightweight fusion;

[0144] In-situ feature extraction: Each sensor node calculates local eigenvalues ​​(such as humidity change rate, displacement acceleration, and pressure gradient) in real time. Dynamic data compression: A sliding window mechanism is used to transmit only eigenvectors that exceed a threshold. Cross-modal alignment: Time stamp synchronization and spatial coordinate mapping are used to achieve spatiotemporal alignment of multi-source data at the edge layer. Lightweight fusion: Confidence fusion is performed using an improved DS evidence theory to output preliminary anomaly indicators.

[0145] S200: Builds a three-level analysis architecture for rapid response at the edge layer, correlation analysis at the fog layer, and deep verification at the cloud layer;

[0146] Fast response at the edge layer: Deploy a micro-decision tree model to immediately trigger local warnings for obvious anomalies (such as a sudden increase of 30% in humidity and displacement exceeding the limit);

[0147] Fog layer correlation analysis: A lightweight LSTM network is run on regional gateway nodes to analyze the temporal correlation characteristics of multiple sensors;

[0148] In-depth cloud verification: Upload complex working conditions (such as multi-parameter coupling anomalies) to the cloud for mechanism model verification;

[0149] Dynamic weight adjustment: Automatically adjust the computing load of each layer according to the network status to ensure real-time requirements;

[0150] S300: Establish a dynamically updated edge intelligence model that includes cloud-based model slimming, incremental learning, cross-node knowledge sharing, and adaptive working conditions.

[0151] Cloud model slimming: Through channel pruning and quantization, complex models trained on the cloud are compressed into edge-deployable versions.

[0152] Incremental learning: New case data collected by edge nodes is encrypted and used for iterative model updates;

[0153] Cross-node knowledge sharing: Using a federated learning mechanism, multiple edge nodes collaboratively optimize local models;

[0154] Adaptive working condition: Automatically switch model parameter sets according to changes in the on-site environment (such as rainy season / dry season).

[0155] In the above-mentioned embodiment, the multi-sensor fusion slope stability edge computing method achieves full-chain optimization from data perception to decision response through the deep coupling of edge computing architecture and dynamic intelligent models. Its technical core lies in: in-situ feature extraction and dynamic compression based on micro-computing units (S100) significantly reduces redundant data transmission. Simultaneously, lightweight fusion based on an improved DS evidence theory ensures real-time alignment and confidence integration of multimodal data at the edge layer. Combined with the dynamic load balancing mechanism (S200) of the three-level analysis architecture, the edge-layer micro-decision tree can respond to sudden deformations within 50ms. The fog-layer LSTM network reduces the false alarm rate through temporal correlation analysis, and the cloud-based mechanism model improves the verification accuracy of multi-parameter coupling conditions. Furthermore, through the dynamically updated edge intelligent model system (S300), the incremental update driven by federated learning shortens the model's response time during flood / drought season switching, and the slimmed-down model, which undergoes channel pruning, reduces inference energy consumption on the RK3399 edge device. The technical effect of the synergy of the three is reflected in the following: in a 2km² slope monitoring scenario, the overall energy consumption of the system is reduced by 76%, the abnormality detection delay is controlled within 200ms, and the prediction accuracy of complex landslide events is improved by 41.8% compared with traditional methods.

[0156] Figure 6A block diagram is shown of an exemplary electronic device suitable for implementing embodiments of the present invention.

[0157] The electronic device may include a central processing unit / microprocessor / main control chip, etc.; a storage medium, coupled to the central processing unit / microprocessor / main control chip, etc., and storing computer-executable instructions therein, for performing the steps of each method of an embodiment of the present invention when executed by the processor.

[0158] The central processing unit / microprocessor / main control chip etc. may include but is not limited to, for example, one or more processors or microprocessors etc.

[0159] The storage medium may include, but is not limited to, for example, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, computer storage media (such as hard disk, floppy disk, solid-state drive, removable disk, CD-ROM, DVD-ROM, Blu-ray disc, etc.).

[0160] In addition, the electronic device may also include (but not limited to) a data bus, an input / output bus / external bus / device bus, a display, and input / output devices (eg, keyboard, mouse, speaker, etc.).

[0161] The central processing unit / microprocessor / main control chip etc. can communicate with external devices via an I / O bus via a wired or wireless network (not shown).

[0162] The storage medium may also store at least one computer-executable instruction for executing the various functions and / or method steps in the embodiments described in this technology when executed by a central processing unit / microprocessor / main control chip, etc.

[0163] In one embodiment, the at least one computer executable instruction may also be compiled into or constitute a software product, wherein one or more computer executable instructions are executed by a processor to perform the various functions and / or method steps in the embodiments described in the present technology.

[0164] Figure 7 A schematic diagram of a computer-readable storage medium according to an embodiment of the present invention is shown.

[0165] like Figure 7As shown, a non-transitory computer-readable storage medium stores instructions, such as computer-readable instructions. When the computer-readable instructions are executed by a processor, the various methods described above can be executed. Non-transitory computer-readable storage media include, but are not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-transitory non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. For example, the non-transitory computer-readable storage medium can be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions stored on the computer-readable storage medium, the various methods described above can be performed.

[0166] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0167] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0168] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0169] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the various embodiments of the method of the present invention via a computer device (which can be a personal computer, server, or network device, etc.). The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0170] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A multi-sensor fusion slope stability intelligent analysis system, characterized by: Include: The environmental perception subsystem uses soil moisture sensors to monitor changes in soil moisture content, displacement sensors to obtain slope movement data, and pressure sensors to sense changes in ground pressure. It also pre-screens abnormal data to form a fused data set. According to the physical coupling law of moisture-displacement-pressure, dynamic thresholds are set; abnormal soil moisture content, micro-movement data and ground pressure data in the slope data detection network are obtained according to the preset fluctuation trend standard and dynamic threshold; the dynamic threshold generation submodule is used to construct adaptive discrimination standards of baseline threshold, trend correction and spatial correction using a sliding time window mechanism; a three-level discrimination mechanism of single point anomaly, local anomaly and system anomaly is implemented; the confirmed abnormal data is structured by adding spatiotemporal tags, additional confidence and recording the evolution process; among them, the baseline threshold: the initial discrimination standard is established based on historical stable period data; trend correction: the threshold sensitivity is adjusted according to the recent data change rate; spatial correction: the risk weight is differentiated according to the location of the sensor; single point anomaly: the data of a single sensor exceeds the dynamic threshold range; local anomaly: the adjacent sensor group has a coordinated anomaly; system anomaly: the three types of parameters show an abnormal combination that conforms to the damage mechanism; The data analysis subsystem is used to convert abnormal data in the fused dataset into quantitative indicators of slope stability. By analyzing the correlation between soil moisture content, movement data and ground pressure, the potential sliding risk of the slope can be inferred. The coupling feature extraction submodule is used to establish a correlation analysis framework for three types of parameters, namely moisture-displacement correlation, displacement-pressure correlation and moisture-pressure correlation, based on the physical properties of the rock and soil of the slope to be tested. Among them, moisture-displacement correlation: analyzes the soil deformation response caused by seepage; displacement-pressure correlation: monitors the stress redistribution caused by soil sliding; moisture-pressure correlation: evaluates the influence of pore water pressure on effective stress. Through the three types of coupling analysis, the main causes of slope instability are systematically covered. It is checked whether the anomalies of moisture, displacement and pressure meet the typical slope instability combination pattern. Based on the number, coupling strength and spatial distribution of abnormal parameters, local low risk, local high risk and systemic risk are divided. The main causes include seepage, sliding and pore water pressure anomalies. The slope instability combination pattern includes humidity increase → displacement acceleration → pressure drop. Local low risk is not a single parameter anomaly, local high risk is not a double parameter anomaly, and systemic risk is not a three parameter anomaly. The decision support subsystem is used to propose specific reinforcement measures based on the analysis results of potential sliding risks.

2. The multi-sensor fusion slope stability intelligent analysis system according to claim 1 is characterized in that: Environmental perception subsystem, including: The multi-dimensional perception module is used to obtain a three-dimensional model of the slope to be detected and determine the layout coordinates of soil moisture sensors, displacement sensors, and pressure sensors based on historical slope landslide risks. This forms a slope data detection network composed of soil moisture sensors, displacement sensors, and pressure sensors. The anomaly pre-screening module is used to initialize the slope data detection network and compare the fluctuation trend of slope data within the same time window. It sets a dynamic threshold based on the physical coupling law of moisture-displacement-pressure. It then obtains abnormal soil moisture content, micro-displacement data, and ground pressure data in the slope data detection network based on the preset fluctuation trend standard and dynamic threshold. The data fusion module is used to receive abnormal soil moisture content, micro-movement data and ground pressure data, and use the layout coordinates of soil moisture sensors, displacement sensors and pressure sensors as tags to form a fused data set containing multi-dimensional associations of tags.

3. The multi-sensor fusion slope stability intelligent analysis system according to claim 2, characterized in that: Multi-dimensional perception module, including: The 3D model construction submodule is used to obtain the surface geometric data of the slope to be tested. Through point cloud processing and surface reconstruction, a 3D model is generated, which includes the elevation, slope, and curvature geometric characteristics of the terrain. It also uses historical landslide data including the location of the slip surface and the extent of the damage to demarcate potential risk areas, forming a digital slope model with geological risk markers. The risk weight calculation submodule is used to calculate the risk weights of different slope areas to be tested based on the digital slope model, taking into account the location of the slip surface and the spatial distribution of the damage range, the damage depth and the triggering factors; The sensor coordinate optimization submodule is used to determine the mechanical coupling relationship between soil moisture, displacement, and pressure, and set the corresponding sensor layout conditions; Through iterative optimization, the spatial matching coordinates of the three types of sensors are determined.

4. The multi-sensor fusion slope stability intelligent analysis system according to claim 2, characterized in that: Abnormal pre-screening module, including: The multi-source data spatiotemporal alignment submodule is used to unify the spatiotemporal benchmarks of three types of data: soil moisture, displacement, and ground pressure, based on the constructed slope data monitoring network; The coupled feature extraction submodule is used to establish a correlation analysis framework for three types of parameters: moisture-displacement correlation, displacement-pressure correlation, and moisture-pressure correlation, based on the physical properties of the rock and soil mass of the slope to be tested; The dynamic threshold generation submodule uses a sliding time window mechanism to construct adaptive discrimination criteria for baseline thresholds, trend corrections, and spatial corrections; implements a three-level discrimination mechanism for single-point anomalies, local anomalies, and system anomalies; and performs structured processing on confirmed anomaly data by adding spatiotemporal tags, adding confidence levels, and recording the evolution process.

5. The multi-sensor fusion slope stability intelligent analysis system according to claim 4, characterized in that: Abnormal pre-screening module, including: The time dimension of the multi-source data spatiotemporal alignment submodule uses the data acquisition timestamp as the benchmark to establish a millisecond-level synchronization mechanism; the spatial dimension uses the sensor layout coordinate information to construct a three-dimensional spatial reference system; the data dimension converts raw data of different dimensions into standard monitoring indicators; The moisture-displacement correlation of the coupled feature extraction submodule is used to analyze the soil deformation response caused by seepage; Displacement-pressure correlation monitoring of stress redistribution caused by soil sliding; Moisture-pressure correlation to evaluate the effect of pore water pressure on effective stress; The baseline threshold of the dynamic threshold generation submodule establishes the initial discrimination criteria based on historical stable period data; Trend correction adjusts the threshold sensitivity based on the rate of change of recent data; Spatial correction is set differently based on the risk weight of the sensor location; single-point anomaly means that the data of a single sensor exceeds the dynamic threshold range; local anomaly means that a coordinated anomaly occurs in a group of adjacent sensors; system anomaly: the three types of parameters present an abnormal combination that is consistent with the damage mechanism.

6. The multi-sensor fusion slope stability intelligent analysis system according to claim 5, characterized in that: Coupled feature extraction submodule, including: The parameter association model construction unit is used to gradually establish the association framework of three types of parameters based on the physical characteristics of the slope rock and soil. The coupled feature extraction submodule adopts a progressive analysis strategy; A correlation logic confirmation unit is used to set the analysis logic of humidity-displacement correlation, displacement-pressure correlation and humidity-pressure correlation; The anomaly coupling verification unit is used to cross-validate the results of the three types of correlation analysis; if the anomaly signal is valid in multiple correlation models, it is marked as a high-confidence anomaly.

7. The multi-sensor fusion slope stability intelligent analysis system according to claim 1, characterized in that: Data analysis subsystem, including: A logic startup module is used to receive abnormal data, start the corresponding analysis logic of humidity-displacement correlation, displacement-pressure correlation and humidity-pressure correlation, and filter abnormal data according to the corresponding input content; Coupling analysis module, used for multi-analysis coupling of humidity, displacement, and pressure and risk quantification; The collaborative verification module is used for multi-parameter collaborative verification and risk classification, checking whether the three types of parameter anomalies meet the typical combination pattern of slope instability; and dividing the risk level according to the number of abnormal parameters, coupling strength and spatial distribution range.

8. The multi-sensor fusion slope stability intelligent analysis system according to claim 7, characterized in that: Local low risk of the collaborative verification module: single parameter abnormality; local high risk dual parameter coupling abnormality; systemic risk: all three types of parameters are abnormal and consistent with the damage mechanism.

9. The multi-sensor fusion slope stability intelligent analysis system according to claim 1, characterized in that: Decision support subsystem, including: The risk feature analysis and classification module is used to receive risk assessment results from the data analysis subsystem, including risk type labeling, risk level classification, and abnormal characteristics of key parameters, and automatically construct a risk feature matrix; Targeted measures matching and optimization module, used to initiate graded response strategies based on the risk feature matrix; The spatial adaptation and strength calibration module is used to adjust the coverage density of engineering measures according to the spatial distribution density of abnormal parameters, determine the design strength level of the structure based on the quantitative value of the risk level, consider the temporal characteristics of parameter anomalies, and plan the priority of measure implementation.

10. A multi-sensor fusion slope stability edge computing method of the multi-sensor fusion slope stability intelligent analysis system according to any one of claims 1 to 9, characterized in that: The multi-sensor fusion slope stability edge computing method includes the following steps: The edge computing method of slope stability based on multi-sensor fusion includes the following steps: Deploy micro-computing units at the edge nodes of sensor networks to achieve in-situ feature extraction, dynamic data compression, cross-modal alignment, and lightweight fusion; Build a three-level analysis architecture with rapid response at the edge layer, correlation analysis at the fog layer, and deep verification at the cloud layer; Establish a dynamically updated edge intelligence model that includes cloud model slimming, incremental learning, cross-node knowledge sharing, and working condition adaptation.

Images