A slope collapse real-time evaluation method and system based on state consistency damage degree

Abstract

This invention discloses a real-time assessment method and system for slope collapse based on state consistency failure, belonging to the field of disaster prevention, mitigation, and safety assessment technology. It addresses the lack of effective characterization in existing technologies regarding whether various state elements within a slope still maintain reasonable coordination and consistency. When the slope is in a stable operating phase, this invention establishes, verifies, and fixes a state consistency relationship model to obtain a stable-period state consistency relationship model, forming a stable-period benchmark. During slope operation, based on the stable-period state consistency relationship model, it calculates the coordination deviation and state consistency failure index among state variables obtained from real-time multi-source monitoring data. Combining the time dimension, it analyzes the continuous changes of the state consistency failure index, extracting the trend, persistence, and stage characteristics of the failure. The above results are used for real-time slope classification and risk assessment. This invention is used for real-time slope collapse assessment.

Description

Technical Field

[0001] A real-time assessment method and system for slope collapse based on state consistency failure degree is proposed, which is used for real-time assessment of slope collapse and belongs to the field of disaster prevention, mitigation and safety assessment technology. Background Technology

[0002] As a crucial engineering structure in highways, railways, water conservancy, mining, and urban engineering projects, slopes are constantly subjected to the combined effects of the natural environment and engineering disturbances. They are susceptible to various factors such as rainfall seepage, earthquakes, construction disturbances, and material degradation, exhibiting significant time-varying and uncertainties. Once a slope collapses, it not only causes traffic disruptions and damage to engineering facilities but can also trigger serious casualties and secondary disasters. Therefore, timely and reliable assessment of slope collapse risks is a vital technical requirement for ensuring the safe operation of major projects and public safety.

[0003] Current slope collapse assessment and monitoring technologies mainly rely on the analysis of single or multi-source monitoring information such as displacement, deformation, dip angle, cracks, and rainfall, and judge slope stability by setting thresholds or empirical rules. However, as a typical complex geological body, the internal structure and mechanical state of a slope exhibit significant spatial heterogeneity and evolutionary characteristics. Single monitoring quantities or simply superimposed multi-source indicators often fail to fully reflect the true structural changes of the slope. When a slope evolves from a stable state to an unstable state, the instability process is usually not directly triggered by a sudden change in a single monitoring quantity, but rather accompanied by the gradual weakening or even decoupling of the synergistic relationship between various internal state elements.

[0004] In recent years, with the enrichment of monitoring methods and the improvement of data acquisition capabilities, multi-source monitoring and data fusion methods have been applied in slope safety assessment. However, existing methods mostly focus on weighting, scoring, or classifying the monitoring data itself, and the assessment results still mainly rely on the absolute change range or statistical anomaly characteristics of the monitored quantities. While these methods improve information utilization to some extent, they still struggle to accurately depict the continuous evolution of the slope structure from "self-consistent" to "instability" when facing complex working conditions and sudden disturbances. The assessment results are highly dependent on monitoring conditions and empirical parameters. Furthermore, some model-based or data-driven slope assessment methods often require a large amount of historical samples or prior information under specific working conditions, limiting their applicability and transferability. Simultaneously, these methods often focus on predicting whether collapse events will occur, lacking a real-time depiction of the rationality and stability of the current slope structure, making it difficult to provide continuous and interpretable decision-making basis for engineering management.

[0005] In summary, the existing technology has the following technical problems: 1. It focuses on monitoring and judging the "state values" of the slope, but lacks effective characterization of whether the various state elements inside the slope still maintain reasonable coordination and consistency. 2. When a slope is in a stable state, its surface response, deep deformation, structural integrity, and external disturbance response usually exhibit a relatively stable correlation. When the slope gradually approaches an unstable state, the inherent consistency between these states will be disrupted first, and will exhibit perceptible and quantifiable evolutionary characteristics before the collapse occurs. 3. Analyzing multi-source monitoring data as independent information lacks a mechanism for unified modeling and comprehensive identification of multi-state information, making it difficult to reflect the true operating status of the slope structure from the overall system level; 4. In slope monitoring and assessment methods, there is a lack of technical means to solidify the operational characteristics during the stable period as a reference benchmark and to verify the consistency of changes in the operational state based on this benchmark. This results in frequent changes in assessment results as monitoring data fluctuates, leading to insufficient stability and interpretability. 5. Existing slope stability assessment methods generally rely on setting thresholds for single or a few monitoring indicators. The selection of thresholds is highly empirical and regionally dependent, making it difficult to adapt to different slope structural conditions and state changes during long-term operation, which can easily lead to misjudgment or omission. Summary of the Invention

[0006] The purpose of this invention is to provide a real-time assessment method and system for slope collapse based on the degree of state consistency failure, which solves the problem that the existing technology focuses on monitoring and judging the "state value" of the slope, but lacks an effective characterization of whether the various state elements inside the slope still maintain reasonable coordination and consistency.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A real-time assessment method for slope collapse based on state consistency failure degree includes the following steps: Step 1: When the slope is in a stable operation phase, the slope multi-state variable set is constructed by multi-source monitoring data, and time alignment and scale normalization are performed. The coordinated change law between different state variables is identified by statistical analysis methods, a state consistency relationship model is established and verified and fixed to obtain the stable period state consistency relationship model, forming a stable period benchmark. Step 2: During the slope operation, based on the stable period state consistency relationship model, calculate the coordinated deviation degree between the state variables obtained from the real-time multi-source monitoring data, and construct a state consistency destruction index for the coordinated deviation degree between the slope operation period state and the stable period benchmark. Step 3: Analyze the continuous changes in the state consistency violation index in conjunction with the time dimension, and extract the trend, persistence and stage characteristics of the violation. Step 4: Based on the results obtained in Step 3, perform real-time grading and risk assessment of the slope.

[0008] Furthermore, the specific steps of step 1 are as follows: Step 1.1: When the slope is in a stable operating phase, the acquired multi-source monitoring data will be used to construct a multi-state variable set for the slope. Among them, the multi-source monitoring data includes deformation state information reflecting the geometric deformation characteristics of soil or rock, stress and strain information reflecting the stress state of slope, seepage state information characterizing the internal hydrodynamic characteristics of slope, external environmental information describing external disturbances or environmental driving factors, and structural integrity information reflecting the degree of damage to slope materials. Deformation state information includes surface displacement, deep displacement and dip angle changes; stress and strain information includes strain values, stress distribution or stress changes measured by strain gauges of soil or support structure; seepage state information includes pore water pressure, groundwater level and seepage rate; external environmental information includes rainfall, temperature and ground motion acceleration; and structural integrity information includes crack width, acoustic emission energy and microseismic count. Multi-state variable set of slope Represented as: in, express The first moment One state variable, This represents the total number of state variables. Step 1.2: Use linear interpolation or moving average to align the time of each state variable in the multi-state variable set of the slope, so as to unify the multi-source monitoring data to the same time step; Step 1.3: Normalize the time-aligned state variables to obtain the stable multi-state data sequence. The normalization formula is: in, To The normalized state variables, which represent the state data during the steady-state period, have a range of values ​​limited to [a certain value range]. Inside, Indicates the first One state variable, and State variables The maximum and minimum values; Step 1.4: Based on the multi-state data sequence during the stable period, statistical analysis methods are used to identify significant synergistic relationships between different state variables during the stable operation phase. The statistical analysis methods include calculating the Pearson correlation coefficient, mutual information analysis, and canonical correlation analysis. The specific steps for identifying significant cooperative relationships between different state variables during the stable operation phase are as follows: During the stable operation phase, the time series is divided into multiple sub-time periods, and the correlation coefficient between any two state variables within the same sub-time period is calculated. The specific determination method is as follows: If the Pearson correlation coefficient of a pair of state variables is greater than a given threshold and the mutual information obtained from mutual information analysis is greater than a given threshold, then the pair of state variables is determined to be a significantly co-linear pair of state variables. If the Pearson correlation coefficient of a pair of state variables is less than or equal to a given threshold, and the mutual information obtained from mutual information analysis is greater than the given threshold, then the pair of state variables is determined to be a nonlinear cooperative state variable pair. If the Pearson correlation coefficient of a pair of state variables is less than or equal to a given threshold, and the mutual information obtained from mutual information analysis is less than or equal to a given threshold, then the pair of state variables is determined to have no stable relationship and is included in the identification of the state variable group. If the Pearson correlation coefficient of a pair of state variables is greater than a given threshold, and the mutual information obtained from mutual information analysis is less than or equal to the given threshold, then the pair of state variables is determined to be a random co-state pair and included in the identification of the state variable group. The state variables included in the identification of the state variable group are formed into multiple sets of state variable sequences of different types or different physical meanings. Canonical correlation analysis is performed on any two sets of state variable sequences to obtain the maximum canonical correlation coefficient. If the maximum canonical correlation coefficient is greater than a given threshold, it is determined that there is a stable multivariate synergistic relationship between the two sets of state variable sequences. Based on this, in the canonical correlation analysis results, the canonical loading values ​​of any two state variables are calculated. If the canonical loading values ​​of the two state variables are greater than a preset threshold, it is determined that the state variable makes a significant contribution to the multivariate synergistic relationship and is included in the state variable group. Otherwise, the state variables already in the state variable group are retained, and the state variables not in the state variable group are not included. During the stable operation phase, the state relationships of each pair of state variables and each group of state variables that satisfy the corresponding conditions are determined across multiple sub-time periods, which indicates that the pair of state variables or each group of state variables has a significant synergistic relationship. Among them, "across multiple sub-time periods" means that when the total number of sub-time periods in the stable period exceeds 5, it is more than 80% of the total number of sub-time periods; when the number of sub-time periods in the stable period is less than or equal to 5, it is no less than 3 consecutive sub-time periods. The stable period refers to the stable operation phase. Step 1.5: Establish a state consistency relationship model describing the coupling constraints between state variables during the stable operation phase. Establish a stable state consistency relationship for state variable pairs and state variable groups with significant synergistic relationships. The state consistency relationship model includes a linear synergistic model applicable to significantly synergistic linear state variable pairs, a nonlinear synergistic model applicable to nonlinear synergistic state variable pairs, and a multiple regression model applicable to state variable groups. Step 1.6: After extracting the state consistency relationship during the stable period, the stability of the state variable pairs or groups of state variables in the stable period state consistency relationship is verified and the model is solidified based on statistical characteristics through time segment test and residual analysis. When the time segment test and residual analysis simultaneously satisfy the model solidification and stability verification, the stable period state consistency relationship model is obtained. The statistical characteristics include mean and standard deviation. After completing the stability verification and model solidification of the state consistency relationship model in Step 1.7 and Step 1.6, based on the stable state consistency relationship model and its corresponding model parameters, calculate the residual between the predicted and actual values ​​of the stable state variables, and perform statistical analysis on the residual to obtain statistical analysis characteristics as the stable period benchmark. The statistical analysis characteristics include the mean and standard deviation.

[0009] Furthermore, in step 1.4: Pearson correlation coefficient The formula is: in, and They represent the first time. Individual time period No. The and the first State variables and The normalized state variables at all times in the process. and It is the first and the The state variables in the sub-time period Mean of state variables after normalization, Pearson correlation coefficient The range of values ​​is , Indicates a perfect positive correlation. Indicates a completely negative correlation. This indicates that there is no linear correlation; The formula for obtaining mutual information through mutual information analysis is: in, Indicates the first Individual time period No. The and the first State variables and The probability density function of the joint distribution. and They are and marginal distribution; Canonical correlation analysis refers to setting a sequence of state variables... and , and These represent state variables of different types or with different physical meanings in the first... Individual time segment The canonical correlation analysis will look for a new pair of variables in the sequence of state variables formed above. and And the maximum canonical correlation coefficient is obtained by solving the formula for maximizing the canonical correlation coefficient; The formula for maximizing the canonical correlation coefficient is: in, To maximize the canonical correlation coefficient, For variables and The covariance matrix between them and Representing variables respectively and The cross covariance matrix, and Given the weight vector to be solved, obtained through the generalized eigenvalue method, the objective is to maximize... and The correlation coefficient between them.

[0010] Furthermore, in step 1.5: The linear cooperative model is as follows: in, and They represent the first time. Individual time period No. The and the first State variables and The mean of the normalized state variables at all times in the time series, and and It has a significant synergistic relationship. For linear coordination in sub-time periods No. State variables The predicted value of the normalized mean is affected by the state variable. The target state affected Represents the coupling coefficient, reflecting the state variables. For state variables Intensity of influence Represents state variables With state variables The offset item; The nonlinear cooperative model is as follows: in, This indicates that in the nonlinear cooperative model, during the sub-time period... No. State variables The predicted value of the normalized mean. For random disturbance terms, The mean of the state variable and The mapping function for the nonlinear consistency relationship between state variables can be a polynomial mapping, basis function expansion, kernel function mapping or other continuous nonlinear function forms, used to characterize the nonlinear cooperative relationship between state variables during the steady period. The multiple regression model is as follows: in, This indicates that in multiple regression, within a sub-time period No. State variables The predicted value of the normalized mean. Indicates the first The mean of the state variables The mean of the state variable The linear contribution coefficient, i.e., the degree of its influence, This indicates the number of state variables participating in collaborative modeling. This represents the system baseline term in a multivariate linear relationship, used to describe the theoretical baseline level of the target state when all input states are zero.

[0011] Furthermore, step 1.6 specifically includes: Time-segment testing involves dividing the set of multi-state variables during the stable period into multiple sub-time periods. For each sub-time period, the model parameters of the state consistency relationship model are calculated, including the coupling coefficients in the linear cooperative model. and offset terms Linear contribution coefficient in a multiple regression model and system benchmark items and the random perturbation term in the nonlinear cooperative model After calculating the above parameters in each sub-time period, the stability was verified using the following method: During the stable period, a parameter sample sequence is formed based on the model parameters calculated from all sub-time periods: ,in, Indicates the first The model parameter sample sequence consists of all model parameters calculated in each sub-time period. The average value of the same model parameter is calculated across all sub-time periods. And the degree of dispersion, where the degree of dispersion is the standard deviation. or coefficient of variation ; Based on average Calculate the relative change ratio of each model parameter within each sub-time period: in, Indicates the first The first time period calculated within each sub-time period Model parameters, The first time interval calculated based on all sub-time intervals The mean of each model parameter is used to determine the difference between the relative change ratio and a pre-set threshold value for the parameter mean. If a comparison is made, If the mean of the model parameters remains consistent within the given sub-period, and this condition is met for more than 80% of the sub-periods, then the mean of the model parameters is considered stable during the stable period. The value is determined based on the upper quantile of the relative change ratio of the model parameters during the stable period or is given manually. Among them, the standard deviation of each model parameter in the parameter sample sequence or coefficient of variation Perform statistical analysis and compare it with the corresponding dispersion threshold. or In comparison, if or If the fluctuation range of the model parameters during the stable period is considered to be controlled, then the dispersion threshold is set. and Determined based on the upper quantile value of the model parameter distribution during the stable period or given manually; If the model parameters of each model are stable at the mean level and the fluctuation range is controlled, then the model is considered to remain stable during the stable period. When a model parameter is determined to be unstable, the following procedure is followed: First, perform model-level processing: If the unstable model parameters continue to show significant drift within a given sub-time period, it indicates that the state variable pair does not have long-term consistency constraints during the stable period. The state variable sequence corresponding to the state variable pair or state variable group should be removed from the stable period state consistency relation set. After model-level processing, data-level processing is performed: If unstable model parameters are concentrated in a specific sub-time period, it indicates that the stable period has not been sufficiently identified. The time segment test should be performed again. If the statistical instability is caused by insufficient number of model parameters, adjacent time segments should be merged and the time segment test should be performed again. After data processing, model solidification processing is performed: Only model parameters that pass the stability test are solidified. Model parameters that fail the stability test are marked as unstable consistency models and isolated for storage, and are not included in the subsequent calculation of state consistency violation index. At the same time, they are continuously monitored as candidate consistency relationships. When they meet the stability judgment conditions again with the addition of new data or during the extended stability period, the stability verification process can be retried. If the model fails the stability judgment in multiple verifications, it is completely removed and no longer participates in subsequent processing. Residual analysis refers to the process of modeling state consistency relationships. Based on historical multi-source monitoring data collected during the stable operation phase, a state consistency relationship model is used to invert and predict the target state variables, obtaining predicted values. These predicted values ​​are then compared with the actual observed values ​​to calculate the difference between the predicted and actual values, forming a residual sequence. The residuals are calculated for the state variables inverted from each model in the state consistency relationship model. The formula is as follows: Among them, the mean of the state variables That is, the actual observed value. , and For the corresponding residuals; during the stable period, statistical analysis is performed on the corresponding residual sequence. When the total number of sub-time periods exceeds 5, more than 80% of the sub-time periods, when the number of sub-time periods is less than or equal to 5, and when there are no less than 3 consecutive sub-time periods, the residual mean is close to zero, the residual variance is within the stable range, and the residuals are randomly distributed without obvious trends or periodic characteristics, then the state consistency relationship model is considered to have good fitting stability and robustness during the stable period, that is, the stable period state consistency relationship model is finally obtained. When the stability and robustness of the state consistency relationship model are poor, the model parameters are re-optimized, data noise and outliers are processed, or thresholds and biases are corrected to improve the stability and robustness of the state consistency relationship model.

[0012] Furthermore, the specific steps of step 2 are as follows: Step 2.1: During slope operation, based on the stable-period state consistency relationship model, examine the coordination relationship between each group and each pair of state variables in the multi-source monitoring data collected in real time during slope operation, and calculate their coordination deviation. When the When the state consistency relationship is a significantly cooperative linear pair of state variables, its calculation formula is: in, The predicted values ​​obtained from the linear cooperation model in the stable-period state consistency relationship model in the sub-time period Residuals during operation and These are the mean and standard deviation of the stable-period baseline for significantly co-linear state variable pairs, respectively. To prevent small constants with a denominator of zero, the degree of coordination deviation is... Reflects the first The degree of coordination change in the relationship between state variables, i.e., the degree of deviation between the current relationship between state variables and the baseline during the stable period, is called the degree of coordination deviation. The larger the value, the higher the degree of damage to the linear state variable, indicating that the slope stability is threatened; When the When the state consistency relationship is a nonlinear pair of state variables, the calculation steps are as follows: in, The predicted values ​​obtained from the nonlinear cooperative model in the stable-period state consistency relationship model in the sub-time period Residuals during operation and These represent the mean and standard deviation of the nonlinear state variable pair during the stable period, respectively, in the benchmark during the stable period. To prevent the use of tiny constants with a denominator of zero; When the When the state consistency relation is a group of state variables, the calculation formula is: The state variable set is mapped to a collaborative relation quantity for consistency testing, and the predicted values ​​for the runtime period are obtained using the stable-period state consistency relation model. Sub-time periods are then calculated. Operating residuals Based on the statistical stability period residuals, the residuals are standardized using the stability period benchmark, and the cooperative deviation of the state variable group is defined as: in, The predicted values ​​obtained from the multiple regression model in the stable-period state consistency relationship model in the sub-time period Residuals during operation and These represent the mean and standard deviation of the stable-period baseline for the stable-period state variable group, respectively. To prevent tiny constants with a denominator of zero; Step 2.2: Comprehensively analyze the cooperative deviation of each pair or group of state variables to obtain the state consistency destruction index used to quantify the cooperative relationship of slope structure. The calculation formula is: in, For the first The nth linear state variable pair, the nth The nth nonlinear state variable pair and the nth The weights of the relationships between state variables. The state consistency violation index is the total number of state relations. This reflects the overall degree of damage to the multi-state collaborative relationship of the slope. A lower value indicates a stronger cooperative relationship between states, suggesting the slope is in a stable state. A higher value indicates that the state coordination relationship has been disrupted, and the slope may be in an unstable stage.

[0013] Furthermore, the specific steps of step 3 are as follows: Step 3.1: Based on the state consistency violation index, form a violation time series in chronological order; Step 3.2: The time series of damage is segmented using a sliding window method, specifically by setting the window length. and sliding step size The time series of damage severity is divided into multiple overlapping time windows in chronological order. Each time window corresponds to a continuous sequence of destructiveness, which includes multiple state consistency destructiveness indicators. Step 3.3, within each time window The dynamic change characteristics of the destruction degree sequence are calculated. Specifically, the rate of change of the destruction degree is obtained by fitting or differentiating the trend of the state consistency destruction degree index with time within the time window. At the same time, the volatility index of the state consistency destruction degree index within the time window is calculated, including the standard deviation and amplitude range. The rate of change is used to reflect the upward or downward trend of the destruction degree within the time period, and the volatility index characterizes the intensity of the fluctuation of the destruction degree within the time period. The formula for the rate of change is: in, For the rate of change, For the time increment of change, express State consistency violation index; Step 3.4: Based on the rate of change and volatility indices calculated within the time window, comprehensively analyze the evolution characteristics of the state consistency violation index, and uniformly extract its dynamic evolution characteristics, including trend characteristics, persistence characteristics, and stage characteristics. The time window refers to a sub-time period, specifically: Trend characteristics: Based on the rate of change within multiple time windows, it is determined whether the rate of change is positive in more than three consecutive sub-time periods. If so, it indicates that the slope state consistency failure index is on a continuous upward trend, reflecting the continuous deterioration of the slope structure coordination relationship. If the rate of change is zero or negative in the above sub-time periods, the state consistency failure index is not on a continuous upward trend. Persistent characteristics: To determine whether the state consistency violation index is satisfied at all times within each time window. If so, then the degree of disruption within the corresponding time window is in a high-level range, and the volatility index of the state consistency disruption index is in a relatively stable range, indicating that the change in disruption degree has a significant and continuous characteristic; otherwise, it is not. The average value of the index representing the degree of disruption to the consistency of the state during the stable period. The standard deviation of the index representing the degree of disruption of state consistency during the stable period These are statistical coefficients; Stage-specific characteristics: Identifying significant temporal changes, i.e., inflection points or abrupt changes in the evolution of the state consistency failure index, serves as a stage-specific characteristic of the slope's transition from a stable to an unstable phase. The formula for the evolution of the state consistency failure index is as follows: in, for Indicators of state consistency violation within a time window. For time window The state consistency violation index, if Exceeding the set threshold indicates that the stability of the slope is seriously threatened.

[0014] Furthermore, the specific steps of step 4 are as follows: Step 4.1: Using the state consistency destruction index and the dynamic evolution characteristics obtained in Step 3.4 as the core criteria, the slope structure state is quantitatively graded and risk warning is achieved through "static threshold division and dynamic trend verification". Specifically: Static grading: During slope operation, the statistical characteristics of the state consistency failure index during the stable period are used as a benchmark to determine the range of thresholds. The range of the thresholds at which the state consistency failure index falls is then assessed. This indicates that the slope is in a stable state. This indicates that the slope is in a state of potential instability. This indicates the slope's height and instability state, and its statistical characteristics include the mean value of the state consistency failure index during the stable period. and standard deviation ; Dynamic verification: based on static hierarchical structure In each case, trend characteristics, persistence characteristics, and stage characteristics are assessed. If any one of these assessments indicates a potential instability state, it means the slope is in a highly unstable state. Specifically: Determine whether the trend is a continuous upward trend. If so, the slope is entering a state of instability; otherwise, it is not. Determine whether the persistent characteristic is in a high-level range. If so, then conduct a secondary confirmation of potential instability; otherwise, it is not. Determine whether the stage-specific characteristics represent a sudden increase. If so, it indicates that the stability of the slope is seriously threatened; otherwise, it is not. Step 4.2: Based on the final classification results of Step 4.1, the slope is classified as low risk if it is in a stable state, medium risk if it is in a potentially unstable state, and high risk if it is in a highly unstable state.

[0015] A real-time slope collapse assessment system based on state consistency failure degree includes a memory, a processor, and a computer program stored in the memory, characterized in that: the processor executes the computer program to implement the steps of the real-time slope collapse assessment method based on state consistency failure degree.

[0016] Compared with the prior art, the advantages of the present invention are as follows: I. This invention addresses the real-time assessment of slope collapse risk. It tackles the common problems in existing technologies, such as reliance on single monitoring quantities and fixed threshold judgments, difficulty in characterizing the rationality of multi-state collaborative relationships, and lack of quantifiable representation of pre-instability consistency disruption. This invention proposes a real-time assessment method and system that uses the stable-period consistency benchmark as a reference and the degree of consistency disruption as the core. This achieves a shift from "judging abnormal monitoring values" to "judging the disruption of structural collaborative relationships." Through modeling and solidifying the long-term collaborative constraint relationships between multiple state variables, and through continuous quantification and evolutionary analysis of consistency deviation during slope operation, this invention can identify the evolutionary signs of slopes transitioning from stability to instability earlier and more stably. It outputs risk classification and early warning results in a unified index format, thereby improving the timeliness, reliability, and interpretability of slope safety management. Second, this invention expresses the multi-source monitoring information of slope stability operation stage into a unified state variable system, and extracts the long-term synergistic laws under stable conditions to form a reusable state consistency benchmark, so that the operation period assessment has a "comparison standard", avoiding the difficulty of scenario transfer caused by relying solely on empirical thresholds. Through time segment verification and residual analysis, the consistency relationship model is verified and solidified to eliminate spurious relationships caused by accidental correlation or short-term noise, making the benchmark more robust and improving the credibility of subsequent damage degree calculation from the source. Third, this invention transforms the assessment of "whether the multi-state collaborative relationship is disrupted" into a continuously calculable damage index, shifting slope risk assessment from discrete threshold triggering to continuous measurement, making it more suitable for real-time monitoring and dynamic updates. The damage index uses the consistency relationship during the stable period as a reference, reflecting the systematic decoupling process, reducing misjudgments caused by short-term anomalies in a single monitoring quantity, and facilitating the identification of gradual instability processes. By integrating the deviation results of multiple sets of state variable relationships, it can maintain assessment stability even when multi-source monitoring data contains noise, missing data, or local anomalies. It not only uses the instantaneous value of the damage index but also combines its evolutionary characteristics for identification, enabling the differentiation between short-term disturbances and persistent risks, reflecting the true operating state of the slope structure from an overall perspective, and improving the reliability and interpretability of early warnings. Fourth, through comprehensive analysis of trend, continuity and stage characteristics, this invention can capture the key turning point from stability to instability, providing a clearer time window and basis for handling slope safety, realizing real-time graded output of stability, potential instability and high instability, so that the early warning results can directly serve graded response and operation and maintenance decisions, and improve the efficiency of closed-loop management. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a technical roadmap for the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] like Figure 1 As shown: A real-time assessment method for slope collapse based on state consistency failure degree, comprising the following steps: Step 1: When the slope is in a stable operation phase, the slope multi-state variable set is constructed by multi-source monitoring data, and time alignment and scale normalization are performed. The coordinated change law between different state variables is identified by statistical analysis methods, a state consistency relationship model is established and verified and fixed to obtain the stable period state consistency relationship model, forming a stable period benchmark. The specific steps are as follows: Step 1.1: When the slope is in a stable operating phase, the acquired multi-source monitoring data will be used to construct a multi-state variable set for the slope. Among them, the multi-source monitoring data includes deformation state information reflecting the geometric deformation characteristics of soil or rock, stress and strain information reflecting the stress state of slope, seepage state information characterizing the internal hydrodynamic characteristics of slope, external environmental information describing external disturbances or environmental driving factors, and structural integrity information reflecting the degree of damage to slope materials. Deformation state information includes surface displacement, deep displacement and dip angle changes; stress and strain information includes strain values, stress distribution or stress changes measured by strain gauges of soil or support structure; seepage state information includes pore water pressure, groundwater level and seepage rate; external environmental information includes rainfall, temperature and ground motion acceleration; and structural integrity information includes crack width, acoustic emission energy and microseismic count. Multi-state variable set of slope Represented as: in, express The first moment One state variable, This represents the total number of state variables. Step 1.2: Use linear interpolation or moving average to align the time of each state variable in the multi-state variable set of the slope, so as to unify the multi-source monitoring data to the same time step; Step 1.3: Normalize the time-aligned state variables to obtain the stable multi-state data sequence. The normalization formula is: in, To The normalized state variables, which represent the state data during the steady-state period, have a range of values ​​limited to [a certain value range]. Inside, Indicates the first One state variable, and State variables The maximum and minimum values; Step 1.4: Based on the multi-state data sequence during the stable period, statistical analysis methods are used to identify significant synergistic relationships between different state variables during the stable operation phase. The statistical analysis methods include calculating the Pearson correlation coefficient, mutual information analysis, and canonical correlation analysis. The specific steps for identifying significant cooperative relationships between different state variables during the stable operation phase are as follows: During the stable operation phase, the time series is divided into multiple sub-time periods, and the correlation coefficient between any two state variables within the same sub-time period is calculated. The specific determination method is as follows: If the Pearson correlation coefficient of a pair of state variables is greater than a given threshold and the mutual information obtained from mutual information analysis is greater than a given threshold, then the pair of state variables is determined to be a significantly co-linear pair of state variables. If the Pearson correlation coefficient of a pair of state variables is less than or equal to a given threshold, and the mutual information obtained from mutual information analysis is greater than the given threshold, then the pair of state variables is determined to be a nonlinear cooperative state variable pair. If the Pearson correlation coefficient of a pair of state variables is less than or equal to a given threshold, and the mutual information obtained from mutual information analysis is less than or equal to a given threshold, then the pair of state variables is determined to have no stable relationship and is included in the identification of the state variable group. If the Pearson correlation coefficient of a pair of state variables is greater than a given threshold, and the mutual information obtained from mutual information analysis is less than or equal to the given threshold, then the pair of state variables is determined to be a random co-state pair and included in the identification of the state variable group. The state variables included in the identification of the state variable group are formed into multiple sets of state variable sequences of different types or different physical meanings. Canonical correlation analysis is performed on any two sets of state variable sequences to obtain the maximum canonical correlation coefficient. If the maximum canonical correlation coefficient is greater than a given threshold, it is determined that there is a stable multivariate synergistic relationship between the two sets of state variable sequences. Based on this, in the canonical correlation analysis results, the canonical loading values ​​of any two state variables are calculated. If the canonical loading values ​​of the two state variables are greater than a preset threshold, it is determined that the state variable makes a significant contribution to the multivariate synergistic relationship and is included in the state variable group. Otherwise, the state variables already in the state variable group are retained, and the state variables not in the state variable group are not included. For example, state variables , , , , , ,like , , and Incorporate it into the state variable group for judgment. and The preceding steps identified either significantly cooperative linear state variable pairs or nonlinear cooperative state variable pairs. This constitutes... and ,in, Represents values ​​at multiple times. , , same or The expression, for and Canonical correlation analysis is performed. If the maximum canonical correlation coefficient between the two groups is greater than a given threshold, then it is determined that the two groups have a multivariate relationship, and then the following calculations are performed. , , , Given any two typical load values, determine which ones will ultimately be included in the state variable group.

[0021] When calculating the canonical correlation coefficient in this section, it is equivalent to transforming the vector into a matrix as follows: Similarly .

[0022] During the stable operation phase, the state relationships of each pair of state variables and each group of state variables that satisfy the corresponding conditions are determined across multiple sub-time periods, which indicates that the pair of state variables or each group of state variables has a significant synergistic relationship. Among them, "across multiple sub-time periods" means that when the total number of sub-time periods in the stable period exceeds 5, it is more than 80% of the total number of sub-time periods; when the number of sub-time periods in the stable period is less than or equal to 5, it is no less than 3 consecutive sub-time periods. The stable period refers to the stable operation phase. Pearson correlation coefficient The formula is: in, and They represent the first time. Individual time period No. The and the first State variables and The normalized state variables at all times in the process. and It is the first and the The state variables in the sub-time period Normalized mean of state variables, Pearson correlation coefficient The range of values ​​is , Indicates a perfect positive correlation. Indicates a completely negative correlation. This indicates that there is no linear correlation; This means adding up all the times. There is a moment in time and , The actual result is: , , and and State variables and At any moment and The normalized state quantity.

[0023] The formula for obtaining mutual information through mutual information analysis is: in, Indicates the first Individual time period No. The and the first State variables and The probability density function of the joint distribution. and They are and marginal distribution; Canonical correlation analysis refers to setting a sequence of state variables... and , and These represent state variables of different types or with different physical meanings in the first... Individual time segment The canonical correlation analysis will look for a new pair of variables in the sequence of state variables formed above. and And the maximum canonical correlation coefficient is obtained by solving the formula for maximizing the canonical correlation coefficient; The formula for maximizing the canonical correlation coefficient is: in, To maximize the canonical correlation coefficient, For variables and The covariance matrix between them and Representing variables respectively and The cross covariance matrix, and Given the weight vector to be solved, obtained through the generalized eigenvalue method, the objective is to maximize... and The correlation coefficient between them.

[0024] Step 1.5: Establish a state consistency relationship model describing the coupling constraints between state variables during the stable operation phase. Establish a stable state consistency relationship for state variable pairs and state variable groups with significant synergistic relationships. The state consistency relationship model includes a linear synergistic model applicable to significantly synergistic linear state variable pairs, a nonlinear synergistic model applicable to nonlinear synergistic state variable pairs, and a multiple regression model applicable to state variable groups. The linear cooperative model is as follows: in, and They represent the first time. Individual time period No. The and the first State variables and The mean of the normalized state variables at all times in the time series, and and It has a significant synergistic relationship. For linear coordination in sub-time periods No. State variables The predicted value of the normalized mean is affected by the state variable. The target state affected Represents the coupling coefficient, reflecting the state variables. For state variables Intensity of influence Represents state variables With state variables The offset item; The nonlinear cooperative model is as follows: in, This indicates that in the nonlinear cooperative model, during the sub-time period... No. State variables The predicted value of the normalized mean. For random disturbance terms, The mean of the state variable and The mapping function for the nonlinear consistency relationship between state variables can be a polynomial mapping, basis function expansion, kernel function mapping or other continuous nonlinear function forms, used to characterize the nonlinear cooperative relationship between state variables during the steady period. The multiple regression model is as follows: in, This indicates that in multiple regression, within a sub-time period No. State variables The predicted value of the normalized mean. Indicates the first The mean of the state variables The mean of the state variable The linear contribution coefficient, i.e., the degree of its influence, This indicates the number of state variables participating in collaborative modeling. This represents the system baseline term in a multivariate linear relationship, used to describe the theoretical baseline level of the target state when all input states are zero.

[0025] Step 1.6: After extracting the state consistency relationship during the stable period, the stability of the state variable pairs or groups of state variables in the stable period state consistency relationship is verified and the model is solidified based on statistical characteristics through time segment test and residual analysis. When the time segment test and residual analysis simultaneously satisfy the model solidification and stability verification, the stable period state consistency relationship model is obtained. The statistical characteristics include mean and standard deviation. Time-segment testing involves dividing the set of multi-state variables during the stable period into multiple sub-time periods. For each sub-time period, the model parameters of the state consistency relationship model are calculated, including the coupling coefficients in the linear cooperative model. and offset terms Linear contribution coefficient in a multiple regression model and system benchmark items and the random perturbation term in the nonlinear cooperative model After calculating the above parameters in each sub-time period, the stability was verified using the following method: During the stable period, a parameter sample sequence is formed based on the model parameters calculated from all sub-time periods: ,in, Indicates the first The model parameter sample sequence consists of all model parameters calculated in each sub-time period. The average value of the same model parameter is calculated across all sub-time periods. And the degree of dispersion, where the degree of dispersion is the standard deviation. or coefficient of variation ; Based on average Calculate the relative change ratio of each model parameter within each sub-time period: in, Indicates the first The first time period calculated within each sub-time period Model parameters, The first time interval calculated based on all sub-time intervals The mean of each model parameter is used to determine the difference between the relative change ratio and a pre-set threshold value for the parameter mean. If a comparison is made, If the mean of the model parameters remains consistent within the given sub-period, and this condition is met for more than 80% of the sub-periods, then the mean of the model parameters is considered stable during the stable period. The upper quantile value of the relative change ratio of the model parameters during the stable period is determined or given manually, taking into account factors such as whether the sample is sufficient. If so, the same model parameters correspond to the calculation of the mean, assuming there are three sub-time periods; = This is the first sub-time period; = The second sub-time period; = The third sub-time period; So, The model parameter sequence for all time periods; Corresponding to , ...the mean, etc., the model parameters corresponding to each sub-time period are compared with the mean calculated for all sub-time periods, such as... and , and The calculated means are compared.

[0026] Among them, the standard deviation of each model parameter in the parameter sample sequence or coefficient of variation Perform statistical analysis and compare it with the corresponding dispersion threshold. or In comparison, if or If the fluctuation range of the model parameters during the stable period is considered to be controlled, then the dispersion threshold is set. and Determined based on the upper quantile value of the model parameter distribution during the stable period or given manually; If the model parameters of each model are stable at the mean level and the fluctuation range is controlled, then the model is considered to remain stable during the stable period. When a model parameter is determined to be unstable, the following procedure is followed: First, perform model-level processing: If the unstable model parameters continue to show significant drift within a given sub-time period, it indicates that the state variable pair does not have long-term consistency constraints during the stable period. The state variable sequence corresponding to the state variable pair or state variable group should be removed from the stable period state consistency relation set. After model-level processing, data-level processing is performed: If unstable model parameters are concentrated in a specific sub-time period, it indicates that the stable period has not been sufficiently identified. The time segment test should be performed again. If the statistical instability is caused by insufficient number of model parameters, adjacent time segments should be merged and the time segment test should be performed again. After data processing, model solidification processing is performed: Only model parameters that pass the stability test are solidified. Model parameters that fail the stability test are marked as unstable consistency models and isolated for storage, and are not included in the subsequent calculation of state consistency violation index. At the same time, they are continuously monitored as candidate consistency relationships. When they meet the stability judgment conditions again with the addition of new data or during the extended stability period, the stability verification process can be retried. If the model fails the stability judgment in multiple verifications, it is completely removed and no longer participates in subsequent processing. Residual analysis refers to the process of modeling state consistency relationships. Based on historical multi-source monitoring data collected during the stable operation phase, a state consistency relationship model is used to invert and predict the target state variables, obtaining predicted values. These predicted values ​​are then compared with the actual observed values ​​to calculate the difference between the predicted and actual values, forming a residual sequence. The residuals are calculated for the state variables inverted from each model in the state consistency relationship model. The formula is as follows: Among them, the mean of the state variables That is, the actual observed value. , and For the corresponding residuals; during the stable period, statistical analysis is performed on the corresponding residual sequence. When the total number of sub-time periods exceeds 5, more than 80% of the sub-time periods, when the number of sub-time periods is less than or equal to 5, and when there are no less than 3 consecutive sub-time periods, the residual mean is close to zero, the residual variance is within the stable range, and the residuals are randomly distributed without obvious trends or periodic characteristics, then the state consistency relationship model is considered to have good fitting stability and robustness during the stable period, that is, the stable period state consistency relationship model is finally obtained. When the stability and robustness of the state consistency relationship model are poor, the model parameters are re-optimized, data noise and outliers are processed, or thresholds and biases are corrected to improve the stability and robustness of the state consistency relationship model.

[0027] Re-optimize model parameters: The parameters in the model are readjusted by optimizing algorithms (such as least squares, gradient descent, etc.); data-driven methods (such as cross-validation, model selection, etc.) are used to select the optimal parameters to avoid overfitting or underfitting; and the model structure is adjusted.

[0028] Handling data noise and outliers: Remove noise and outliers from the data. Use statistical methods (such as box plots and Z-scores) to detect and remove outlier data points. For monitoring data with large fluctuations, apply filters (such as median filters and moving average filters) to smooth the data and reduce the impact of noise.

[0029] Correction threshold and deviation: When correcting the stationary parameters, the initial threshold can be modified.

[0030] Time segment test and residual analysis are two independent judgment conditions. Time segment test determines whether the model parameters are stable over time; residual analysis determines whether the model fit is stable; only when both are satisfied can the model be solidified and stabilized.

[0031] After completing the stability verification and model solidification of the state consistency relationship model in Step 1.7 and Step 1.6, based on the stable state consistency relationship model and its corresponding model parameters, calculate the residual between the predicted and actual values ​​of the stable state variables, and perform statistical analysis on the residual to obtain statistical analysis characteristics as the stable period benchmark. The statistical analysis characteristics include the mean and standard deviation.

[0032] Step 2: During the slope operation, based on the stable period state consistency relationship model, calculate the coordinated deviation degree between the state variables obtained from the real-time multi-source monitoring data, and construct a state consistency destruction index for the coordinated deviation degree between the slope operation period state and the stable period benchmark. The specific steps are as follows: Step 2.1: During slope operation, based on the stable-period state consistency relationship model, examine the coordination relationship between each group and each pair of state variables in the multi-source monitoring data collected in real time during slope operation, and calculate their coordination deviation. When the When the state consistency relationship is a significantly cooperative linear pair of state variables, its calculation formula is: in, The predicted values ​​obtained from the linear cooperation model in the stable-period state consistency relationship model in the sub-time period Residuals during operation and These are the mean and standard deviation of the stable-period baseline for significantly co-linear state variable pairs, respectively. To prevent small constants with a denominator of zero, the degree of coordination deviation is... Reflects the first The degree of coordination change in the relationship between state variables, i.e., the degree of deviation between the current relationship between state variables and the baseline during the stable period, is called the degree of coordination deviation. The larger the value, the higher the degree of damage to the linear state variable, indicating that the slope stability is threatened; When the When the state consistency relationship is a nonlinear pair of state variables, the calculation steps are as follows: in, The predicted values ​​obtained from the nonlinear cooperative model in the stable-period state consistency relationship model in the sub-time period Residuals during operation and These represent the mean and standard deviation of the nonlinear state variable pair during the stable period, respectively, in the benchmark during the stable period. To prevent the use of tiny constants with a denominator of zero; When the When the state consistency relation is a group of state variables, the calculation formula is: The state variable set is mapped to a collaborative relation quantity for consistency testing, and the predicted values ​​for the runtime period are obtained using the stable-period state consistency relation model. Sub-time periods are then calculated. Operating residuals Based on the statistical stability period residuals, the residuals are standardized using the stability period benchmark, and the cooperative deviation of the state variable group is defined as: in, The predicted values ​​obtained from the multiple regression model in the stable-period state consistency relationship model in the sub-time period Residuals during operation and These represent the mean and standard deviation of the stable-period baseline for the stable-period state variable group, respectively. To prevent tiny constants with a denominator of zero.

[0033] Step 2.2: Comprehensively analyze the cooperative deviation of each pair or group of state variables to obtain the state consistency destruction index used to quantify the cooperative relationship of slope structure. The calculation formula is: in, For the first The nth linear state variable pair, the nth The nth nonlinear state variable pair and the nth The weights of the relationships between state variables. The state consistency violation index is the total number of state relations. This reflects the overall degree of damage to the multi-state collaborative relationship of the slope. A lower value indicates a stronger cooperative relationship between states, suggesting the slope is in a stable state. A higher value indicates that the state coordination relationship has been disrupted, and the slope may be in an unstable stage.

[0034] Step 3: Analyze the continuous changes in the state consistency violation index in conjunction with the time dimension, and extract the trend, persistence and stage characteristics of the violation. The specific steps are as follows: Step 3.1: Based on the state consistency violation index, form a violation time series in chronological order; Step 3.2: The time series of damage is segmented using a sliding window method, specifically by setting the window length. and sliding step size The time series of damage severity is divided into multiple overlapping time windows in chronological order. Each time window corresponds to a continuous sequence of destructiveness, which includes multiple state consistency destructiveness indicators. Step 3.3, within each time window The dynamic change characteristics of the destruction degree sequence are calculated. Specifically, the rate of change of the destruction degree is obtained by fitting or differentiating the trend of the state consistency destruction degree index with time within the time window. At the same time, the volatility index of the state consistency destruction degree index within the time window is calculated, including the standard deviation and amplitude range. The rate of change is used to reflect the upward or downward trend of the destruction degree within the time period, and the volatility index characterizes the intensity of the fluctuation of the destruction degree within the time period. The formula for the rate of change is: in, For the rate of change, For the time increment of change, express State consistency violation index; Step 3.4: Based on the rate of change and volatility indices calculated within the time window, comprehensively analyze the evolution characteristics of the state consistency violation index, and uniformly extract its dynamic evolution characteristics, including trend characteristics, persistence characteristics, and stage characteristics. The time window refers to a sub-time period, specifically: Trend characteristics: Based on the rate of change within multiple time windows, it is determined whether the rate of change is positive in more than three consecutive sub-time periods. If so, it indicates that the slope state consistency failure index is on a continuous upward trend, reflecting the continuous deterioration of the slope structure coordination relationship. If the rate of change is zero or negative in the above sub-time periods, the state consistency failure index is not on a continuous upward trend. Persistent characteristics: To determine whether the state consistency violation index is satisfied at all times within each time window. If so, then the degree of disruption within the corresponding time window is in a high-level range, and the volatility index of the state consistency disruption index is in a relatively stable range, indicating that the change in disruption degree has a significant and continuous characteristic; otherwise, it is not. The average value of the index representing the degree of disruption to the consistency of the state during the stable period. The standard deviation of the index representing the degree of disruption of state consistency during the stable period These are statistical coefficients; Stage-specific characteristics: Identifying significant temporal changes, i.e., inflection points or abrupt changes in the evolution of the state consistency failure index, serves as a stage-specific characteristic of the slope's transition from a stable to an unstable phase. The formula for the evolution of the state consistency failure index is as follows: in, for Indicators of state consistency violation within a time window. For time window The state consistency violation index, if Exceeding the set threshold indicates that the stability of the slope is seriously threatened.

[0035] Step 4: Based on the results obtained in Step 3, perform real-time grading and risk assessment of the slope.

[0036] The specific steps are as follows: Step 4.1: Using the state consistency destruction index and the dynamic evolution characteristics obtained in Step 3.4 as the core criteria, the slope structure state is quantitatively graded and risk warning is achieved through "static threshold division and dynamic trend verification". Specifically: Static grading: During slope operation, the statistical characteristics of the state consistency failure index during the stable period are used as a benchmark to determine the range of thresholds. The range of the thresholds at which the state consistency failure index falls is then assessed. This indicates that the slope is in a stable state. This indicates that the slope is in a state of potential instability. This indicates the slope's height and instability state, and its statistical characteristics include the mean value of the state consistency failure index during the stable period. and standard deviation ; Dynamic verification: based on static hierarchical structure In each case, trend characteristics, persistence characteristics, and stage characteristics are assessed. If any one of these assessments indicates a potential instability state, it means the slope is in a highly unstable state. Specifically: Determine whether the trend is a continuous upward trend. If so, the slope is entering a state of instability; otherwise, it is not. Determine whether the persistent characteristic is in a high-level range. If so, then conduct a secondary confirmation of potential instability; otherwise, it is not. Determine whether the stage-specific characteristics represent a sudden increase. If so, it indicates that the stability of the slope is seriously threatened; otherwise, it is not. The secondary verification of dynamic validation further confirms the preliminary state assessment results by analyzing the trend, persistence, and stage characteristics of the damage, avoiding misjudgments caused by external disturbances or sudden changes. This method can effectively enhance the accuracy and robustness of real-time slope assessment and provide reliable technical support for slope safety management, operation and maintenance, and engineering decision-making through comprehensive verification.

[0037] Step 4.2: Based on the final classification results of Step 4.1, the slope is classified as low risk if it is in a stable state, medium risk if it is in a potentially unstable state, and high risk if it is in a highly unstable state.

[0038] After completing the real-time assessment, the system will provide real-time and reliable technical support for slope safety management through visualization output and decision support functions. ① Real-time assessment result output: The real-time assessment results of the slope, including damage degree, state identification, and trend analysis, will be displayed through a graphical interface. The system presents the current stability level of the slope (stable, potentially unstable, or highly unstable) and related trends in the form of charts, facilitating engineering managers to quickly obtain key information. ② Early warning and decision support: When the slope is in a potentially unstable or highly unstable state, the system will automatically trigger an early warning, generate a report, and send it to relevant personnel via SMS, email, etc. In addition, based on real-time data, the system will also generate decision suggestions to help managers take timely emergency measures, such as on-site inspections or suspension of construction. Through this decision support mechanism, slope risk management can be more scientific and efficient. Through the above steps, a real-time assessment framework based on state consistency damage degree is constructed, which can identify the structural state of the slope in real time and predict the risk of instability. This method provides effective technical support for dynamic monitoring, real-time assessment and emergency management of slopes, and can help relevant departments achieve early warning of slope instability and take timely countermeasures to reduce disaster losses.

Claims

1. A real-time assessment method for slope collapse based on state consistency failure degree, characterized in that, Includes the following steps: Step 1: When the slope is in a stable operation phase, the slope multi-state variable set is constructed by multi-source monitoring data, and time alignment and scale normalization are performed. The coordinated change law between different state variables is identified by statistical analysis methods, a state consistency relationship model is established and verified and fixed to obtain the stable period state consistency relationship model, forming a stable period benchmark. Step 2: During the slope operation, based on the stable period state consistency relationship model, calculate the coordinated deviation degree between the state variables obtained from the real-time multi-source monitoring data, and construct a state consistency destruction index for the coordinated deviation degree between the slope operation period state and the stable period benchmark. Step 3: Analyze the continuous changes in the state consistency violation index in conjunction with the time dimension, and extract the trend, persistence and stage characteristics of the violation. Step 4: Based on the results obtained in Step 3, perform real-time grading and risk assessment of the slope.

2. The real-time assessment method for slope collapse based on state consistency failure degree according to claim 1, characterized in that, The specific steps of step 1 are as follows: Step 1.1: When the slope is in a stable operating phase, the acquired multi-source monitoring data will be used to construct a multi-state variable set for the slope. Among them, the multi-source monitoring data includes deformation state information reflecting the geometric deformation characteristics of soil or rock, stress and strain information reflecting the stress state of slope, seepage state information characterizing the internal hydrodynamic characteristics of slope, external environmental information describing external disturbances or environmental driving factors, and structural integrity information reflecting the degree of damage to slope materials. Deformation state information includes surface displacement, deep displacement and dip angle changes; stress and strain information includes strain values, stress distribution or stress changes measured by strain gauges of soil or support structure; seepage state information includes pore water pressure, groundwater level and seepage rate; external environmental information includes rainfall, temperature and ground motion acceleration; and structural integrity information includes crack width, acoustic emission energy and microseismic count. Multi-state variable set of slope Represented as: in, express The first moment One state variable, This represents the total number of state variables. Step 1.2: Use linear interpolation or moving average to align the time of each state variable in the multi-state variable set of the slope, so as to unify the multi-source monitoring data to the same time step; Step 1.3: Normalize the time-aligned state variables to obtain the stable multi-state data sequence. The normalization formula is: in, To The normalized state variables, which represent the state data during the steady-state period, have a range of values ​​limited to [a certain value range]. Inside, Indicates the first One state variable, and State variables The maximum and minimum values; Step 1.4: Based on the multi-state data sequence during the stable period, statistical analysis methods are used to identify significant synergistic relationships between different state variables during the stable operation phase. The statistical analysis methods include calculating the Pearson correlation coefficient, mutual information analysis, and canonical correlation analysis. The specific steps for identifying significant cooperative relationships between different state variables during the stable operation phase are as follows: During the stable operation phase, the time series is divided into multiple sub-time periods, and the correlation coefficient between any two state variables within the same sub-time period is calculated. The specific determination method is as follows: If the Pearson correlation coefficient of a pair of state variables is greater than a given threshold and the mutual information obtained from mutual information analysis is greater than a given threshold, then the pair of state variables is determined to be a significantly co-linear pair of state variables. If the Pearson correlation coefficient of a pair of state variables is less than or equal to a given threshold, and the mutual information obtained from mutual information analysis is greater than the given threshold, then the pair of state variables is determined to be a nonlinear cooperative state variable pair. If the Pearson correlation coefficient of a pair of state variables is less than or equal to a given threshold, and the mutual information obtained from mutual information analysis is less than or equal to a given threshold, then the pair of state variables is determined to have no stable relationship and is included in the identification of the state variable group. If the Pearson correlation coefficient of a pair of state variables is greater than a given threshold, and the mutual information obtained from mutual information analysis is less than or equal to the given threshold, then the pair of state variables is determined to be a random co-state pair and included in the identification of the state variable group. The state variables included in the identification of the state variable group are formed into multiple sets of state variable sequences of different types or different physical meanings. Canonical correlation analysis is performed on any two sets of state variable sequences to obtain the maximum canonical correlation coefficient. If the maximum canonical correlation coefficient is greater than a given threshold, it is determined that there is a stable multivariate synergistic relationship between the two sets of state variable sequences. Based on this, in the canonical correlation analysis results, the canonical loading values ​​of any two state variables are calculated. If the canonical loading values ​​of the two state variables are greater than a preset threshold, it is determined that the state variable makes a significant contribution to the multivariate synergistic relationship and is included in the state variable group. Otherwise, the state variables already in the state variable group are retained, and the state variables not in the state variable group are not included. During the stable operation phase, the state relationships of each pair of state variables and each group of state variables that satisfy the corresponding conditions are determined across multiple sub-time periods, which indicates that the pair of state variables or each group of state variables has a significant synergistic relationship. Among them, "across multiple sub-time periods" means that when the total number of sub-time periods in the stable period exceeds 5, it is more than 80% of the total number of sub-time periods; when the number of sub-time periods in the stable period is less than or equal to 5, it is no less than 3 consecutive sub-time periods. The stable period refers to the stable operation phase. Step 1.5: Establish a state consistency relationship model describing the coupling constraints between state variables during the stable operation phase. Establish a stable state consistency relationship for state variable pairs and state variable groups with significant synergistic relationships. The state consistency relationship model includes a linear synergistic model applicable to significantly synergistic linear state variable pairs, a nonlinear synergistic model applicable to nonlinear synergistic state variable pairs, and a multiple regression model applicable to state variable groups. Step 1.6: After extracting the state consistency relationship during the stable period, the stability of the state variable pairs or groups of state variables in the stable period state consistency relationship is verified and the model is solidified based on statistical characteristics through time segment test and residual analysis. When the time segment test and residual analysis simultaneously satisfy the model solidification and stability verification, the stable period state consistency relationship model is obtained. The statistical characteristics include mean and standard deviation. After completing the stability verification and model solidification of the state consistency relationship model in Step 1.7 and Step 1.6, based on the stable state consistency relationship model and its corresponding model parameters, calculate the residual between the predicted and actual values ​​of the stable state variables, and perform statistical analysis on the residual to obtain statistical analysis characteristics as the stable period benchmark. The statistical analysis characteristics include the mean and standard deviation.

3. The real-time assessment method for slope collapse based on state consistency failure degree according to claim 2, characterized in that, In step 1.4: Pearson correlation coefficient The formula is: in, and They represent the first time. Individual time period No. The and the first State variables and The normalized state variables at all times in the process. and It is the first and the The state variables in the sub-time period Mean of state variables after normalization, Pearson correlation coefficient The range of values ​​is , Indicates a perfect positive correlation. Indicates a completely negative correlation. This indicates that there is no linear correlation; The formula for obtaining mutual information through mutual information analysis is: in, Indicates the first Individual time period No. The and the first State variables and The probability density function of the joint distribution. and They are and marginal distribution; Canonical correlation analysis refers to setting a sequence of state variables... and , and These represent state variables of different types or with different physical meanings in the first... Individual time segment The canonical correlation analysis will look for a new pair of variables in the sequence of state variables formed above. and And the maximum canonical correlation coefficient is obtained by solving the formula for maximizing the canonical correlation coefficient; The formula for maximizing the canonical correlation coefficient is: in, To maximize the canonical correlation coefficient, For variables and The covariance matrix between them and Representing variables respectively and The cross covariance matrix, and Given the weight vector to be solved, obtained through the generalized eigenvalue method, the objective is to maximize... and The correlation coefficient between them.

4. The real-time assessment method for slope collapse based on state consistency failure degree according to claim 3, characterized in that, In step 1.5: The linear cooperative model is as follows: in, and They represent the first time. Individual time period No. The and the first State variables and The mean of the normalized state variables at all times in the time series, and and It has a significant synergistic relationship. For linear coordination in sub-time periods No. State variables The predicted value of the normalized mean is affected by the state variable. The target state affected Represents the coupling coefficient, reflecting the state variables. For state variables Intensity of influence Represents state variables With state variables The offset item; The nonlinear cooperative model is as follows: in, This indicates that in the nonlinear cooperative model, during the sub-time period... No. State variables The predicted value of the normalized mean. For random disturbance terms, The mean of the state variable and The mapping function for the nonlinear consistency relationship between state variables can be a polynomial mapping, basis function expansion, kernel function mapping or other continuous nonlinear function forms, used to characterize the nonlinear cooperative relationship between state variables during the steady period. The multiple regression model is as follows: in, This indicates that in multiple regression, within a sub-time period No. State variables The predicted value of the normalized mean. Indicates the first The mean of the state variables The mean of the state variable The linear contribution coefficient, i.e., the degree of its influence, This indicates the number of state variables participating in collaborative modeling. This represents the system baseline term in a multivariate linear relationship, used to describe the theoretical baseline level of the target state when all input states are zero.

5. The real-time assessment method for slope collapse based on state consistency failure degree according to claim 4, characterized in that, Step 1.6 specifically includes: Time-segment testing involves dividing the set of multi-state variables during the stable period into multiple sub-time periods. For each sub-time period, the model parameters of the state consistency relationship model are calculated, including the coupling coefficients in the linear cooperative model. and offset terms Linear contribution coefficient in a multiple regression model and system benchmark items and the random perturbation term in the nonlinear cooperative model After calculating the above parameters in each sub-time period, the stability was verified using the following method: During the stable period, a parameter sample sequence is formed based on the model parameters calculated from all sub-time periods: ,in, Indicates the first The model parameter sample sequence consists of all model parameters calculated in each sub-time period. The average value of the same model parameter is calculated across all sub-time periods. And the degree of dispersion, where the degree of dispersion is the standard deviation. or coefficient of variation ; Based on average Calculate the relative change ratio of each model parameter within each sub-time period: in, Indicates the first The first time period calculated within each sub-time period Model parameters, The first time interval calculated based on all sub-time intervals The mean of each model parameter is used to determine the difference between the relative change ratio and a pre-set threshold value for the parameter mean. If a comparison is made, If the mean of the model parameters remains consistent within the given sub-period, and this condition is met for more than 80% of the sub-periods, then the mean of the model parameters is considered stable during the stable period. The value is determined based on the upper quantile of the relative change ratio of the model parameters during the stable period or is given manually. Among them, the standard deviation of each model parameter in the parameter sample sequence or coefficient of variation Perform statistical analysis and compare it with the corresponding dispersion threshold. or In comparison, if or If the fluctuation range of the model parameters during the stable period is considered to be controlled, then the dispersion threshold is set. and Determined based on the upper quantile value of the model parameter distribution during the stable period or given manually; If the model parameters of each model are stable at the mean level and the fluctuation range is controlled, then the model is considered to remain stable during the stable period. When a model parameter is determined to be unstable, the following procedure is followed: First, perform model-level processing: If the unstable model parameters continue to show significant drift within a given sub-time period, it indicates that the state variable pair does not have long-term consistency constraints during the stable period. The state variable sequence corresponding to the state variable pair or state variable group should be removed from the stable period state consistency relation set. After model-level processing, data-level processing is performed: If unstable model parameters are concentrated in a specific sub-time period, it indicates that the stable period has not been sufficiently identified. The time segment test should be performed again. If the statistical instability is caused by insufficient number of model parameters, adjacent time segments should be merged and the time segment test should be performed again. After data processing, model solidification processing is performed: Only model parameters that pass the stability test are solidified. Model parameters that fail the stability test are marked as unstable consistency models and isolated for storage, and are not included in the subsequent calculation of state consistency violation index. At the same time, they are continuously monitored as candidate consistency relationships. When they meet the stability judgment conditions again with the addition of new data or during the extended stability period, the stability verification process can be retried. If the model fails the stability judgment in multiple verifications, it is completely removed and no longer participates in subsequent processing. Residual analysis refers to the process of modeling state consistency relationships. Based on historical multi-source monitoring data collected during the stable operation phase, a state consistency relationship model is used to invert and predict the target state variables, obtaining predicted values. These predicted values ​​are then compared with the actual observed values ​​to calculate the difference between the predicted and actual values, forming a residual sequence. The residuals are calculated for the state variables inverted from each model in the state consistency relationship model. The formula is as follows: Among them, the mean of the state variables That is, the actual observed value. , and For the corresponding residuals; during the stable period, statistical analysis is performed on the corresponding residual sequence. When the total number of sub-time periods exceeds 5, more than 80% of the sub-time periods, when the number of sub-time periods is less than or equal to 5, and when there are no less than 3 consecutive sub-time periods, the residual mean is close to zero, the residual variance is within the stable range, and the residuals are randomly distributed without obvious trends or periodic characteristics, then the state consistency relationship model is considered to have good fitting stability and robustness during the stable period, that is, the stable period state consistency relationship model is finally obtained. When the stability and robustness of the state consistency relationship model are poor, the model parameters are re-optimized, data noise and outliers are processed, or thresholds and biases are corrected to improve the stability and robustness of the state consistency relationship model.

6. The real-time assessment method for slope collapse based on state consistency failure degree according to claim 5, characterized in that, The specific steps of step 2 are as follows: Step 2.1: During slope operation, based on the stable-period state consistency relationship model, examine the coordination relationship between each group and each pair of state variables in the multi-source monitoring data collected in real time during slope operation, and calculate their coordination deviation. When the When the state consistency relationship is a significantly cooperative linear pair of state variables, its calculation formula is: in, The predicted values ​​obtained from the linear cooperation model in the stable-period state consistency relationship model in the sub-time period Residuals during operation and These are the mean and standard deviation of the stable-period baseline for significantly co-linear state variable pairs, respectively. To prevent small constants with a denominator of zero, the degree of coordination deviation is... Reflects the first The degree of coordination change in the relationship between state variables, i.e., the degree of deviation between the current relationship between state variables and the baseline during the stable period, is called the degree of coordination deviation. The larger the value, the higher the degree of damage to the linear state variable, indicating that the slope stability is threatened; When the When the state consistency relationship is a nonlinear pair of state variables, the calculation steps are as follows: in, The predicted values ​​obtained from the nonlinear cooperative model in the stable-period state consistency relationship model in the sub-time period Residuals during operation and These represent the mean and standard deviation of the nonlinear state variable pair during the stable period, respectively, in the benchmark during the stable period. To prevent the use of tiny constants with a denominator of zero; When the When the state consistency relation is a group of state variables, the calculation formula is: The state variable set is mapped to a collaborative relation quantity for consistency testing, and the predicted values ​​for the runtime period are obtained using the stable-period state consistency relation model. Sub-time periods are then calculated. Operating residuals Based on the statistical stability period residuals, the residuals are standardized using the stability period benchmark, and the cooperative deviation of the state variable group is defined as: in, The predicted values ​​obtained from the multiple regression model in the stable-period state consistency relationship model in the sub-time period Residuals during operation and These represent the mean and standard deviation of the stable-period baseline for the stable-period state variable group, respectively. To prevent tiny constants with a denominator of zero; Step 2.2: Comprehensively analyze the cooperative deviation of each pair or group of state variables to obtain the state consistency destruction index used to quantify the cooperative relationship of slope structure. The calculation formula is: in, For the first The nth linear state variable pair, the nth The nth nonlinear state variable pair and the nth The weights of the relationships between state variables. The state consistency violation index is the total number of state relations. This reflects the overall degree of damage to the multi-state collaborative relationship of the slope. A lower value indicates a stronger cooperative relationship between states, suggesting the slope is in a stable state. A higher value indicates that the state coordination relationship has been disrupted, and the slope may be in an unstable stage.

7. The real-time assessment method for slope collapse based on state consistency failure degree according to claim 6, characterized in that, The specific steps of step 3 are as follows: Step 3.1: Based on the state consistency violation index, form a violation time series in chronological order; Step 3.2: The time series of damage is segmented using a sliding window method, specifically by setting the window length. and sliding step size The time series of damage severity is divided into multiple overlapping time windows in chronological order. Each time window corresponds to a continuous sequence of destructiveness, which includes multiple state consistency destructiveness indicators. Step 3.3, within each time window The dynamic change characteristics of the destruction degree sequence are calculated. Specifically, the rate of change of the destruction degree is obtained by fitting or differentiating the trend of the state consistency destruction degree index with time within the time window. At the same time, the volatility index of the state consistency destruction degree index within the time window is calculated, including the standard deviation and amplitude range. The rate of change is used to reflect the upward or downward trend of the destruction degree within the time period, and the volatility index characterizes the intensity of the fluctuation of the destruction degree within the time period. The formula for the rate of change is: in, For the rate of change, For the time increment of change, express State consistency violation index; Step 3.4: Based on the rate of change and volatility indices calculated within the time window, comprehensively analyze the evolution characteristics of the state consistency violation index, and uniformly extract its dynamic evolution characteristics, including trend characteristics, persistence characteristics, and stage characteristics. The time window refers to a sub-time period, specifically: Trend characteristics: Based on the rate of change within multiple time windows, it is determined whether the rate of change is positive in more than three consecutive sub-time periods. If so, it indicates that the slope state consistency failure index is on a continuous upward trend, reflecting the continuous deterioration of the slope structure coordination relationship. If the rate of change is zero or negative in the above sub-time periods, the state consistency failure index is not on a continuous upward trend. Persistent characteristics: To determine whether the state consistency violation index is satisfied at all times within each time window. If so, then the degree of disruption within the corresponding time window is in a high-level range, and the volatility index of the state consistency disruption index is in a relatively stable range, indicating that the change in disruption degree has a significant and continuous characteristic; otherwise, it is not. The average value of the index representing the degree of disruption to the consistency of the state during the stable period. The standard deviation of the index representing the degree of disruption of state consistency during the stable period These are statistical coefficients; Stage-specific characteristics: Identifying significant temporal changes, i.e., inflection points or abrupt changes in the evolution of the state consistency failure index, serves as a stage-specific characteristic of the slope's transition from a stable to an unstable phase. The formula for the evolution of the state consistency failure index is as follows: in, for Indicators of state consistency violation within a time window. For time window The state consistency violation index, if Exceeding the set threshold indicates that the stability of the slope is seriously threatened.

8. The real-time assessment method for slope collapse based on state consistency failure degree according to claim 7, characterized in that, The specific steps of step 4 are as follows: Step 4.1: Using the state consistency disruption index and the dynamic evolution characteristics obtained in Step 3.4 as the core criteria, the slope structure state is quantitatively graded and risk warning is achieved through "static threshold division and dynamic trend verification". Specifically: Static grading: During slope operation, the statistical characteristics of the state consistency failure index during the stable period are used as a benchmark to determine the range of thresholds. The range of the thresholds at which the state consistency failure index falls is then assessed. This indicates that the slope is in a stable state. This indicates that the slope is in a state of potential instability. This indicates the slope's height and instability state, and its statistical characteristics include the mean value of the state consistency failure index during the stable period. and standard deviation ; Dynamic verification: based on static hierarchical structure In each case, trend characteristics, persistence characteristics, and stage characteristics are assessed. If any one of these assessments indicates a potential instability state, it means the slope is in a highly unstable state. Specifically: Determine whether the trend is a continuous upward trend. If so, the slope is entering a state of instability; otherwise, it is not. Determine whether the persistent characteristic is in a high-level range. If so, then conduct a secondary confirmation of potential instability; otherwise, it is not. Determine whether the stage-specific characteristics represent a sudden increase. If so, it indicates that the stability of the slope is seriously threatened; otherwise, it is not. Step 4.2: Based on the final classification results of Step 4.1, the slope is classified as low risk if it is in a stable state, medium risk if it is in a potentially unstable state, and high risk if it is in a highly unstable state.

9. A real-time slope collapse assessment system based on state consistency failure degree, comprising a memory, a processor, and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1-8.

Images