Slope excavation dynamic support early warning method based on real-time monitoring

Abstract

The application discloses a kind of based on real-time monitoring's side slope excavation dynamic support early warning method, belong to geotechnical engineering informatization monitoring technical field.For the technical problem that existing side slope monitoring early warning method cannot accurately reflect the real mechanical state of side slope, leading to low early warning accuracy, the application is arranged monitoring point in key area of side slope, is configured displacement sensor, stress sensor, pore water pressure gauge and moisture content sensor;Continuous acquisition sensor signal;After filtering denoising and abnormal value rejection to original monitoring data, calculate surface displacement rate, deep displacement increment, stress change rate, pore water pressure ratio and moisture content fluctuation amplitude;Input side slope stability analysis model, solve real-time safety factor;When safety factor is lower than first threshold value, yellow early warning is sent, lower than second threshold value, red early warning is sent and support feedback control is started.The application is used for the stability monitoring and early warning of side slope excavation construction period, improves early warning accuracy.

Description

Technical Field

[0001] This invention relates to the field of information-based monitoring technology in geotechnical engineering. More specifically, this invention relates to a method for early warning of dynamic support during slope excavation based on real-time monitoring. Background Technology

[0002] Slope excavation is a common construction step in projects such as transportation, water conservancy, and mining. The excavation process disrupts the original mechanical balance of the slope, potentially triggering disasters such as landslides. Therefore, monitoring and early warning of slope stability during excavation is crucial.

[0003] Currently, conventional slope monitoring methods mainly involve deploying a single type of sensor on or inside the slope to acquire data on changes in a specific physical parameter, such as displacement, stress, or pore water pressure. In practical engineering applications, threshold values ​​for these parameters are typically preset, and when the monitored values ​​exceed these thresholds, an appropriate level of warning signal is triggered.

[0004] However, the aforementioned monitoring and early warning methods based on single-parameter threshold judgments have significant shortcomings in practice. Slope stability is determined by the combined effects of multiple factors, including the mechanical properties of the soil and rock, groundwater conditions, and stress field distribution; it is a complex mechanical system. Relying solely on changes in a single parameter (such as displacement or stress value) is insufficient to comprehensively and accurately describe the true mechanical state of the slope. For example, an increase in displacement may be caused by various factors and may not directly correspond to a sharp decline in overall stability; similarly, the impact of increased pore water pressure on slope stability is closely related to its stress environment. Judging solely based on whether a single parameter exceeds its limit often fails to accurately assess the overall stability of the slope, easily leading to delayed or false alarms, resulting in low accuracy and failing to meet the actual safety risk management needs of engineering projects. Summary of the Invention

[0005] One object of the present invention is to solve at least the above-mentioned problems and to provide at least the advantages that will be described later.

[0006] To achieve these objectives and other advantages according to the present invention, a method for early warning of dynamic support for slope excavation based on real-time monitoring is provided, comprising the following steps: Step 1: Set up multiple monitoring points in the potential sliding surface area, the stress concentration area at the toe of the slope, and the tension crack area at the top of the slope in the area affected by the slope excavation. Each monitoring point is equipped with a displacement sensor, a stress sensor, a pore water pressure gauge, and a moisture content sensor to measure the physical parameters of the slope rock and soil. Step 2: The data acquisition unit continuously acquires the output signals of each sensor at a set sampling frequency, and transmits the acquired raw monitoring data to the remote data processing center through a wireless communication network. Step 3: The remote data processing center performs filtering and noise reduction and outlier removal on the received raw monitoring data, and then calculates the surface displacement rate, deep displacement increment, stress change rate, pore water pressure ratio, and water content fluctuation range. Step 4: Input the surface displacement rate, deep displacement increment, stress change rate, pore water pressure ratio, and water content fluctuation amplitude obtained in Step 3 into the pre-built slope stability analysis model, and calculate the real-time safety factor of the current slope through the mechanical equilibrium relationship built into the slope stability analysis model. Step 5: Compare the calculated real-time safety factor with the preset first threshold and second threshold. When the safety factor is lower than the first threshold, a yellow warning signal is issued. When the safety factor is lower than the second threshold, a red warning signal is issued and support feedback control is activated. The second threshold is less than the first threshold.

[0007] Preferably, in step one, the displacement sensor at each monitoring point includes a GNSS displacement meter for measuring surface displacement and a fixed inclinometer for measuring deep displacement. The fixed inclinometer obtains horizontal displacement data at different depths by drilling into a borehole buried deep in the slope.

[0008] Preferably, in step two, the data acquisition unit includes a microcontroller, an analog-to-digital converter module connected to the output terminals of each sensor, and a 4G communication module connected to the microcontroller. The microcontroller reads the voltage signals of the sensors through the analog-to-digital converter module and converts them into raw monitoring data in digital form.

[0009] Preferably, in step three, the outlier removal process adopts the Grubbs criterion, which calculates the mean and standard deviation of the original monitoring data, and identifies data points whose absolute residual value is greater than the product of the standard deviation and the lookup coefficient as outliers and removes them.

[0010] Preferably, the slope stability analysis model in step four is a model constructed based on the limit equilibrium method. It corrects the shear strength parameters of the soil and rock mass with the pore water pressure ratio obtained in step three and corrects the sliding force and anti-sliding force in the model with the stress change rate obtained in step three. Among them, correcting the shear strength parameters of soil and rock by pore water pressure ratio means determining the pore water pressure according to the pore water pressure ratio, subtracting the pore water pressure from the total stress to obtain the effective stress, substituting the effective stress into the Mohr-Coulomb strength criterion, and using the corrected effective cohesion and effective internal friction angle as the shear strength parameters of soil and rock. Correcting the sliding force and anti-slip force in the model using the rate of stress change refers to: based on the sampling interval of the data acquisition unit. The stress change rate obtained in step three Multiply by the time interval The stress change during that period is obtained. Then Multiplying by the corresponding area of ​​action A yields the increment of additional force caused by excavation unloading during that time period. ; Increment the additional force These are respectively superimposed onto the sliding force calculation term and anti-slip force calculation term at the bottom of each strip in the limit equilibrium strip method.

[0011] Preferably, in step five, when the safety factor is lower than the first threshold and continues to rise, only a yellow warning signal is issued; when the safety factor is lower than the first threshold and continues to fall, a red warning signal is issued directly.

[0012] Preferably, the support feedback control initiated in step five refers to the remote data processing center sending a control command to the prestressed anchor tension control system located at the toe of the slope via a wireless communication network to increase the anchor tension force.

[0013] Preferably, in step two, the data acquisition unit is equipped with a local storage module and a breakpoint resume mechanism. When the wireless communication network is interrupted, the microcontroller temporarily stores the acquired raw monitoring data in the local storage module and automatically uploads the missing data after the network is restored.

[0014] Preferably, in step three, the filtering and denoising process employs an improved wavelet transform denoising method based on the low-frequency trend characteristics of the slope monitoring signal, specifically including: Using the db4 or sym5 wavelet basis, the original monitoring signal is decomposed into 5 to 7 layers of wavelet decomposition to obtain the high-frequency detail coefficients of each layer and the first-layer approximation coefficients. For the high-frequency detail coefficients of each layer, the threshold of each layer is determined based on unbiased risk estimation, and a soft thresholding function is used to shrink the high-frequency detail coefficients of each layer to obtain the processed wavelet detail coefficients. Polynomial fitting is performed on the first-level approximation coefficients to obtain the fitting curve. The first-level approximation coefficients are subtracted from the fitting curve to obtain the approximation coefficients after removing trend drift. The processed wavelet detail coefficients and the approximation coefficients after removing trend drift are reconstructed using wavelet reconstruction to obtain the filtered monitoring signal.

[0015] Preferably, in step four, the slope stability analysis model combines slope geological stratification information and historical monitoring data, and dynamically corrects the geotechnical parameters in the model using a Bayesian update method to achieve adaptive calibration of the model.

[0016] In step four, the slope stability analysis model undergoes dynamic parameter calibration using the following steps: S1. Obtain geological stratification information of the slope and determine the initial shear strength parameters of each soil and rock layer, including initial cohesion. and initial internal friction angle And set its prior distribution; S2. Collect historical monitoring data within the current window according to the preset sliding time window. The historical monitoring data includes the surface displacement rate, deep displacement increment, pore water pressure ratio sequence and stress change rate sequence obtained in step three. S3. Constructing a forward numerical model: Based on the geological stratification information of the slope, establish a numerical analysis model of the slope, take the pore water pressure ratio and stress change rate as known input loads that change with time, take the shear strength parameters of each layer of rock and soil as variables to be inverted, and calculate the theoretical displacement values ​​at the corresponding time by running the numerical model, including theoretical surface displacement and theoretical deep displacement. S4. Based on the prior distribution of step S1, the surface displacement rate and deep displacement increment collected in step S2 are used as observation data. The likelihood function is constructed using the deviation between the theoretical displacement calculated in step S3 and the observed displacement. The posterior distribution of the shear strength parameters of each layer of rock and soil is calculated using Bayes' formula. S5. Use the expected value of the posterior distribution as the updated cohesion. and internal friction angle Substitute the values ​​into the slope stability analysis model in step four, replace the initial shear strength parameters, and recalculate the real-time safety factor. S6. As the time window slides, after each acquisition of new monitoring data, repeat steps S2 to S5 to recursively update the shear strength parameters.

[0017] The present invention has at least the following beneficial effects: First, this invention overcomes the fundamental deficiency of traditional methods that rely solely on single parameter thresholds for early warning by combining multi-source sensor collaborative monitoring with mechanical model calculation. Traditional methods can only reflect local changes in a certain aspect of the slope and cannot describe the overall stability of the soil and rock mechanical system, easily leading to false alarms or missed alarms due to the one-sidedness of parameter selection. This invention incorporates multiple physical parameters such as displacement, stress, pore water pressure, and water content into the analysis simultaneously, and calculates the real-time safety factor through mechanical equilibrium relationships, so that early warning judgments are based on the actual mechanical state of the slope. This significantly improves the scientific nature and accuracy of early warnings, demonstrating the creativity of leaping from empirical thresholds to mechanistic models.

[0018] Secondly, this invention solves the problem that static models cannot reflect the softening effect of rainfall infiltration and the redistribution of excavation unloading stress by dynamically correcting the mechanical model using pore water pressure ratio and stress change rate. Traditional limit equilibrium methods use fixed parameters and cannot consider the real-time impact of external environmental changes on the mechanical properties of soil and rock, leading to deviations between the calculated safety factor and actual stability. This invention uses the real-time monitored pore water pressure ratio to correct the effective stress parameters and the stress change rate to correct the sliding force and anti-sliding force, enabling the mechanical model to dynamically respond to environmental changes. This represents a creative leap from static analysis to dynamic tracking, significantly improving the real-time performance and accuracy of safety factor calculation.

[0019] Third, this invention solves the problem that relying solely on instantaneous thresholds cannot distinguish between short-term fluctuations and continuous deterioration trends by using a safety factor change trend judgment mechanism. Traditional methods only focus on whether the safety factor is below a preset threshold. When the safety factor fluctuates briefly due to accidental factors, it is easy to trigger unnecessary red alerts, causing construction interference and resource waste. This invention monitors the change trend of the safety factor within a time window. When the safety factor is below the first threshold but continues to rise, only a yellow alert is issued, while a red alert is issued directly when it continues to fall. This can accurately identify the danger signal of deteriorating stability. This method upgrades from "single-point judgment" to "trend recognition," reflecting a creative progress in early warning logic from static comparison to dynamic analysis.

[0020] Fourth, this invention solves the problem of time lag between the discovery of a hazard and the manual operation of support equipment by remotely and automatically initiating prestressed anchor cable tensioning control. Traditional methods require manual intervention on-site after an early warning is issued, which often results in slow response during the critical window of slope instability, potentially missing the optimal support opportunity. This invention directly links the early warning system with the support control system. Once a red warning is triggered, the remote data processing center immediately sends an instruction to increase the tension force to the prestressed anchor cable control system at the slope toe. This method seamlessly connects the "early warning" and "response" stages, upgrading from passive alarm to proactive prevention and control.

[0021] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0022] Figure 1 This is a flowchart of the dynamic support and early warning method for slope excavation based on real-time monitoring, as described in this invention.

[0023] Figure 2 This is a flowchart of the improved wavelet transform denoising method in step three of this invention.

[0024] Figure 3This is a flowchart of the method for dynamically correcting geotechnical mechanical parameters based on Bayesian updates in step four of this invention. Detailed Implementation

[0025] The present invention will now be described in further detail with reference to examples, so that those skilled in the art can implement it based on the description.

[0026] It should be noted that, unless otherwise specified, the experimental methods described in the following implementation plan are all conventional methods, and the reagents and materials described are all commercially available unless otherwise specified.

[0027] like Figure 1 to Figure 3 As shown, this invention provides a real-time monitoring-based dynamic support early warning method for slope excavation, applicable to stability monitoring and early warning during slope excavation construction in engineering projects such as transportation, water conservancy, and mining. This method uses multi-source sensors to monitor the physical parameters of key slope components in real time, dynamically calculates the safety factor using an improved limit equilibrium analysis model, and issues graded early warnings based on the safety factor value and its changing trend, automatically initiating support feedback control when necessary. This method includes: Step 1: Monitoring Point Deployment and Sensor Configuration Within the area affected by slope excavation, the first step is to obtain the layering and initial mechanical parameters of the soil and rock mass based on geological survey data. Preliminary stability calculations are then performed using limit equilibrium methods (such as the simplified Bishop method or the Swedish circular arc method) to identify the location and extent of the most dangerous sliding surface, which serves as the potential sliding surface area. Simultaneously, in conjunction with the excavation design, the possible locations of stress concentration zones at the slope toe and tension crack zones at the slope crest are identified. Multiple monitoring points are deployed in these key areas, each equipped with a comprehensive sensor array to measure various physical parameters of the slope's soil and rock mass.

[0028] Specifically, each monitoring point includes: A GNSS displacement meter is installed on the ground surface to measure the three-dimensional displacement of the slope surface. The sampling frequency can be set to 1 time / minute, and the surface displacement rate is obtained through differential calculation.

[0029] A fixed inclinometer is installed deep into the slope via a borehole, which must penetrate the stable rock layer below the potential sliding surface. Multiple measuring points are arranged along the borehole at different depths (e.g., one every 2 meters) to acquire horizontal displacement data at different depths, thereby obtaining the deep displacement increment. This data can distinguish between surface creep and deep sliding, solving the problem that single surface displacement monitoring cannot accurately determine the deformation mode.

[0030] An earth pressure cell, embedded in the stress concentration zone at the toe of a slope, is used to measure changes in soil stress, with a sampling frequency of once per minute. Simultaneously, pore water pressure is measured using a nearby embedded pore water pressure gauge, which can be used to determine the effective stress principle. Calculate the change in effective stress.

[0031] A pore water pressure gauge, buried near a potential sliding surface, is used to measure pore water pressure at a sampling frequency of 1 time per minute.

[0032] A moisture content sensor is embedded in the surface soil near the tension crack zone at the top of the slope to measure the volumetric moisture content, with a sampling frequency of 1 time per minute.

[0033] All sensors were calibrated before installation to ensure measurement accuracy.

[0034] Step 2: Data Acquisition and Transmission Each monitoring point is equipped with a data acquisition unit, which includes a microcontroller (MCU), an analog-to-digital converter (ADC) connected to the output of each sensor, a local storage module, and a 4G communication module. The MCU continuously reads the output signals (such as voltage, current, and pulses) of each sensor at a sampling frequency of 1 time per minute through the ADC, and converts these signals into raw digital monitoring data. Since the output signal types of different sensors may be inconsistent (e.g., strain gauge sensors output millivolt voltage, while vibrating wire sensors output frequency signals), the ADC can uniformly convert them into digital quantities, solving the technical problem of data integration difficulties caused by inconsistent signal types.

[0035] The data acquisition unit is equipped with a local storage module and a breakpoint resume mechanism to cope with unstable communication signals. The specific workflow is as follows: Set up a non-volatile storage medium (such as an SD card) in the unit, establish a circular storage queue, and store the raw monitoring data in chronological order. Each data record includes the sensor ID, acquisition timestamp, raw monitoring value, and cyclic redundancy check code.

[0036] The microcontroller monitors the network connection status of the 4G communication module in real time. When a network interruption is detected, the currently collected raw monitoring data is sequentially written into a circular storage queue, and the interruption start time is recorded. If the circular storage queue reaches its storage limit, the oldest data record is overwritten according to the first-in, first-out principle.

[0037] The microcontroller continuously monitors the network status. When the network recovers, it reads all unuploaded data records from the interruption start time to the current time from the circular storage queue and uploads them sequentially to the remote data processing center via the 4G communication module according to the order of the collected timestamps. After receiving a successful reception confirmation signal from the center, it marks the data record as uploaded. If the network is interrupted again during the upload process, the upload is paused and the interruption time is re-recorded. Once the network is restored, the upload continues from the last interrupted data record.

[0038] The aforementioned local storage and breakpoint resume mechanism effectively prevents the loss of real-time monitoring data due to communication network interruptions, ensuring the continuity of subsequent safety factor calculations and the reliability of early warnings.

[0039] Step 3: Data Preprocessing and Feature Calculation After receiving the raw monitoring data, the remote data processing center first performs preprocessing, including filtering and noise reduction and outlier removal.

[0040] The filtering and noise reduction process employs a digital low-pass filter (such as a Butterworth filter), with the cutoff frequency set according to the frequency characteristics of the slope deformation signal. Considering that the sensor's sampling frequency is 1 time / minute (approximately 0.0167 Hz), and based on the Nyquist sampling theorem, to avoid signal aliasing, the cutoff frequency of the low-pass filter is set to 0.005 Hz to filter out high-frequency environmental noise and retain the low-frequency trend components reflecting slope deformation.

[0041] Outlier removal was performed using the Grubbs criterion. This was done for the monitoring data sequences of each sensor. Calculate its mean And standard deviation s. For each data point, calculate the residual. The critical value G(α, n) corresponding to the sample size n and the significance level α (usually taken as 0.05) can be obtained by consulting the Grubbs critical value table. Then determine x i Outliers are identified and removed, and then replaced with linear interpolation or neighboring values ​​to ensure data continuity. This process eliminates random noise and spike interference, avoiding misjudgments of stability caused by distortion of subsequent feature parameters due to drastic data fluctuations.

[0042] The preprocessed data is used to calculate the following five characteristic physical quantities. To improve the signal-to-noise ratio and eliminate the influence of high-frequency random fluctuations on trend judgment, before calculating rate-related indicators, the preprocessed displacement and stress sequences are first subjected to moving average filtering. The moving window length can be set to 10-30 minutes based on engineering experience.

[0043] Surface displacement rate v s It is obtained by differencing the displacement sequence after moving average. The unit is mm / min.

[0044] Deep displacement increment Δ d d The displacement difference between the current moment and the previous moment is calculated using horizontal displacement data at different depths measured by a fixed inclinometer, with the unit being mm.

[0045] Stress change rate : Obtained by differential analysis of stress sensor data, The unit is kPa / min.

[0046] Pore ​​water pressure ratio r u Defined as pore water pressure u With total stress s The ratio, , dimensionless. The total stress σ is taken from the measured value of stress sensors (earth pressure cells) at the same or adjacent monitoring points. If no stress sensor is installed at that point, it can be calculated based on the corrected unit weight and burial depth of the overlying soil and rock.

[0047] Moisture content fluctuation range Δ w The difference between the current moment and the previous moment is calculated from the moisture content sensor data, expressed in percent.

[0048] The above-mentioned features are calculated once every minute to form a real-time data stream.

[0049] Step 4: Solving the slope stability analysis model The surface displacement rate, deep displacement increment, stress change rate, pore water pressure ratio, and water content fluctuation amplitude obtained in step three are input into the pre-constructed slope stability analysis model. The slope stability analysis model is constructed based on the limit equilibrium method (specifically, the simplified Bishop method), which can dynamically correct the mechanical parameters of the soil and rock using real-time monitoring data, thereby reflecting the strength softening effect caused by rainfall infiltration and the stress redistribution effect caused by excavation unloading.

[0050] The model correction and solution process is as follows: 1. Dynamic correction of shear strength parameters: based on pore water pressure ratio r u Determine the pore water pressure at the current moment. (in s The total stress can be obtained by actual measurement using a stress sensor or by calculation from the weight of the soil. According to the effective stress principle, the effective stress... Substituting the effective stress into the Mohr-Coulomb strength criterion, we obtain the modified shear strength parameters: in and These are the effective cohesion and the effective internal friction angle, which can be determined through back analysis or empirical formulas (e.g., using empirical relationships) based on initial values ​​from indoor tests combined with field monitoring data. , ,in c 0、 f 0 represents the initial strength parameter. α、β(These are empirical coefficients and can be selected based on the specific soil and rock type). In this embodiment, for simplicity, the saturated-unsaturated strength parameters obtained from laboratory tests are directly used. r u Interpolation is performed based on the corresponding relationship. The specific interpolation method is as follows: Based on a series of direct shear or triaxial test results under different saturation levels (or matrix suction), shear strength parameters are established (…). c' , f′ ) and pore water pressure ratio r u Empirical relationship curves or lookup tables. Based on real-time calculations. r u The effective shear strength parameter at the current moment is determined from the relationship curve using a linear interpolation method. c' and f′ .

[0051] 2. Dynamic correction of sliding force and anti-slip force: adjusting the rate of change of stress. This is considered as the rate of change of additional stress caused by excavation unloading. In this embodiment, the stress sensor is an earth pressure cell, buried vertically, used to measure the change in total vertical stress. Since the safety factor is calculated based on the instantaneous state of static equilibrium, the rate of change needs to be converted into a specific force increment. Let the sampling time interval of the data acquisition unit be Δ. t (For example, 1 minute), then the change in vertical stress generated during this period is: Assuming that the normal and tangential stresses at the bottom of each strip are related to the rate of stress change at that location, σ is multiplied by the area of ​​the corresponding strip's bottom. A (i.e., the arc length of the bottom of the strip) l i Multiply by the unit thickness to obtain the change in vertical additional force caused by excavation unloading during that period. Because stress sensors are only deployed in the stress concentration area at the toe of the slope, it is impossible to directly obtain the precise stress change rate at the bottom of each segment. Therefore, this method assumes that the additional stress field caused by excavation unloading is continuously distributed near the sliding surface and is related to the distance from the excavation face. Specifically, the stress change rate at the stress monitoring point at the toe of the slope is used as the basis for the calculation. Based on the baseline, for any _th i The rate of change of stress at the bottom of each strip. The value is obtained through spatial interpolation based on its location. The interpolation weights can be determined based on the distance between the block and the slope toe, as well as the slope geometry (e.g., using the inverse square distance method). ,in w i Spatial interpolation weighting coefficients determined based on distance attenuation (0 < w i ≤ 1). The vertical additional force Δ FThe direction of its action is vertically downwards. In the limit equilibrium slice method, this vertical force needs to be decomposed onto the bottom surface of each slice: for the i Each strip has a base angle of inclination of . i i Then the increment of the normal force perpendicular to the bottom surface is The increment of the tangential force parallel to the bottom surface is Therefore, the increase in downward force Δ T i It is mainly contributed by the increment of tangential force, that is The increase in anti-slip force Δ R i This is derived from the increase in frictional force provided by the increase in normal force, i.e. The incremental force is added to the original force to obtain the corrected sliding force and anti-slip force.

[0052] Dynamic correction of water content to the unit weight of soil and rock mass: The water content fluctuation range Δ obtained in step three is... w Used to correct the natural unit weight of sliding soil. c Let the initial natural unit weight of the soil be... c 0, saturated bulk density is c sat The current volumetric water content is w The corrected bulk density is ,in w max The maximum possible change in moisture content is set in advance based on the difference between the maximum and minimum values ​​in historical monitoring data, or based on the maximum possible change in moisture content estimated from the soil porosity, to ensure that the correction factor is between 0 and 1. The corrected unit weight... c' Substitute these values ​​into the limit equilibrium slice method and recalculate the weight of each slice. W i This affects the calculation results of sliding force and anti-sliding force. The correction takes into account the direct impact of soil weight gain due to rainfall infiltration on sliding force.

[0053] 3. Safety factor calculation: The simplified Bishop method is used to iteratively solve for the safety factor. F s Its expression is: in l i The arc length of the bottom of the strip. b i The width of the strip. W i For the weight of the strip, α i The bottom inclination angle of the strip. During the iteration process, the pore water pressure ratio will be corrected.c' i , f' i and after stress change rate correction w i (Implicitly contained in the change in the weight of the strip, or by directly adjusting the sliding force term) is substituted into the calculation. Specifically, the additional force Δ caused by the rate of change of stress is... F i This is considered a correction to the weight of the strip, i.e., the equivalent weight. At this point, the downward force becomes The normal force becomes The friction term in the anti-slip force is updated accordingly. To maintain consistency with the previous decomposition, an incremental form can also be used directly: the increment of the sliding force. Increase in anti-slip force These increments are then added to the original sliding force and anti-slip force. The two methods are equivalent; this embodiment uses the equivalent weight method, i.e. Substitute the values ​​into the Bishop's method formula for calculation. After multiple iterations, until the result of two consecutive calculations is obtained... F s If the difference is less than the allowable error (e.g., 0.001), the real-time safety factor for the current moment is obtained. F s ( t ).

[0054] Through the above dynamic corrections, the model can reflect the intensity reduction caused by rainfall infiltration in real time (through...). r u and stress adjustment caused by excavation unloading (through This makes the calculated safety factor more consistent with the actual mechanical state of the slope, and solves the problem that the static mechanical model cannot reflect these dynamic effects, which leads to the distortion of the calculated safety factor value.

[0055] Step 5: Early Warning Judgment and Support Feedback Control The real-time security factor calculated in step four F s With the preset first threshold F th1 and the second threshold F th2 A comparison is made. In this embodiment, a first threshold is set according to engineering design safety requirements and relevant specifications. F th1 = 1.20, second threshold F th2 =1.05, where the second threshold is less than the first threshold. The warning logic is as follows: like F s ≥ Fth1 If the slope is stable, no warning signal will be issued.

[0056] like F s < F th1 Then, we further determine its changing trend. We set a time window and take 10 consecutive safety factor calculation values ​​before the current moment (corresponding to monitoring data within the most recent 10 minutes, since the sampling frequency is 1 time / minute), and calculate the trend index. We use the Mann-Kendall trend test to perform a monotonic trend test on the safety factor sequence within the window. We set the significance level α = 0.05. If the test results show that the sequence has a significant monotonically increasing trend (i.e., the MK statistic), then... Z >0 and p If the value is less than 0.05 and the current safety factor is below the first threshold, a yellow warning signal will be issued, prompting the construction party to pay attention to monitoring and strengthen inspections. If the test results show that the sequence has a significant monotonically decreasing trend (i.e., MK statistic), a yellow warning signal will be issued. Z <0 and p If the value is less than 0.05 and the current safety factor is below the first threshold, a red warning signal will be issued directly, indicating that stability is deteriorating and immediate countermeasures are required. If the significance test is not met, but the safety factor is below the first threshold, only the anomaly will be recorded, and no trend-related warning will be triggered, or manual judgment will be made based on engineering experience.

[0057] like F s < F th2 Regardless of the trend of change, a red warning signal will be issued immediately.

[0058] The aforementioned trend judgment mechanism avoids false alarms that may result from relying solely on a single instantaneous safety factor threshold. For example, when the safety factor falls below the first threshold due to a brief load fluctuation but then rebounds, the continuously rising judgment will only trigger a yellow warning, rather than a red one, thereby reducing unnecessary emergency responses in non-hazardous operating conditions.

[0059] When a red alert signal is issued, the remote data processing center automatically initiates support feedback control. Specifically, the center sends a control command to the prestressed anchor tensioning control system located at the toe of the slope via a wireless communication network, increasing the anchor tension. This command may include a target tension value, for example, calculating the required increase in anchoring force based on the deviation between the current safety factor and a second threshold. ,in kWhere W is the coefficient and W is the total weight of the sliding body; or tensioning is gradually increased in preset steps (e.g., 50kN increments each time) until the safety factor rises above the threshold or the maximum design tension of the anchor cable is reached. Upon receiving the command, the anchor cable tensioning control system automatically starts the hydraulic pump station, increasing the anchor cable tension according to the command value, thereby providing timely support force and enhancing slope stability. This process requires no manual on-site operation, eliminating the time lag between the discovery of a hazard and the implementation of support, ensuring that support measures take effect promptly within the critical window period of slope instability.

[0060] It should be noted that the above thresholds F th1 , F th2 Parameters such as the length of the time window for trend judgment and the coefficient of tension increment should be set in advance according to the geological conditions, slope importance level and design requirements of the specific project, and can be dynamically adjusted according to the feedback of monitoring data during construction to achieve the best early warning and support effect.

[0061] In another embodiment of the present invention, considering that slope excavation construction areas are usually located in the wild or mountainous areas, wireless communication signals are easily affected by factors such as terrain and weather, and network interruptions or instability may occur. To ensure the integrity and continuity of monitoring data and avoid the loss of original monitoring data due to communication interruptions, which would affect the accuracy of subsequent safety factor calculations and the reliability of early warnings, the data acquisition unit is equipped with a local storage module and a breakpoint resume mechanism. Specifically, the data acquisition unit implements local storage and breakpoint resume according to the following steps: ① A non-volatile storage medium, such as an SD card or Flash memory, is provided in the data acquisition unit, and a circular storage queue is established. The circular storage queue stores the raw monitoring data in chronological order, and each data record includes at least the sensor ID, acquisition timestamp, raw monitoring value, and cyclic redundancy check (CRC) code. The CRC code is used for data integrity verification to ensure that the data is not corrupted during storage and transmission.

[0062] ② The microcontroller monitors the network connection status of the 4G communication module in real time, that is, periodically checks whether a communication link can be established with the remote data processing center. When a network interruption is detected (e.g., multiple consecutive connection attempts fail), the microcontroller immediately writes the currently collected raw monitoring data into a circular storage queue in sequence and records the interruption start time. The write operation uses an append method to ensure that the data is arranged in chronological order.

[0063] ③ Due to the limited storage space of non-volatile storage media, the circular storage queue adopts a circular overwrite mechanism: when the queue storage space reaches a preset upper limit (e.g., storage capacity utilization reaches 95% or the number of stored records reaches the maximum value), the oldest data record is automatically overwritten according to the first-in, first-out principle. To ensure that critical data is not lost, the upper limit of storage capacity should be designed based on the longest estimated network interruption time, ensuring that all data collected within this time period can be completely saved. For example, if the estimated longest network interruption time is 72 hours and the sampling frequency is 1 time / minute, the storage capacity should be able to save at least 4320 records. This maximizes the retention of the most recent data while preventing storage space exhaustion that would prevent the writing of new data.

[0064] ④ During data storage, the microcontroller continuously checks the network status at fixed time intervals (e.g., every 10 seconds). When network recovery is detected (e.g., successful connection to a remote data processing center and receipt of a handshake response), the microcontroller reads all unuploaded data records from the interruption start time to the current time from the circular storage queue. To avoid duplicate uploads, uploaded data records are marked as "uploaded" in the queue, or the range of unuploaded data is determined by recording the timestamp of the last successful upload.

[0065] ⑤ The read, unuploaded data records are uploaded sequentially to the remote data processing center via the 4G communication module according to the order of their collection timestamps. During the upload process, after each data record is sent, the microcontroller waits for a successful reception confirmation signal from the remote data processing center. If a confirmation signal is received within the specified time, the data record is marked as uploaded; if no confirmation is received, the record is kept in the queue for later retransmission.

[0066] ⑥ If a network interruption occurs again during the upload process (e.g., signal loss midway through the upload), pause the current upload task and return to step ② to re-enter the data storage mode and record the new interruption start time. Once the network is restored, resume uploading from the last interrupted data record to ensure no duplication or omission.

[0067] Through the aforementioned local storage and breakpoint resume mechanism, this invention effectively solves the problem of real-time monitoring data loss during periods of unstable communication signals or network interruptions in the monitoring area. Even in the event of a prolonged network outage, all original monitoring data is completely preserved in the local storage medium and automatically retransmitted to the remote data processing center after network recovery, thereby ensuring the continuity of safety factor calculation and the reliability of early warning in subsequent steps.

[0068] In another embodiment of the present invention, after receiving the raw monitoring data, the remote data processing center first performs preprocessing, including filtering and denoising, and outlier removal. Since slope monitoring signals typically exhibit low-frequency trend characteristics (reflecting slow deformation of the soil and rock), while environmental noise (such as wind vibration, mechanical vibration, electromagnetic interference, etc.) is mostly high-frequency, traditional filtering methods (such as low-pass filters or moving averages) easily lose useful information from the actual deformation signal while filtering out noise. This leads to distortion of characteristic parameters such as surface displacement rate and stress change rate in subsequent calculations, thus affecting the input accuracy of the slope stability analysis model. To solve this technical problem, this embodiment employs an improved wavelet transform denoising method, the specific steps of which are as follows: Wavelet decomposition: Based on the low-frequency trend characteristics of the slope monitoring signal, the db4 or sym5 wavelet basis suitable for geological signal processing is selected to perform 5-7 levels of wavelet decomposition on the original monitoring signal. The number of decomposition levels is determined according to the signal sampling frequency and noise level. In this embodiment, the sampling frequency is 1 time / minute, and the number of decomposition levels is determined accordingly. J =6. High-frequency detail coefficients of each layer are obtained through wavelet decomposition. d 1, d 2, ..., d J and the first-level approximation coefficient a 1 Among them, the high-frequency detail coefficients mainly contain noise and local abrupt change information, while the approximation coefficients reflect the overall trend of the signal.

[0069] Thresholding of high-frequency detail coefficients: For each layer of high-frequency detail coefficients d j ( j =1, 2, ..., J The threshold for this layer is determined using a method based on Stein's Unbiased Risk Estimate (SURE). l j The SURE criterion selects the optimal threshold by using an unbiased estimate of the mean square error, adaptively balancing noise suppression and signal preservation. Specifically, for a given wavelet coefficient sequence, the SURE value is calculated under different thresholds, and the threshold that minimizes the SURE is selected as the optimal threshold. l j Then, a soft thresholding function is used to shrink the high-frequency detail coefficients of each layer: in, d j ( i ) is the first j Layer i A detailed coefficient, These are the processed wavelet detail coefficients. The soft thresholding function effectively suppresses noise components while maintaining the continuity of the coefficients, avoiding the pseudo-Gibbs phenomenon that may occur with hard thresholding.

[0070] Trend drift removal of approximation coefficients: Level 1 approximation coefficients a 1 It contains the main trend components of the signal, but low-frequency trend drift may be introduced due to factors such as sensor zero-point drift and temperature changes. To eliminate this drift, [the following steps are taken]. a 1 Perform polynomial fitting (usually using a first- or second-order polynomial) to obtain the fitted curve. p ( t Taking a first-order polynomial as an example, let the fitted curve be... The parameters α and β are estimated using the least squares method. Then, the fitted curve is subtracted from the first-level approximation coefficients to obtain the approximation coefficients after removing trend drift. This step can effectively correct the systematic drift of the sensor, making the reconstructed signal more accurately reflect the slope deformation.

[0071] Wavelet reconstruction: The processed wavelet detail coefficients... Approximation coefficient after removing trend drift Inverse wavelet transform (i.e., wavelet reconstruction) is performed to obtain the filtered monitoring signal. This signal effectively removes high-frequency noise and low-frequency drift while preserving the true deformation trend of the slope, providing a high-quality data source for subsequent calculations of characteristic physical quantities such as surface displacement rate and stress change rate.

[0072] By using the improved wavelet transform denoising method described above, this invention can accurately distinguish between real deformation signals and environmental noise, solving the technical problem of characteristic parameter distortion caused by traditional filtering methods, thereby improving the input accuracy and early warning reliability of the slope stability analysis model.

[0073] In another embodiment of the present invention, in order to overcome the technical problem that static mechanical models cannot adaptively adjust parameters with excavation progress or environmental changes, leading to a decline in model prediction ability and an accumulation of safety factor calculation deviations, the slope stability analysis model further combines slope geological stratification information and historical monitoring data, and dynamically corrects the geotechnical mechanical parameters in the model through a Bayesian update method to achieve adaptive calibration of the model. Specifically, this includes the following steps: S1. Obtain geological stratification information of the slope, determine the initial shear strength parameters of each layer of rock and soil, and set its prior distribution.

[0074] Based on the engineering geological survey report, geological stratification information of the slope is obtained, clarifying the distribution range, thickness, and physical and mechanical properties of each soil and rock layer. For each soil and rock layer, its initial shear strength parameters, including initial cohesion, are obtained through indoor triaxial compression tests or direct shear tests. c 0 and initial internal friction angle f 0. Considering the spatial variability of soil and rock parameters and the error in test sampling, c 0 and f 0 is treated as a random variable, and its prior probability distribution is defined. Based on the statistical characteristics of soil and rock parameters, cohesion can usually be assumed. c Following a log-normal distribution, the internal friction angle f It follows a normal distribution. For the th k The a priori distribution of the soil layers can be represented as: , in , and , It is estimated from the sample mean and variance of the initial experimental data. This prior distribution reflects the understanding of the parameters before any monitoring data is obtained.

[0075] S2. Collect historical monitoring data within the current window according to the preset sliding time window.

[0076] Set a sliding time window with a length of T w The value can be determined based on the slope deformation rate and construction progress, for example, it can be taken as follows: T w = 30 days. Based on the current time. t Use the endpoint as the starting point and extract the time interval forward. Historical monitoring data within the area. This data includes the surface displacement rate sequence calculated in step three. Deep displacement increment sequence Pore ​​water pressure ratio sequence and stress change rate sequence ,in t i For each sampling time, i = 1, 2, ..., N, where N is the total number of data points within the window. The surface displacement rate and the deep displacement increment are used as the observation dataset. D obs Using the pore water pressure ratio and stress change rate as the input condition set I .

[0077] S3. Construct a forward numerical model.

[0078] Based on the geological stratification information of the slope, a numerical analysis model of the slope is established (e.g., using finite element software or finite difference programs such as FLAC, ABAQUS, etc.). In this model, the shear strength parameters (cohesion) of each layer of soil and rock are included. c k internal friction angle f k The pore water pressure ratio sequence obtained in step S2 is considered as a random variable to be inverted. and stress change rate sequence A time-varying input load is applied to the model (e.g., by changing the pore water pressure field and stress boundary conditions). By running this numerical model, for any set of assumed shear strength parameters ( c k , f k This allows us to calculate the theoretical displacement value at the corresponding time, including the theoretical surface displacement and the theoretical deep displacement, denoted as... D calc ( c k , f k ; I This forward model establishes a mapping relationship from shear strength parameters to theoretical displacement.

[0079] S4. Construct the likelihood function and calculate the posterior distribution of the shear strength parameters.

[0080] The observed displacement collected in step S2 D obs Assuming the error between the theoretical and observed displacements follows a Gaussian distribution with a mean of zero, and that the errors at each time point are independent, the likelihood function can be expressed as: in, D obs ( t i (time) t i The measured displacement (surface displacement rate or deep displacement increment). Given shear strength parameters When calculating the theoretical displacement using a forward numerical model, it's important to note that the measured displacement used for likelihood function comparisons is different. D obs ( t i and theoretical displacement D calc ( t iThe quantities must be the same physical quantity and have consistent dimensions. Therefore, if the observation data collected in step S2 are "surface displacement rate" and "deep displacement increment", the cumulative displacement time series calculated by the forward numerical model needs to be differentiated or derived to convert it into the rate or increment value at the corresponding time, and then substituted into the likelihood function for calculation. Conversely, the measured rate / increment data can also be integrated to obtain a cumulative displacement series for comparison. This embodiment adopts the former, that is, converting the theoretical cumulative displacement into the theoretical displacement increment. Let σ represent the variance of the observation error, and N represent the total number of data points within the time window. σ is the standard deviation of the observation error, which can be preset based on sensor accuracy and field testing. According to Bayes' theorem, the posterior distribution is proportional to the product of the prior distribution and the likelihood function: Since the posterior distribution typically lacks an analytical form, numerical methods are employed for solution. This embodiment uses the Metropolis-Hastings algorithm within the Markov Chain Monte Carlo (MCMC) method to extract a large number of samples from the posterior distribution and obtain its characteristics through sample statistics. Based on these samples, the posterior expected values ​​of the shear strength parameters for each soil and rock layer are calculated: Where M is the number of valid samples. , For the m-th posterior sample value, , The updated shear strength parameters are used (for substitution into the stability analysis model).

[0081] S5. Use the expected value of the posterior distribution as the updated shear strength parameter, and substitute it into the model to recalculate the real-time safety factor.

[0082] The posterior expected value obtained in step S4 c k ' and f k ' The updated shear strength parameters are used to replace the original initial parameters of the slope stability analysis model in step four. Then, combined with the real-time monitoring data (including surface displacement rate, deep displacement increment, water content fluctuation amplitude, etc., but not limited to these), the real-time safety factor is recalculated according to the limit equilibrium method described in step four. F s ( t Because the updated parameters incorporate historical monitoring information from the most recent time window, they can better reflect the actual mechanical properties of the current soil and rock mass, thus the calculated safety factor has higher accuracy.

[0083] S6. As the time window slides, after each acquisition of new monitoring data, repeat steps S2 to S5 to recursively update the shear strength parameters.

[0084] As time goes on, new monitoring data is continuously generated. The time window sliding step size Δ is set. t (e.g., Δ) t =1 day), after each swipe, the window will move forward Δ t The newly collected data is incorporated, while the oldest data outside the window is discarded, forming a new observation dataset. D obsnew and input condition set I new Then, based on the prior distribution set in step S1 (or the posterior distribution of the previous window can be used as the prior for the current window to achieve recursive Bayesian update), steps S2 to S5 are repeated to recalculate the posterior distribution and update the shear strength parameters. This process is repeated, allowing the model parameters to adaptively adjust with the dynamic changes in the slope excavation process and the external environment (such as rainfall and water level changes), maintaining high prediction accuracy and effectively avoiding the accumulation of safety factor calculation deviations caused by fixed parameters.

[0085] Through the aforementioned Bayesian dynamic update mechanism, the slope stability analysis model possesses self-learning and adaptive capabilities, enabling it to continuously correct geotechnical parameters based on real-time monitoring data. This allows it to maintain high prediction accuracy during long-term monitoring, solving the problem of declining prediction capabilities caused by fixed parameters in traditional static models.

[0086] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. Other modifications can be easily made by those skilled in the art. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and examples shown and described herein.

Claims

1. A method for early warning of dynamic support in slope excavation based on real-time monitoring, characterized in that, Includes the following steps: Step 1: Set up multiple monitoring points in the potential sliding surface area, the stress concentration area at the toe of the slope, and the tension crack area at the top of the slope in the area affected by the slope excavation. Each monitoring point is equipped with a displacement sensor, a stress sensor, a pore water pressure gauge, and a moisture content sensor to measure the physical parameters of the slope rock and soil. Step 2: The data acquisition unit continuously acquires the output signals of each sensor at a set sampling frequency, and transmits the acquired raw monitoring data to the remote data processing center through a wireless communication network. Step 3: The remote data processing center performs filtering and noise reduction and outlier removal on the received raw monitoring data, and then calculates the surface displacement rate, deep displacement increment, stress change rate, pore water pressure ratio, and water content fluctuation range. Step 4: Input the surface displacement rate, deep displacement increment, stress change rate, pore water pressure ratio, and water content fluctuation amplitude obtained in Step 3 into the pre-built slope stability analysis model, and calculate the real-time safety factor of the current slope through the mechanical equilibrium relationship built into the slope stability analysis model. Step 5: Compare the calculated real-time safety factor with the preset first threshold and second threshold. When the safety factor is lower than the first threshold, a yellow warning signal is issued. When the safety factor is lower than the second threshold, a red warning signal is issued and support feedback control is activated. The second threshold is less than the first threshold.

2. The method for early warning of dynamic support for slope excavation based on real-time monitoring according to claim 1, characterized in that, In step one, the displacement sensors at each monitoring point include a GNSS displacement meter for measuring surface displacement and a fixed inclinometer for measuring deep displacement. The fixed inclinometer obtains horizontal displacement data at different depths by drilling into boreholes buried deep in the slope.

3. The method for early warning of dynamic support for slope excavation based on real-time monitoring according to claim 1, characterized in that, In step two, the data acquisition unit includes a microcontroller, an analog-to-digital converter module connected to the output terminals of each sensor, and a 4G communication module connected to the microcontroller. The microcontroller reads the voltage signals of the sensors through the analog-to-digital converter module and converts them into raw monitoring data in digital form.

4. The method for early warning of dynamic support for slope excavation based on real-time monitoring according to claim 1, characterized in that, In step three, the outlier removal process uses the Grubbs criterion to calculate the mean and standard deviation of the original monitoring data. Data points whose absolute residual value is greater than the product of the standard deviation and the lookup table coefficient are identified as outliers and removed.

5. The method for early warning of dynamic support for slope excavation based on real-time monitoring according to claim 1, characterized in that, The slope stability analysis model in step four is a model constructed based on the limit equilibrium method. It uses the pore water pressure ratio obtained in step three to correct the shear strength parameters of the soil and rock mass and the stress change rate obtained in step three to correct the sliding force and anti-sliding force in the model. Among them, correcting the shear strength parameters of soil and rock by pore water pressure ratio means determining the pore water pressure according to the pore water pressure ratio, subtracting the pore water pressure from the total stress to obtain the effective stress, substituting the effective stress into the Mohr-Coulomb strength criterion, and using the corrected effective cohesion and effective internal friction angle as the shear strength parameters of soil and rock. Correcting the sliding force and anti-slip force in the model using the rate of stress change refers to: based on the sampling interval of the data acquisition unit. The stress change rate obtained in step three Multiply by the time interval Δ t The stress change during that period was obtained. Then Multiplying by the corresponding area of ​​action A yields the increment of additional force caused by excavation unloading during that time period. The additional force increment Δ F These are respectively superimposed onto the sliding force calculation term and anti-slip force calculation term at the bottom of each strip in the limit equilibrium strip method.

6. The method for early warning of dynamic support for slope excavation based on real-time monitoring according to claim 1, characterized in that, In step five, when the safety factor is below the first threshold and continues to rise, only a yellow warning signal is issued; when the safety factor is below the first threshold and continues to fall, a red warning signal is issued directly.

7. The method for early warning of dynamic support for slope excavation based on real-time monitoring according to claim 1, characterized in that, The support feedback control initiated in step five refers to the remote data processing center sending a control command to the prestressed anchor tension control system located at the toe of the slope via a wireless communication network to increase the anchor tension force.

8. The method for early warning of dynamic support for slope excavation based on real-time monitoring according to claim 1, characterized in that, In step two, the data acquisition unit is equipped with a local storage module and a breakpoint resume mechanism. When the wireless communication network is interrupted, the microcontroller will temporarily store the acquired raw monitoring data in the local storage module and automatically upload the missing data after the network is restored.

9. The method for early warning of dynamic support for slope excavation based on real-time monitoring according to claim 1, characterized in that, In step three, the filtering and denoising process, based on the low-frequency trend characteristics of the slope monitoring signal, employs an improved wavelet transform denoising method, specifically including: Using the db4 or sym5 wavelet basis, the original monitoring signal is decomposed into 5 to 7 layers of wavelet decomposition to obtain the high-frequency detail coefficients of each layer and the first-layer approximation coefficients. For the high-frequency detail coefficients of each layer, the threshold of each layer is determined based on unbiased risk estimation, and a soft thresholding function is used to shrink the high-frequency detail coefficients of each layer to obtain the processed wavelet detail coefficients. Polynomial fitting is performed on the first-level approximation coefficients to obtain the fitting curve. The first-level approximation coefficients are subtracted from the fitting curve to obtain the approximation coefficients after removing trend drift. The processed wavelet detail coefficients and the approximation coefficients after removing trend drift are reconstructed using wavelet reconstruction to obtain the filtered monitoring signal.

10. The method for early warning of dynamic support for slope excavation based on real-time monitoring according to claim 1, characterized in that, In step four, the slope stability analysis model combines slope geological stratification information and historical monitoring data, and dynamically corrects the geotechnical parameters in the model through the Bayesian update method to achieve adaptive calibration of the model. In step four, the slope stability analysis model undergoes dynamic parameter calibration using the following steps: S1. Obtain geological stratification information of the slope and determine the initial shear strength parameters of each soil and rock layer, including initial cohesion. and initial internal friction angle And set its prior distribution; S2. Collect historical monitoring data within the current window according to the preset sliding time window. The historical monitoring data includes the surface displacement rate, deep displacement increment, pore water pressure ratio sequence and stress change rate sequence obtained in step three. S3. Constructing a forward numerical model: Based on the geological stratification information of the slope, establish a numerical analysis model of the slope, take the pore water pressure ratio and stress change rate as known input loads that change with time, take the shear strength parameters of each layer of rock and soil as variables to be inverted, and calculate the theoretical displacement values ​​at the corresponding time by running the numerical model, including theoretical surface displacement and theoretical deep displacement. S4. Based on the prior distribution of step S1, the surface displacement rate and deep displacement increment collected in step S2 are used as observation data. The likelihood function is constructed using the deviation between the theoretical displacement calculated in step S3 and the observed displacement. The posterior distribution of the shear strength parameters of each layer of rock and soil is calculated using Bayes' formula. S5. Use the expected value of the posterior distribution as the updated cohesion. and internal friction angle Substitute the values ​​into the slope stability analysis model in step four, replace the initial shear strength parameters, and recalculate the real-time safety factor. S6. As the time window slides, after each acquisition of new monitoring data, repeat steps S2 to S5 to recursively update the shear strength parameters.

Images