Slope settlement data processing method under infiltration surface effect based on water-soil coupling model

Abstract

The invention discloses a slope settlement data processing method under an infiltration surface effect based on a water-soil coupling model, and relates to the technical field of civil engineering geomechanical analysis. And establishing a water-soil coupling numerical model containing a mutual coupling relationship between water flow permeation and soil deformation. According to the method, the multi-source monitoring data and the sudden change recognition algorithm are fused, so that the recognition precision and real-time performance of the sudden change point of the slope infiltration surface are improved. And a parameter adaptive correction mechanism is introduced, so that the model can dynamically adjust key parameters, and the adaptability to a complex geological environment and the prediction stability are enhanced. Meanwhile, a dynamic processing flow integrating data driving, model calculation and risk identification is constructed, intelligent linkage early warning of the side slope settlement risk is achieved, the landslide accident risk is effectively reduced, and the scientificity and safety of geological disaster monitoring are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of geotechnical engineering geomechanics analysis, and particularly relates to a slope settlement data processing method under the effect of a phreatic surface based on a water-soil coupling model. BACKGROUND

[0002] The "slope settlement data processing under the effect of a phreatic surface based on a water-soil coupling model" is an analysis method that comprehensively considers the changes in slope settlement caused by the interaction between water and soil during the process of surface water infiltration into the soil. Specifically, the method constructs a water-soil coupling model that can dynamically simulate the process of water infiltration and the formation of a phreatic surface under the action of rainfall or other water sources. As the position, shape and expansion speed of the phreatic surface within the slope change, the stress-strain state of the soil also changes, thereby causing different degrees of settlement and displacement. By collecting slope settlement monitoring data and combining the coupling model for processing and fitting analysis, the settlement mechanism can be more accurately identified, the slope stability can be evaluated, and scientific basis can be provided for landslide warning and geological disaster control. This processing method improves the scientificity of data interpretation and prediction accuracy.

[0003] The prior art has the following disadvantages: In the existing slope settlement data processing process under the effect of a phreatic surface based on a water-soil coupling model, the identification of phreatic surface mutation points is a problem that is easily overlooked but has a profound impact. This problem often occurs under sudden heavy rainfall or artificial drainage intervention, and the phreatic surface undergoes nonlinear transition in a short period of time, causing serious deviation in the model's evaluation of water migration paths and soil state. Once the identification fails, the subsequent data processing and settlement prediction will have significant errors, which may mask early signs of slope instability, interfere with early warning judgments, and thus cause serious safety hazards in high-risk areas.

[0004] The above information disclosed in the background section is only used to enhance the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0005] The purpose of the present application is to provide a slope settlement data processing method under the effect of a phreatic surface based on a water-soil coupling model, which improves the identification accuracy and real-time performance of the phreatic surface mutation points of the slope by fusing multi-source monitoring data and mutation identification algorithms. The introduction of a parameter self-adaptive correction mechanism enables the model to dynamically adjust key parameters, enhancing the adaptability and prediction stability of complex geological environments. At the same time, a dynamic processing flow integrating data-driven, model calculation and risk identification is constructed, realizing intelligent linkage warning of slope settlement risk, effectively reducing the risk of landslide accidents, and improving the scientificity and safety of geological disaster monitoring to solve the problems in the above background.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for processing slope settlement data under the wetting surface effect based on a water-soil coupling model, comprising the following steps: Based on the slope geological structure, hydrological characteristics and historical settlement data, a water-soil coupling numerical model that includes the coupling relationship between water infiltration and soil deformation is established; Use distributed sensors to obtain pore water pressure, moisture content, and settlement within the slope, with a data sampling frequency of no less than one minute to obtain dynamic change information with high temporal resolution. Using a multi-physics data fusion algorithm, the collected moisture distribution data is spatially reconstructed to construct a time-evolving wetted surface distribution map. A sliding window-based temporal mutation recognition algorithm is introduced, combined with a joint threshold judgment mechanism of sedimentation rate and pore pressure change to identify the location and time of nonlinear mutation of the infiltration surface. Based on the mutation identification results, the permeability coefficient, shear strength parameters and boundary conditions in the water-soil coupling model are automatically adjusted to enhance the model's ability to respond to emergencies; The revised water-soil coupling model is used to predict future short-term settlement, and an early warning output of the slope instability risk level is given by predicting the trend of the settlement curve.

[0007] Preferably, the unsaturated seepage characteristics and stress-strain history effects of the slope soil are further considered in the process of establishing the water-soil coupling numerical model. By setting up a dual constitutive relationship corresponding to the soil response parameters in the saturated and unsaturated states respectively, a variable rainfall boundary, evaporation boundary and groundwater dynamic boundary are introduced into the model boundary conditions to ensure that the hydrodynamic boundary and the settlement response respond synchronously. At the same time, in order to improve the initial accuracy of the modeling, various geological parameters are used. The interpolation method is used for spatial mapping, which makes the initial conditions of the model continuous and has the ability to express regional details, thereby enhancing the adaptability of the model to the differences in topography and landforms.

[0008] Preferably, laser scanning measurement and GNSS high-precision displacement monitoring system are further integrated in the real-time monitoring data collection process to capture the surface settlement and deformation of the slope at the point cloud level, supplement the surface information that is difficult to obtain with traditional embedded sensors, build a multi-source heterogeneous monitoring data channel, and achieve high-precision time synchronization between different monitoring systems through timestamp calibration technology; at the same time, in order to improve data quality, a Kalman filter algorithm based on anomaly recognition is used to eliminate high-frequency interference and sudden errors, ensuring the stability and physical consistency of the data input to the model.

[0009] Preferably, in the dynamic fitting step of the infiltration surface, the moisture distribution body is constructed by introducing the spatial three-dimensional interpolation method, and the spatial variation trend of pore pressure and water content is converted into a three-dimensional infiltration zone evolution diagram by adopting multi-level linear interpolation combined with volume finite element subdivision technology; at the same time, an "effective infiltration surface identification threshold" is set in the model. When the moisture gradient change rate reaches the set dynamic threshold, the boundary morphology of the current infiltration surface is automatically reconstructed to ensure that it reflects the dynamic evolution process of the real seepage channel and avoid the risk of misidentification caused by static layered division.

[0010] Preferably, a synchronous mutation detection algorithm of the first-order derivative function of sedimentation rate and the second-order derivative function of pore pressure is adopted in the mutation identification mechanism, and the joint mutation index JCI is defined as the triggering criterion for mutation events; when the JCI value exceeds the double threshold within the sliding window range, it is determined that the infiltration surface has transitioned, and the start and end time and spatial range of the mutation interval are recorded, and then a rapid parameter resetting process is entered to ensure the immediate response of the data-driven process.

[0011] Preferably, the mutation identification mechanism uses a weighted dynamic time-delay cross-correlation method and an adaptive time-segmentation entropy algorithm to extract mutation points on the infiltration surface. The specific steps are as follows: Calculate the original sedimentation sequence and pore pressure series The weighted cross-correlation function is calculated as follows: , where is the weighted dynamic time-lag cross-correlation function, which measures the correlation between two time series at different time lags. The degree of matching under It's time The slope settlement monitoring value, is the time weight function, is the total length of the time series, is the slope settlement monitoring value The average value of It's time The pore water pressure, is the time-lagged variable, is the pore water pressure The average value of extract The lag time corresponding to the extreme value in the curve , as the candidate time window of time series anomaly points; The adaptive time-segmented entropy algorithm is used for the time window to calculate the rate of change of information entropy. The calculation expression is as follows: , where is the rate of change of entropy after the time step, is the time step, is the information entropy function, which measures the state uncertainty of the system at time . If is greater than a set threshold , it is determined as a mutation point; Integrate the mutation point set to construct the mutation event set , wherein is the total number of times, which is used for model response correction and early warning triggering.

[0012] Preferably, in the adaptive parameter correction step, the model automatically calls the Bayesian inference module, combines the historical evolution sequence and the current abnormal input state, updates the posterior distribution of the permeability coefficient and the shear strength index, and performs parameter inversion through the Markov chain Monte Carlo algorithm; after parameter correction, the system performs sensitivity analysis on the key control parameters, verifies their stability by setting a change threshold, and if the change is too large, prompts the manual confirmation mechanism to intervene in correction, effectively preventing false alarms caused by parameter drift.

[0013] Preferably, in the adaptive parameter correction, the parameter correction is performed through a multi-objective optimization function based on the disturbance propagation model, and the calculation steps are as follows: Set the objective function group, the expression is as follows: , , wherein is the observed slope settlement data sequence, is the settlement prediction value sequence calculated by the water-soil coupling model based on the current parameter set , is the observed pore water pressure data sequence, is the pore water pressure prediction value sequence calculated by the model based on the current parameter set , is the fitting error objective function of the settlement data, is the fitting error objective function of the pore pressure data; Construct the disturbance propagation function, the formula is as follows: , wherein is the disturbance propagation function, is the parameter set to be optimized, is the slope settlement at time , is the pore water pressure value at time , is the penalty coefficient; A multi-objective iterative algorithm based on particle swarm optimization is used to dynamically optimize and update θ, and the target is as follows: , wherein, is the optimal parameter set obtained by final optimization, is the difference between the simulated settlement value and the measured settlement value, is the difference between the simulated pore pressure value and the measured pore pressure value, is , is the disturbance propagation tolerance threshold; the optimization result is applied to the original water-soil coupling model, and the prediction module is re-run after the parameters are automatically corrected, so that the response of the model to the mutation event has rapidness and accuracy.

[0014] Preferably, a long short-term memory neural network and physical coupling prediction hybrid model is used for settlement trend estimation in the settlement prediction and instability early warning process; The LSTM part is responsible for learning the nonlinear time series characteristics of the settlement data, and the coupling model provides boundary constraints and physical interpretation mechanism, and the outputs of the two are combined as the predicted settlement values for the next one hour, three hours and six hours; The prediction results are also input into the risk assessment module, and the site level, safety factor and historical deformation limit are combined to construct a graded early warning output, and the risk level is quantified into three levels of I, II and III.

[0015] In the above technical solution, the technical effects and advantages provided by the present application are: By introducing a multi-source monitoring data fusion mechanism and a mutation recognition algorithm, the present application can accurately capture the rapid transition behavior of the infiltration surface of the slope under the condition of heavy rainfall or sudden intervention, and make up for the defects of the traditional static water-soil coupling model in response rate and recognition accuracy. Especially through sliding window analysis, information entropy change rate evaluation and weighted dynamic correlation calculation and other algorithm means, the sensitivity and anti-misjudgment ability of the mutation point of the infiltration surface are improved. This way can effectively prevent the delay of early warning caused by model lag in actual engineering, and improve the real-time and scientific nature of geological disaster monitoring from the source.

[0016] By introducing a parameter self-adaptive correction mechanism, the water-soil coupling model has dynamic adjustment capability, and can automatically update the permeability coefficient, strength parameter and boundary condition according to the mutation event, so as to ensure that the model always fits the current actual geological state. In multiple feedback updates, Bayesian inversion and multi-objective disturbance propagation optimization function are used to make the model response more consistent with the real physical evolution law, effectively improving the stability and reliability of the settlement prediction result. This mechanism is particularly suitable for high-risk slope environments with complex and frequently changing geological conditions, and provides a solid technical guarantee for long-term operation of the model.

[0017] The application constructs a dynamic processing flow of trinity of data driving, model calculation and risk identification, breaks through the previous single data dependent early warning mode, and organically fuses physical mechanism, machine learning prediction and engineering early warning standard. Through real-time tracking of the settlement curve trend and multi-level risk grade judgment, the system can automatically link the early warning response measures according to the prediction results, realizes the early intervention to the potential instability event. This mechanism greatly reduces the probability of landslide accidents caused by judgment lag, and effectively improves the intelligent level of infrastructure operation safety and slope disaster prevention and control. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art according to these drawings.

[0019] Figure 1 The method flow chart of the slope settlement data processing method under the effect of the water-soil coupling model based on the wetting surface effect of the present application. DETAILED DESCRIPTION

[0020] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations may, however, be implemented in many different forms and should not be construed as limited to the examples set forth herein; rather, these example implementations are provided so that this disclosure will be thorough and complete, and will fully convey the scope of example implementations to those skilled in the art.

[0021] The present application provides a slope settlement data processing method under the effect of the water-soil coupling model based on the wetting surface effect as shown in Figure 1 The slope settlement data processing method under the effect of the water-soil coupling model based on the wetting surface effect comprises the following steps: Based on the geological structure of the slope, the hydrological characteristics and the historical settlement data, a water-soil coupling numerical model containing the mutual coupling relationship between water flow penetration and soil deformation is established; In the process of establishing the water-soil coupling numerical model, the unsaturated seepage characteristics and the stress-strain history effect of the slope soil are further considered, the double constitutive relationship is set to correspond to the soil response parameters in the saturated and unsaturated states, the variable rainfall boundary, the evaporation boundary and the underground water dynamic boundary are introduced in the boundary conditions of the model, so that the water dynamic boundary and the settlement response can be synchronously responded; at the same time, in order to improve the initial accuracy of modeling, the interpolation method is used for spatial mapping of various geological parameters, so that the model initial conditions have continuity and regional fine expression ability, thereby enhancing the adaptability of the model to the differences in topography and geomorphology.

[0022] ​The pore water pressure, water content and settlement in the slope body are obtained by using distributed sensors, and the data sampling frequency is not less than the minute level, so as to obtain high time resolution dynamic change information. During the real-time monitoring data acquisition process, the laser scanning measurement and GNSS high-precision displacement monitoring system are further integrated, the point cloud level is captured for the surface layer settlement deformation of the slope, the surface layer information which is difficult to obtain by the traditional embedded sensor is supplemented, the multi-source heterogeneous monitoring data channel is constructed, and the high-precision time synchronization between different monitoring systems is realized through the time stamp calibration technology; at the same time, in order to improve the data quality, the Kalman filtering algorithm based on abnormal identification is used to remove high-frequency interference and sudden errors, so as to ensure that the data input into the model has stability and physical consistency.

[0023] The spatial reconstruction of the collected water distribution data is carried out by using the multi-physical field data fusion algorithm, and the distribution map of the infiltration surface evolving with time is constructed. In the dynamic fitting step of the infiltration surface, the spatial three-dimensional interpolation method is introduced to construct the water distribution body, the spatial change trend of the pore pressure and water content is converted into the evolution map of the three-dimensional infiltration zone by using the multi-level linear interpolation combined with the volume finite element subdivision technology; at the same time, the "effective infiltration surface recognition threshold" is set in the model, when the water gradient change rate reaches the set dynamic threshold, the boundary shape of the current infiltration surface is automatically reconstructed, so as to ensure that it reflects the real seepage channel dynamic evolution process, and avoids the misidentification risk caused by the static stratified division.

[0024] The time series mutation recognition algorithm based on sliding window is introduced, and the joint threshold judgment mechanism of the settlement rate and the pore pressure change is combined to identify the position and time of the nonlinear mutation of the infiltration surface. In the mutation recognition mechanism, the synchronous mutation detection algorithm of the first order derivative function of the settlement rate and the second order derivative function of the pore pressure is used, and the joint mutation index JCI (Joint Change Index) is defined as the mutation event triggering criterion; when the JCI value exceeds the double threshold value (i.e. the perturbation threshold value and the outbreak threshold value) in the sliding window range, it is judged that the infiltration surface has transition, and the start and end time and the spatial range of the mutation interval are recorded, and then the fast parameter reorganization process is entered, so as to ensure the immediate response of the data driven process.

[0025] In the mutation recognition mechanism, the weighted dynamic time-lag cross-correlation method (WD-TCC) and the adaptive time segmentation entropy algorithm (ATSEA) are used to extract the mutation point of the infiltration surface, and the specific steps are as follows: The weighted cross-correlation function of the original settlement sequence and the pore pressure sequence is calculated, and the calculation expression is as follows: , in which, is a weighted dynamic time-lag cross-correlation function that measures the correlation between two time series (e.g., slope settlement and pore water pressure ) at different time lags The degree of matching under It's time The slope settlement monitoring value, It is a time weight function, which is used to emphasize the influence of data close to the central time point. is the total length of the time series, is the slope settlement monitoring value The average value of It's time The pore water pressure, is the time-lagged variable, is the pore water pressure The average value of extract The lag time corresponding to the extreme value in the curve , as the candidate time window of time series anomaly points; The adaptive time-segmented entropy algorithm is used for the time window to calculate the rate of change of information entropy. The calculation expression is as follows: , where is the rate of change of entropy after a time step, indicating the degree of mutation of system information, is the time step, Is the information entropy function, which measures the system at time uncertainty of the state; like Greater than the set threshold , it is determined to be a mutation point; Integrate the mutation point set and construct the mutation event set ,in, It is the total number of moments used for model response correction and warning triggering.

[0026] Based on the mutation identification results, the permeability coefficient, shear strength parameters and boundary conditions in the water-soil coupling model are automatically adjusted to enhance the model's ability to respond to emergencies; During the adaptive parameter correction step, the model automatically calls the Bayesian reasoning module, combines the historical evolution sequence with the current abnormal input state, updates the posterior distribution of the permeability coefficient and shear strength index, and performs parameter inversion through the Markov Chain Monte Carlo algorithm (MCMC); after the parameters are corrected, the system conducts a sensitivity analysis on the key control parameters and verifies their stability by setting change thresholds. If the change is too large, the manual confirmation mechanism is prompted to intervene and make corrections, effectively preventing false alarms caused by parameter drift.

[0027] In the adaptive parameter correction, the parameters are corrected through a multi-objective optimization function based on the disturbance propagation model. The calculation steps are as follows: Set the objective function group, the expression is as follows: , , where is the observed slope settlement data series, The water-soil coupling model is based on the current parameter set The calculated settlement prediction value sequence, is the observed pore water pressure data series, Based on the current parameter set The pore water pressure prediction value sequence calculated by the model, is the fitting error objective function of the settlement data, is the fitting error objective function of pore pressure data; Construct the disturbance propagation function, the formula is as follows: , where is the disturbance propagation function, which represents the sensitivity aggregation index of settlement and pore pressure to parameter changes. is the set of parameters to be optimized, It's time The slope settlement, It's time The pore water pressure value, is the penalty coefficient; A multi-objective iterative algorithm based on particle swarm optimization is used to dynamically optimize and update θ, with the following objectives: , where is the optimal parameter set obtained by the final optimization. is the difference between the simulated settlement value and the measured settlement value, is the difference between the simulated pore pressure value and the measured pore pressure value, yes (Constraints are), is the disturbance propagation tolerance threshold; The optimization results Applied to the original water-soil coupling model, the prediction module is re-run after automatic parameter correction to ensure that the model responds quickly and accurately to sudden abnormal events.

[0028] The revised water-soil coupling model is used to predict future short-term settlement, and the slope instability risk level warning output is given by predicting the trend of the settlement curve; The long short-term memory (LSTM) neural network and a physical coupling prediction hybrid model are used in the settlement prediction and instability early warning process to estimate the settlement trend. The LSTM part is responsible for learning the nonlinear time series characteristics of the settlement data, and the coupling model provides boundary constraints and physical interpretation mechanisms. The outputs of the two are combined as the predicted settlement values for the next one hour, three hours, and six hours. The prediction results are also input into the risk assessment module, which combines the site level, safety factor, and historical deformation limit to construct a graded early warning output and quantize the risk level into three levels: I (alert), II (attention), and III (normal).

[0029] Embodiment I: In mountainous or hilly areas where geological disasters frequently occur, heavy rainfall is a key trigger for slope instability and landslides and other geological disasters. Especially during short-term heavy rainfall, rainwater quickly penetrates the surface layer of the soil and spreads downward along the pores, joints, or potential sliding surface, causing the wetting front to rapidly move downward or locally bulge. This drastic change easily induces phenomena such as local softening of the soil, reduction in strength, and weakening of effective stress, thereby inducing accelerated settlement or local sliding. Traditional slope settlement data processing systems generally use static boundaries and fixed seepage path parameters for modeling, which are difficult to respond to the nonlinear evolution process of the wetting front under sudden rainfall conditions, leading to settlement prediction deviating from reality and even false positives or false negatives. Therefore, constructing a water-soil coupling model with real-time perception, mutation recognition, and dynamic response capability under heavy rainfall conditions is the key to achieving accurate prediction and efficient early warning.

[0030] This embodiment proposes a technical system that integrates "high-density monitoring + mutation recognition algorithm + model dynamic parameter adjustment" to address the above problems, which specifically includes the following technical steps: First, a high-density, multi-scale sensor array is deployed in the slope monitoring area. The sensor types include resistance soil moisture meters, miniature pore pressure gauges, soil temperature and humidity composite probes, and deep settlement gauges. These sensors are connected to the central processing unit (data server) through optical fiber, wireless, or wired methods to achieve high-frequency and low-latency data acquisition. The sampling frequency of soil moisture content and pore water pressure is set to 1 time / 5 minutes, and the settlement data acquisition period is not more than 10 minutes to ensure sufficient data timeliness of the system.

[0031] Second, the system integrates a weather data access module to support the acquisition of macro rainfall data from the National Meteorological Bureau API and the joint use of rain gauges and micro-meteorological stations on site to achieve fine perception of the spatial and temporal distribution of rainfall. Once the local rainfall intensity reaches the set threshold (such as 30 mm / h or more), the system immediately enters "emergency response mode" and starts the mutation recognition algorithm.

[0032] The mutation recognition module takes time series mutation detection as the core, uses the weighted sliding window method to extract the time evolution characteristics of the soil response parameters, and combines the pore pressure jump rate (ΔP / Δt) and the sedimentation acceleration (Δ²D / Δt²) as the joint judgment index. If both indexes exceed the adaptive threshold within 5 minutes, the system determines that there is an abnormal jump event of the infiltration surface. In order to improve the recognition accuracy, the system introduces the joint mutation index JCI (Joint Change Index).

[0033] The system continuously calculates JCI and performs difference analysis with the historical average value to ensure that the identified mutation is not affected by noise.

[0034] After identifying the mutation event, the system immediately calls the "adaptive parameter correction module" of the water-soil coupling model. In this module, the model automatically adjusts the following key parameters: rainfall boundary flux, permeability coefficient, porosity, shear strength parameter, etc. Parameter adjustment is realized through real-time updating of Bayesian inference combined with Kalman filtering, ensuring that the corrected model can reflect the force-water state evolution of the soil during the mutation process. At the same time, the infiltration surface boundary line is no longer fixed in layers, but is reconstructed through three-dimensional space interpolation (such as inverse distance weighted interpolation), realizing the dynamic response of the infiltration surface in the time and space domain.

[0035] After completing the model correction, the system immediately re-runs the settlement prediction module, using a hybrid algorithm to estimate the settlement trend in the next 1 hour, 3 hours, and 6 hours. The hybrid algorithm includes a physical mechanism model (based on the corrected water-soil coupling relationship) and a time series prediction model (such as an LSTM neural network), both of which output joint weight prediction values and analyze the first derivative of the prediction curve (representing the settlement trend). Once the settlement rate trend exceeds 70% of the historical maximum value, the system triggers the early warning mechanism, sending risk signals to relevant management platforms, mobile terminals, and on-site alarm devices simultaneously.

[0036] In addition, this embodiment can also be combined with a geographic information system (GIS) platform to realize visual operation. Monitoring data, model results, prediction curves, and mutation events are all displayed in a graphical manner, making it easy for monitoring personnel to keep track of the slope state in real time and assist in decision-making.

[0037] The advantage of this embodiment is that it forms a closed-loop response system from data perception, model recognition, dynamic correction to prediction output, fully solving the problems of "perception delay, recognition lag, and response inaccuracy" in traditional methods in the context of sudden heavy rainfall, significantly improving the real-time prediction and disaster warning capability of slope settlement behavior driven by the mutation of the infiltration surface, and is particularly suitable for mountainous slope environments with frequent rainfall, complex terrain, and high risk levels.

[0038] Embodiment two: In many slope engineering projects, especially artificial fill slope, layered fill slope and interlayer structure soil commonly seen in geotechnical engineering, there are often obvious structural interfaces. These structural interfaces may be formed due to different processes, fillers, compaction differences or natural causes, and they are significantly different from the surrounding soil in physical and mechanical properties, especially in seepage characteristics, such as local permeability coefficient changes, seepage path turning, and stagnant water area gathering. Traditional water-soil coupling models usually adopt the assumption of homogeneous continuous medium, which cannot reflect the disturbance of these structural interfaces to the seepage path and stress field changes, resulting in large prediction deviation and low sensitivity when dealing with phenomena such as sudden change of infiltration surface and seepage stress concentration.

[0039] This embodiment starts from the heterogeneity of structural soil, constructs an intelligent water-soil coupling model with the functions of "structure identification + residual judgment + self-learning parameter adjustment", and systematically solves the problems of infiltration surface identification deviation and prediction lag caused by structural interfaces. The specific implementation path is as follows: First, introduce the structural interface identification module in the modeling stage. Through the fusion analysis of multiple sources of information such as slope history construction records, geological drilling profiles, geophysical imaging (such as seismic wave CT, three-dimensional resistivity imaging), etc., the position, shape and initial physical property parameters of potential structural interfaces are extracted, such as porosity, compression modulus, permeability coefficient, etc. In the model, this part of the unit is given different physical properties from the surrounding, and its parameters are allowed to be dynamically adjusted during calculation, breaking through the limitations of traditional models "fixed parameters, rigid boundaries".

[0040] Second, introduce the residual judgment mechanism in the model calculation process. The system compares the monitoring data (settlement value, pore pressure value, water content) with the model prediction results in real time, and constructs the "model-actual measurement" residual sequence.

[0041] Once the local area residual exceeds the threshold, the system triggers the "structure disturbance analysis" and "parameter inversion" processes. The disturbance analysis part is based on the disturbance propagation function , which traces the contribution of parameter changes in the structural interface region to the overall settlement trend; the inversion part introduces the Bayesian inference framework, taking the current measured data as the observation, and updates the posterior probability of permeability coefficient , shear strength parameters , , etc. A new set of parameter distributions is formed through Markov Chain Monte Carlo (MCMC) sampling, and compared with the historical prior distribution to judge the parameter evolution trend.

[0042] The system further constructs a "self-learning" mechanism, taking the parameter set after each model correction as a new round of training samples to automatically update the structure recognition accuracy and parameter adjustment rules. This mechanism can use the sliding time window technology to integrate residual evolution trends and mutation identification results to model, thereby gradually improving the model's response capability to non-continuous structure area mutation seepage paths. For example, in multiple rainfall-seepage-settlement coupling events, it is observed that the pore pressure response of a certain structure interface area is faster than that of other areas, and the system will give this area a higher sensitivity weight in future prediction, reflecting the mutation process in a timely manner.

[0043] In addition, to achieve stronger generalization capability, the system supports the integration of multi-model learning architecture, such as combining support vector machines (SVM) or convolutional neural networks (CNN) with Bayesian correction modules to fit and predict parameter adjustment results, and to build a "template library" of slope models that can adapt to different terrains and different structural conditions. Each time new monitoring data is encountered, the system can automatically match the most similar template for rapid response.

[0044] Through the above technical path, the present embodiment realizes the transformation of the water-soil coupling model from a "static physical model" to a "dynamic learning model", and is particularly suitable for slope scenarios with complex structural interfaces, historical slip zones, backfill reconstruction projects, etc. In actual engineering applications, this method can significantly reduce the prediction deviation caused by structural misjudgment or parameter mismatch, effectively improve the mutation response speed and settlement prediction accuracy, and provide reliable support for the safety management and engineering control of high-risk slopes.

[0045] Embodiment Three: Fusion Monitoring and Early Warning Application for Slopes Around Major Infrastructure In the operating environment of major infrastructure such as high-speed railways, highways, hydropower stations, and large-scale power transmission channels, slope safety has always been a core link to ensure the stable operation of the system. These infrastructures are usually distributed along the mountain slopes, and some areas have complex geological structures, intense soil disturbance, or frequent human intervention. Once slope settlement anomalies, landslides, or collapses occur, not only will they cause great damage to transportation, water supply, power supply, and other livelihood and economic systems, but also may cause casualties. Therefore, building a slope settlement prediction and early warning system that can cope with complex environments, has mutation identification capability, integrates multi-source monitoring, and supports dynamic response has become an important development direction in the field of slope engineering management.

[0046] This embodiment proposes an integrated technical solution of "multi-source fusion + mutation identification + model-driven + control linkage" for the slope monitoring needs around major infrastructure. The core goal is to use multi-type sensing technology and dynamic water-soil coupling model fusion analysis, combined with mutation identification mechanism and risk early warning rules, to realize early warning and response control of potential slope instability events.

[0047] In terms of system architecture, the method constructs a slope information platform based on the concept of "digital twin". The platform integrates multi-source monitoring data, including but not limited to GNSS high-precision displacement monitoring system, distributed optical fiber settlement sensor, deep displacement meter, pore pressure meter, resistivity profile system, ground radar scanning system, etc. GNSS system is used to capture real-time small displacement changes of surface settlement; optical fiber and deep displacement meter cooperate to monitor the settlement trend and shear displacement of different depth soil layers; pore pressure meter is used to sense the dynamic changes of the phreatic surface; resistivity profile system and radar scanning provide groundwater migration map and surface crack evolution image.

[0048] All monitoring data are integrated and managed with unified time stamp and spatial coordinates, and accessed to the central database for system model calling. In the data fusion layer, the platform uses Bayesian data fusion method and spatio-temporal interpolation reconstruction technology to convert discrete and asynchronous multi-source data into continuous spatio-temporal data stream, and constructs the "state appearance" of the current slope.

[0049] When the system detects that the slope settlement appears trend abnormality (such as continuous rising of settlement rate for multiple time periods or abnormal amplification of displacement curvature), even if it does not exceed the traditional alarm threshold, the system will automatically activate the "mutation recognition module". Based on the joint indicators of settlement acceleration, pore pressure second derivative change and resistivity abnormal area expansion, the module constructs a composite criterion to judge whether there is a potential mutation caused by nonlinear expansion of the phreatic surface in the slope. The system introduces a "phreatic surface mutation scoring function".

[0050] After identifying the real mutation, the platform immediately calls the optimized water-soil coupling model for prediction operation. Here, the model has preset multiple parameter templates for different terrain types, rainfall situations and construction intervention conditions, and has the ability to adjust the boundary conditions adaptively. The model inputs the current multi-source monitoring state, the history of previous response and the predicted meteorological data to carry out the settlement trend prediction for the next 24 hours, and outputs the continuous settlement curve, the phreatic surface change trend graph and the potential instability area distribution graph. At the same time, the platform classifies the risk level according to the predicted settlement and its growth rate, combined with the regional slope danger level, including normal (green), attention (yellow), warning (orange) and emergency (red) four levels.

[0051] To ensure the engineering closed-loop application of early warning effect, the system also connects the early warning results with the operation and control platform. When the prediction result reaches the warning level, the system can link the following response mechanisms: one is to automatically notify the traffic operation control center to adjust the train frequency or speed limit, to ensure safety; two is to push the early warning information to the construction site or slope management unit, to start emergency patrol and soil reinforcement measures; three is to trigger the sound and light alarm device and mobile broadcast system to ensure the safe evacuation of site workers.

[0052] In addition, the system also provides early warning process backtracking and post-analysis functions, records the identification logic, model response and output results of each early warning event in the whole process, and serves as the basis for model continuous optimization and risk prevention and control system construction.

[0053] Through the above technical means and application process, the embodiment not only realizes the whole-process closed loop of "perception-identification-prediction-linkage" in the monitoring of the slope around the major infrastructure, but also significantly improves the advance and accuracy of the slope disaster warning, reduces the false positive rate and false negative rate of the traditional threshold method, and provides a solid technical guarantee for the safe operation of the slope engineering. The method is especially suitable for infrastructure scenes with long distance, cross-terrain and large external interference, and has wide engineering popularization value and social benefits.

[0054] The present application can accurately capture the rapid transition behavior of the slope under the condition of strong rainfall or sudden intervention by introducing a multi-source monitoring data fusion mechanism and a mutation identification algorithm, which makes up for the defects of the traditional static water-soil coupling model in response rate and identification accuracy. Especially through sliding window analysis, information entropy change rate evaluation and weighted dynamic correlation calculation, the sensitivity and anti-misjudgment ability of the mutation point of the infiltration surface are improved. This way can effectively prevent the delay of early warning caused by model lag in practical engineering, and improve the real-time and scientific nature of geological disaster monitoring from the source.

[0055] The present application introduces a parameter self-adaptive correction mechanism, so that the water-soil coupling model has dynamic adjustment capability, can automatically update the permeability coefficient, strength parameter and boundary condition according to the mutation event, and ensures that the model always fits the current actual geological state. In the multi-round feedback update, the Bayesian inversion and multi-target disturbance propagation optimization function are used to make the model response more consistent with the real physical evolution law, effectively improving the stability and reliability of the settlement prediction result. This mechanism is particularly suitable for high-risk slope environments with complex and frequent changes in geological conditions, and provides a solid technical guarantee for long-term operation of the model.

[0056] The present application constructs a dynamic processing flow consisting of data driving, model calculation and risk identification, breaks through the previous single data-dependent early warning method, and organically integrates physical mechanism, machine learning prediction and engineering early warning standard. Through real-time tracking of the settlement curve trend and multi-level risk grade judgment, the system can automatically link the early warning response measures according to the prediction result, realize the early intervention of potential instability events. This mechanism greatly reduces the probability of landslide accidents caused by judgment lag, effectively improves the intelligent level of infrastructure operation safety and slope disaster prevention and control.

[0057] The above formulas are dimensionless to calculate their numerical values, the formula is obtained by collecting a large amount of data to simulate a formula of the latest real situation, and the preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0058] It is apparent that for the person of ordinary skill in the art, modifications can be made to the embodiments described in the foregoing without departing from the spirit and scope of the application. Therefore, the accompanying drawings and descriptions should be regarded in a descriptive sense rather than a restrictive sense.

[0059] It should be noted that, in this text, relational terms such as first and second and the like are used merely to distinguish one entity or action from another, without necessarily requiring or implying any actual such relationship or order between or among the entities or actions. Also, the terms "comprises", "comprising", or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without more limitations, an element defined by the phrase "comprising a..." does not exclude the existence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0060] It should be understood that the size of the sequence number of each process in various embodiments of the present application does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0061] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be realized in electronic hardware, or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0062] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system, device and unit described above can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0063] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e. they can be located in one place or distributed on multiple network units. Part or all of the units can be selected to achieve the purpose of the embodiment according to actual needs.

[0064] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit.

[0065] The above is merely specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0066] The above only describes certain exemplary embodiments of the present application by way of illustration, and it is needless to say that the described embodiments can be modified in various ways without departing from the spirit and scope of the present application for those skilled in the art. Therefore, the above drawings and descriptions are illustrative in nature and should not be understood as limiting the protection scope of the claims of the present application.

Claims

1. A method for processing slope settlement data under the wetting surface effect based on a water-soil coupling model, characterized by: The following steps are involved: Based on the slope geological structure, hydrological characteristics and historical settlement data, a water-soil coupling numerical model that includes the coupling relationship between water infiltration and soil deformation is established; Use distributed sensors to obtain pore water pressure, moisture content, and settlement within the slope, with a data sampling frequency of no less than one minute to obtain dynamic change information with high temporal resolution. Using a multi-physics data fusion algorithm, the collected moisture distribution data is spatially reconstructed to construct a time-evolving wetted surface distribution map. A sliding window-based temporal mutation recognition algorithm is introduced, combined with a joint threshold judgment mechanism of sedimentation rate and pore pressure change, to identify the location and time of nonlinear mutation of the infiltration surface. Based on the mutation identification results, the permeability coefficient, shear strength parameters and boundary conditions in the water-soil coupling model are automatically adjusted to enhance the model's ability to respond to emergencies; The revised water-soil coupling model is used to predict future short-term settlement, and an early warning output of the slope instability risk level is given by predicting the trend of the settlement curve.

2. The method for processing slope settlement data under the wetting surface effect based on the water-soil coupling model according to claim 1 is characterized in that: In the process of establishing the water-soil coupling numerical model, the unsaturated seepage characteristics and stress-strain history effects of the slope soil are further considered. By setting up a dual constitutive relationship corresponding to the soil response parameters in the saturated and unsaturated states respectively, a variable rainfall boundary, evaporation boundary and groundwater dynamic boundary are introduced into the model boundary conditions to ensure that the hydrodynamic boundary responds synchronously with the settlement response. At the same time, in order to improve the initial accuracy of the modeling, various geological parameters are used The interpolation method is used for spatial mapping, which makes the initial conditions of the model continuous and has the ability to express regional details, thereby enhancing the adaptability of the model to the differences in topography and landforms.

3. The method for processing slope settlement data under the wetting surface effect based on the water-soil coupling model according to claim 1 is characterized in that: During the real-time monitoring data collection process, laser scanning measurement and GNSS high-precision displacement monitoring systems are further integrated to capture the surface settlement and deformation of the slope at the point cloud level, supplement the surface information that is difficult to obtain with traditional embedded sensors, build multi-source heterogeneous monitoring data channels, and achieve high-precision time synchronization between different monitoring systems through timestamp calibration technology. At the same time, in order to improve data quality, a Kalman filter algorithm based on anomaly recognition is used to eliminate high-frequency interference and sudden errors, ensuring the stability and physical consistency of the data input to the model.

4. The method for processing slope settlement data under the wetting surface effect based on the water-soil coupling model according to claim 1 is characterized in that: In the dynamic fitting step of the infiltration surface, a spatial three-dimensional interpolation method is introduced to construct a moisture distribution volume. Multi-level linear interpolation combined with volume finite element segmentation technology is used to convert the spatial variation trends of pore pressure and water content into a three-dimensional infiltration zone evolution map. At the same time, an "effective infiltration surface identification threshold" is set in the model. When the rate of change of the moisture gradient reaches the set dynamic threshold, the boundary morphology of the current infiltration surface is automatically reconstructed to ensure that it reflects the dynamic evolution of the actual seepage channel and avoid the risk of misidentification caused by static layered division.

5. The method for processing slope settlement data under the wetting surface effect based on the water-soil coupling model according to claim 1 is characterized in that: The mutation identification mechanism adopts a synchronous mutation detection algorithm of the first-order derivative function of sedimentation rate and the second-order derivative function of pore pressure, and defines the joint mutation index (JCI) as the triggering criterion for mutation events. When the JCI value exceeds the dual threshold within the sliding window, the infiltration surface is judged to have transitioned, and the start and end time and spatial range of the mutation interval are recorded. Then, a rapid parameter resetting process is entered to ensure the immediate response of the data-driven process.

6. The method for processing slope settlement data under the wetting surface effect based on the water-soil coupling model according to claim 1 is characterized in that: In the mutation identification mechanism, the weighted dynamic time-delay cross-correlation method and the adaptive time-segmentation entropy algorithm are combined to extract the mutation points of the infiltration surface. The specific steps are as follows: Calculate the original sedimentation sequence and pore pressure series The weighted cross-correlation function is calculated as follows: , where is the weighted dynamic time-lag cross-correlation function, which measures the correlation between two time series at different time lags. The degree of matching under It's time The slope settlement monitoring value, is the time weight function, is the total length of the time series, is the slope settlement monitoring value The average value of It's time The pore water pressure, is the time-lagged variable, is the pore water pressure The average value of extract The lag time corresponding to the extreme value in the curve , as the candidate time window of time series anomaly points; The adaptive time-segmented entropy algorithm is used for the time window to calculate the rate of change of information entropy. The calculation expression is as follows: , where is the rate of change of entropy after the time step, is the time step, Is the information entropy function, which measures the system at time uncertainty of the state; like Greater than the set threshold , it is determined to be a mutation point; Integrate the mutation point set and construct the mutation event set ,in, It is the total number of moments used for model response correction and warning triggering.

7. The method for processing slope settlement data under the wetting surface effect based on the water-soil coupling model according to claim 1 is characterized in that: During the adaptive parameter correction step, the model automatically calls the Bayesian reasoning module, combines the historical evolution sequence with the current abnormal input state, updates the posterior distribution of the permeability coefficient and shear strength index, and performs parameter inversion through the Markov Chain Monte Carlo algorithm; after the parameters are corrected, the system conducts a sensitivity analysis on the key control parameters and verifies their stability by setting change thresholds. If the change is too large, the manual confirmation mechanism is prompted to intervene and make corrections, effectively preventing false alarms caused by parameter drift.

8. The method for processing slope settlement data under the wetting surface effect based on the water-soil coupling model according to claim 1 is characterized in that: In the adaptive parameter correction, the parameters are corrected through a multi-objective optimization function based on the disturbance propagation model. The calculation steps are as follows: Set the objective function group, the expression is as follows: ， , where is the observed slope settlement data series, The water-soil coupling model is based on the current parameter set The calculated settlement prediction value sequence, is the observed pore water pressure data series, Based on the current parameter set The pore water pressure prediction value sequence calculated by the model, is the fitting error objective function of the settlement data, is the fitting error objective function of pore pressure data; Construct the disturbance propagation function, the formula is as follows: , where is the disturbance propagation function, is the set of parameters to be optimized, It's time The slope settlement, It's time The pore water pressure value, is the penalty coefficient; A multi-objective iterative algorithm based on particle swarm optimization is used to dynamically optimize and update θ, with the following objectives: , where is the optimal parameter set obtained by the final optimization. is the difference between the simulated settlement value and the measured settlement value, is the difference between the simulated pore pressure value and the measured pore pressure value, yes , is the disturbance propagation tolerance threshold; The optimization results Applied to the original water-soil coupling model, the prediction module is re-run after automatic parameter correction to ensure that the model responds quickly and accurately to sudden abnormal events.

9. The method for processing slope settlement data under the wetting surface effect based on the water-soil coupling model according to claim 1 is characterized in that: In the process of settlement prediction and instability warning, a hybrid model based on long-short-term memory neural network and physical coupling prediction is used to estimate the settlement trend; The LSTM part is responsible for learning the nonlinear time series characteristics of the settlement data, while the coupled model provides boundary constraints and physical interpretation mechanisms. The output of the two is combined as the predicted settlement values ​​for the next one, three, and six hours. At the same time, the prediction results are input into the risk assessment module, and a graded warning output is constructed in combination with the site grade, safety factor and historical deformation limit, and the quantified risk levels are divided into three standards: Level I, Level II and Level III.

Images