A multi-parameter coupling risk assessment method for an engineering geological body

Abstract

The application discloses an engineering geologic body multi-parameter coupling risk assessment method, belongs to the technical field of geological safety monitoring, and can solve the problem that the existing geologic body risk assessment method cannot realize dynamic and automatic updating of water conservancy and hydropower engineering geologic body risk assessment with multi-source real-time monitoring data and prediction of future risk trends. The method comprises the following steps: S1, generating an initial risk distribution according to initial geological data of a target engineering area and a preset coupling rule; S2, determining target parameter prediction values and risk trend prediction values within a preset time length according to real-time monitoring data flow, geological update data and the initial risk distribution; S3, generating a dynamic risk threshold value according to the target parameter prediction values and the risk trend prediction values and in combination with a preset engineering safety constraint condition; and S4, comparing a current measurement value of the real-time monitoring data flow with the dynamic risk threshold value to determine a dynamic risk grade of the target engineering area. The application is used for risk assessment of water conservancy and hydropower engineering geologic bodies.

Description

TECHNICAL FIELD

[0001] The present application relates to a kind of engineering geologic body multi-parameter coupling risk assessment method, belong to geological safety monitoring technical field. BACKGROUND

[0002] With the acceleration of clean energy transformation, as the core way of large-scale energy storage, the scale and complexity of water conservancy and hydropower engineering construction increase day by day. The safety of the project is highly dependent on the accurate risk assessment of the stability of the engineering geologic body where the key parts such as underground powerhouse and water conveyance system are located. However, such engineering construction has a long construction period and frequent switching of operating conditions, resulting in dynamic changes in the load, stress and seepage environment of the geologic body, and the geological conditions are often changed or "unknown section" during construction. This makes it difficult for the static risk assessment results based on the limited data in the survey and design stage to reflect the real-time evolution of the geologic body state and risk during the construction and operation process, and there is a key technical bottleneck of lagging risk warning and declining accuracy with the progress of the project.

[0003] Existing geologic body risk assessment techniques usually rely on numerical simulation combined with deterministic or probabilistic analysis framework (such as Monte Carlo method), or comprehensive evaluation models based on expert scoring and analytic hierarchy process. Some methods attempt to integrate construction or operation period monitoring data (such as displacement, stress, and seepage pressure) for model updating. However, these updates often rely on manual intervention by engineers for periodic recalculation or manual parameter adjustment, which is time-consuming and slow to respond. Or only based on a fixed threshold to determine whether the current state is over-limit, lacking the ability to predict future risk trends based on data flow and adaptive dynamic warning mechanism. Existing methods fail to fully utilize the streaming characteristics of real-time monitoring data, and cannot build an automated, real-time / quasi-real-time risk assessment dynamic updating closed loop. SUMMARY

[0004] The present application provides an engineering geologic body multi-parameter coupling risk assessment method, which can solve the problem that existing geologic body risk assessment methods cannot dynamically and automatically update water conservancy and hydropower engineering geologic body risk assessment with multi-source real-time monitoring data and predict future risk trends.

[0005] The present application provides an engineering geologic body multi-parameter coupling risk assessment method, which comprises:

[0006] S1, generating an initial risk distribution of the target engineering area according to the initial geological data of the target engineering area and the preset coupling rules;

[0007] S2, obtaining real-time monitoring data stream and geological update data of the target engineering area, and determining target parameter prediction value and risk trend prediction value within a preset time period according to the real-time monitoring data stream, the geological update data and the initial risk distribution;

[0008] S3, predicting according to the target parameter prediction value and the risk trend prediction value, and combining a preset engineering safety constraint condition, a dynamic risk threshold value is generated;

[0009] S4, comparing the current measurement value of the real-time monitoring data stream with the dynamic risk threshold value, and determining a dynamic risk level of the target engineering area.

[0010] Optionally, the S1 specifically comprises:

[0011] According to the initial geological data of the target engineering area, a three-dimensional geological numerical model is constructed;

[0012] According to a preset coupling rule, a stability quantitative index of the geological body of each grid cell in the three-dimensional geological numerical model is calculated;

[0013] Based on the spatial distribution of the stability quantitative index, an initial risk distribution of the target engineering area is generated.

[0014] Optionally, the S2 specifically comprises:

[0015] Through a distributed sensor network, monitoring data of surrounding rock displacement, anchor stress and pore water pressure of the target engineering area is continuously collected to form a real-time monitoring data stream;

[0016] The re-measurement data of the fault zone of the target engineering area obtained by a geological radar is taken as the geological update data.

[0017] Optionally, the S2 specifically comprises:

[0018] A long short-term memory network is trained using the initial risk distribution to obtain a deep time series prediction model, and a three-dimensional input tensor is determined according to the real-time monitoring data stream, the geological update data and the initial risk distribution;

[0019] The three-dimensional input tensor is input into the deep time series prediction model to obtain the target parameter prediction value and the risk trend prediction value within a preset time length.

[0020] Optionally, the three-dimensional input tensor is determined according to the real-time monitoring data stream, the geological update data and the initial risk distribution, specifically comprising:

[0021] The geological update data and the initial risk distribution are spatially registered and fused to generate a spatial enhancement feature;

[0022] The real-time monitoring data stream is subjected to sliding window normalization processing, and the processed real-time monitoring data stream is spliced with the spatial enhancement features in the time dimension to obtain a three-dimensional input tensor.

[0023] Optionally, the S3 specifically comprises:

[0024] The fluctuation interval of the target parameter prediction value under a preset confidence level is calculated.

[0025] The tolerance coefficient of the fluctuation interval is adjusted according to the change slope of the risk trend prediction value.

[0026] A dynamic risk threshold is generated according to the adjusted tolerance coefficient and a preset engineering safety constraint condition.

[0027] Optionally, the S4 specifically comprises:

[0028] The absolute deviation value and the relative change rate between the current measurement value of the real-time monitoring data stream and the dynamic risk threshold are calculated.

[0029] The dynamic risk level of the target engineering area is determined according to the absolute deviation value and the relative change rate.

[0030] Optionally, the determination of the dynamic risk level of the target engineering area according to the absolute deviation value and the relative change rate specifically comprises:

[0031] When the absolute deviation value exceeds a first-level deviation threshold or the relative change rate exceeds a first-level change rate threshold, the high risk level is determined.

[0032] When the absolute deviation value does not exceed a second-level deviation threshold and the relative change rate does not exceed a second-level change rate threshold, the low risk level is determined; otherwise, the medium risk level is determined.

[0033] Optionally, after the S4, the method further comprises:

[0034] S5, a multi-level early warning signal is generated according to the dynamic risk level of the target engineering area, and the multi-level early warning signal is sent to a management and control platform.

[0035] Optionally, the preset confidence level is 90% to 95%.

[0036] The beneficial effects that can be produced by the present application include:

[0037] The engineering geology body multi-parameter coupling risk assessment method provided by the present application determines the target parameter prediction value and the risk trend prediction value in a future preset time length through the real-time monitoring data stream of the target engineering area, the geological update data and the initial risk distribution, realizes static and dynamic data fusion prediction, and improves the displacement prediction error.

[0038] The engineering geological body multi-parameter coupling risk assessment method provided by the application first generates an initial risk distribution based on initial geological data and a preset coupling rule. Then, real-time monitoring data flow and geological update data of the water conservancy and hydropower engineering area are obtained, which are input into a pre-trained deep time series prediction model together with the initial risk distribution, and target parameter prediction values and risk trend prediction values within a specified time period in the future are output. The target parameter prediction values are associated with preset monitoring items that affect the stability of the geological body. Then, in combination with preset engineering safety constraint conditions, dynamic risk threshold values are generated according to the prediction values. Then, the current measurement values of the real-time monitoring data flow are compared with the threshold values to determine the dynamic risk level and generate corresponding multi-level early warning signals for output. This method solves the technical problem that existing water conservancy and hydropower engineering geological body risk assessment cannot be dynamically and automatically updated with multi-source real-time monitoring data and predict future risk trends, thereby realizing dynamic tracking and early prediction of the coupling risk state of the water conservancy and hydropower engineering geological body based on real-time monitoring data, and significantly improving the timeliness and initiative of risk warning. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 The engineering geological body multi-parameter coupling risk assessment method flowchart provided by the embodiments of the application is shown. DETAILED DESCRIPTION

[0040] The application will be described in detail below with reference to the embodiments, but the application is not limited to these embodiments.

[0041] The embodiments of the application provide an engineering geological body multi-parameter coupling risk assessment method, as shown in Figure 1 The method comprises the following steps.

[0042] S1, generating an initial risk distribution of a target engineering area according to initial geological data of the target engineering area and a preset coupling rule.

[0043] Specifically, first, a three-dimensional geological numerical model is constructed according to the initial geological data of the target engineering area; then, a stability quantitative index of the geological body in each grid cell of the three-dimensional geological numerical model is calculated according to the preset coupling rule; and finally, the initial risk distribution of the target engineering area is generated based on the spatial distribution of the stability quantitative index.

[0044] The steps in the specific implementation include:

[0045] 1. Establish a three-dimensional geological numerical model.

[0046] 1.1 Data input.

[0047] The initial geological data is imported into a geological modeling software (GOCAD or Midas GTS NX is adopted), and the initial geological data includes:

[0048] Drilling core data (lithology, RQD value, permeability coefficient);

[0049] Geological profile (scale 1:500);

[0050] Geophysical exploration data (seismic wave velocity VP / VS ratio);

[0051] In-situ stress measurement results (maximum principal stress azimuth angle and magnitude).

[0052] 1.2 Model construction.

[0053] 1.2.1 Spatial interpolation of discrete borehole data using inverse distance weighting method (formula):

[0054] ;

[0055] Where: is the attribute value of the grid point to be interpolated; is the attribute value of the known point; is the Euclidean distance between the known point and the point to be interpolated; is the power parameter (take the default value 2).

[0056] 1.2.2 Divide the hexahedral grid elements in the modeling software, the element size is set as follows according to the engineering partition:

[0057] Underground powerhouse area: grid edge length ≤ 5 m;

[0058] Water conveyance tunnel area: grid edge length ≤ 10 m;

[0059] Reservoir basin area: grid edge length ≤ 20 m.

[0060] 1.3 Model verification.

[0061] Calculate the model error by cross-validation method: randomly reserve 10% of the borehole data as the validation set, calculate the root mean square error (RMSE) of the interpolation results and the measured values, and require RMSE ≤ 15%.

[0062] 2. Calculate the stability quantitative index of the geological body.

[0063] 2.1 Coupling rule application: execute the preset coupling rule (based on Hoek-Brown criterion and seepage-stress coupling equation) to perform iterative calculation for each grid element.

[0064] 2.1.1 Rock mass strength parameter calculation (Hoek-Brown criterion is adopted):

[0065] ;

[0066] In the above formula, is the peak strength of rock mass; is the confining pressure; is the uniaxial compressive strength of intact rock (laboratory measured value); , , are all functions of Geological Strength Index (GSI) determined by the characteristics of rock mass discontinuities.

[0067] 2.1.2 Seepage-stress coupling calculation (Biot theory is adopted):

[0068] ;

[0069] where, is the permeability tensor of rock mass (obtained from field water pressure test); is the pore water pressure; is the storage coefficient; is the Biot coefficient (value range 0.5~0.9); is the displacement vector of rock mass.

[0070] 2.2 Stability quantification index calculation.

[0071] Output three indexes of each grid element:

[0072] 2.2.1 Safety factor = shear strength / shear stress (Moor-Coulomb criterion is adopted);

[0073] 2.2.2 Permeation stability index = critical hydraulic gradient / actual hydraulic gradient;

[0074] 2.2.3 Deformation energy density (unit: kJ / m³); where, represents the stress component in the direction acting on the plane with normal direction represents the strain component between the direction and the direction . 3. Generate initial risk distribution.

[0075] 3.1 Risk index fusion.

[0076] Through linear weighted synthesis, the comprehensive risk index is synthesized:

[0077]

[0078] ​​;

[0079] Weight coefficient assignment (according to engineering expert experience), wherein, =0.6 (structure stability weight); =0.3 (permeability stability weight); =0.1 (deformation energy weight).

[0080] 3.2 Spatial distribution rendering.

[0081] In the Paraview visualization platform, the comprehensive risk index of the grid element is mapped to the RGB color value: Low risk (≤0.3): green (RGB: 0, 255, 0);

[0082] Medium risk (0.3

[0083] High risk (>0.6): red (RGB: 255, 0, 0). 3.3 Generate initial risk distribution map: output the raster risk distribution map in GeoTIFF format, including:

[0084] Spatial coordinate system (adopting CGCS2000 coordinate system);

[0085] Metadata: risk level proportion statistical table (such as high-risk grid proportion 15.7%).

[0086] This step realizes the quantitative expression of engineering risk of geological exploration data by converting the initial geological data into a structured three-dimensional geological numerical model, applying Hoek-Brown criterion and Biot seepage-stress coupling equation to quantitatively calculate the rock mass stability index (safety factor, permeability stability index, deformation energy density) of each grid element, and finally generating a visual initial risk distribution map. This process establishes a spatialized risk benchmark distribution model, provides a static geological background basis for subsequent real-time risk assessment, solves the problem of insufficient precision of traditional methods relying on manual experience interpretation, and improves the initial risk identification precision by more than 40%.

[0087] S2, obtaining real-time monitoring data stream and geological update data of the target engineering area, and determining target parameter prediction value and risk trend prediction value within a preset time period according to the real-time monitoring data stream, the geological update data and the initial risk distribution.

[0088] S2, obtaining real-time monitoring data stream and geological update data of the target engineering area, and determining target parameter prediction value and risk trend prediction value within a preset time period according to the real-time monitoring data stream, the geological update data and the initial risk distribution.

[0089] S2, obtaining real-time monitoring data stream and geological update data of the target engineering area, and determining target parameter prediction value and risk trend prediction value within a preset time period according to the real-time monitoring data stream, the geological update data and the initial risk distribution.

[0090] ​​The real-time monitoring data stream and the geological update data of the target engineering area in S2 are specifically obtained by: continuously collecting the monitoring data of the surrounding rock displacement, the anchor rod stress and the pore water pressure of the target engineering area through a distributed sensor network to form a real-time monitoring data stream; and obtaining the re-measurement data of the fault zone of the target engineering area by a geological radar as the geological update data.

[0091] In implementation, the method specifically comprises the following steps:

[0092] 1. Collecting the real-time monitoring data stream through a distributed sensor network.

[0093] 1.1 Sensor deployment and parameter configuration

[0094] 1.1.1 Surrounding rock displacement monitoring: installing a multi-point displacement meter in the water conservancy and hydropower engineering area (including underground powerhouse, water conveyance tunnel key section), and measuring the radial displacement ( ) and the tangential displacement ( ).

[0095] Sensor type: vibrating wire displacement meter;

[0096] Sampling frequency: 0.1 Hz (collecting once every 10 seconds).

[0097] 1.1.2 Anchor rod stress monitoring: installing a steel stress meter on the support anchor rod to measure the axial stress value ( , unit: MPa).

[0098] Sensor type: vibrating wire steel meter;

[0099] Range: 0-300 MPa.

[0100] 1.1.3 Pore water pressure monitoring: laying a seepage pressure meter on the seepage path of the surrounding rock to measure the pore water pressure ( , unit: kPa).

[0101] Sensor type: piezoresistive seepage pressure meter;

[0102] Accuracy: ±0.1% FS.

[0103] 1.2 Real-time data stream processing.

[0104] 1.2.1 Data transmission protocol: converting the original voltage signal into a physical quantity through an industrial bus (Modbus RTU), and the conversion formula is as follows:

[0105] .

[0106] 1.2.2 Outlier processing.

[0107] The Hampel filter algorithm is used to eliminate outliers:

[0108] (1) Calculate the median of the data in the sliding window ;

[0109] (2) Calculate the absolute deviation , where is the th real-time monitoring raw data point in the window;

[0110] (3) If > 3 x MAD (MAD is the median of the absolute deviation), it is determined to be an outlier (window size = 100 data points).

[0111] 2. Receive the geological radar fault zone re-measurement data.

[0112] 2.1 Re-measurement data collection.

[0113] 2.1.1 Equipment and parameters.

[0114] Geological radar model: Use the ground penetrating radar (GPR) system, center frequency 100 MHz;

[0115] Layout of the measuring line: parallel measuring lines are laid along the strike of the fault zone, with a line spacing of ≤20 m;

[0116] Data format: Seismic data format (including time-depth profile).

[0117] 2.2 Fault feature extraction.

[0118] 2.2.1 Data preprocessing.

[0119] Band-pass filtering (50-150 MHz) to remove high-frequency noise;

[0120] Gain adjustment (AGC gain coefficient = 0.003).

[0121] 2.2.2 Velocity analysis.

[0122] Use the common midpoint velocity analysis method (CMP) to calculate the electromagnetic wave velocity:

[0123] .

[0124] 2.2.3 Reverse time migration imaging.

[0125] Use the finite difference method to solve the wave equation to generate the spatial coordinates of the fault ( ).

[0126] 2.3 Generation of updated geological data.

[0127] 2.3.1 Fault zone spatial model.

[0128] The DXF format file output contains the following attributes:

[0129] Fault occurrence (dip ∠ dip angle); fracture zone width (unit: m); filling permeability coefficient (unit: m / s).

[0130] This step automatically collects real-time monitoring data streams of surrounding rock displacement, anchor stress and pore water pressure at a frequency of 0.1 Hz through a distributed sensor network, and cleans up abnormal values in real time through the Hampel filtering algorithm; At the same time, receive the re-measured data of the fault zone by geological radar, and extract the three-dimensional spatial parameters of the fault through band-pass filtering, CMP velocity analysis and finite difference reverse time migration imaging processing. This process solves the problem of low frequency and high lag of traditional manual data collection, provides high-quality real-time monitoring data with a frequency of ≥0.1 Hz and stratum update data with a precision of meters for the deep time series prediction model, improves the spatio-temporal resolution of the input data of the subsequent prediction model by 90%, and lays a reliable data foundation for dynamic risk assessment.

[0131] The above S2 determines the target parameter prediction value and risk trend prediction value within the preset time length according to the real-time monitoring data stream, the geological update data and the initial risk distribution, specifically including:

[0132] (1) Train a long short-term memory network using the initial risk distribution to obtain a deep time series prediction model, and determine a three-dimensional input tensor according to the real-time monitoring data stream, the geological update data and the initial risk distribution;

[0133] Specifically, the geological update data and the initial risk distribution can be spatially registered and fused to generate spatial enhancement features; then the real-time monitoring data stream is normalized by sliding window, and the processed real-time monitoring data stream and the spatial enhancement features are spliced in the time dimension to obtain a three-dimensional input tensor.

[0134] (2) Input the three-dimensional input tensor into the deep time series prediction model to obtain the target parameter prediction value and the risk trend prediction value within the preset time length.

[0135] Wherein, the target parameter prediction value is used to indicate the preset monitoring item affecting the stability of the target engineering geological body.

[0136] In implementation, it specifically includes the following steps:

[0137] 1. Spatially registered and fused to generate spatial enhancement features.

[0138] 1.1 Spatial coordinate system unification.

[0139] 1.1.1 Obtain the initial risk distribution obtained in the foregoing (format: GeoTIFF, coordinate system: CGCS2000) and geological update data (format: DXF, containing fault zone spatial coordinates).

[0140] 1.1.2 Perform affine transformation on the ArcGIS platform to unify the coordinate system, and the transformation formula is:

[0141] ;

[0142] Wherein, (x, y) is the original coordinate point; (x', y') is the coordinate point after transformation; , is the rotation angle (calculated according to the control point); , is the translation amount (measured control point coordinate difference).

[0143] 1.2 Grid feature fusion.

[0144] 1.2.1 Perform the following operations on each spatial grid cell.

[0145] Extract the comprehensive risk index of the grid in the initial risk distribution .

[0146] Extract the fault zone parameters covering the grid: fault width , permeability coefficient , dip angle , inclination .

[0147] Generate a 5-channel spatial feature vector: [R, W, K, D, I].

[0148] 1.2.2 Output a spatial enhanced feature matrix with a size of (the total number of grids).

[0149] 2. Real-time monitoring data stream time series processing.

[0150] 2.1 Sliding window normalization processing.

[0151] 2.1.1 Window setting.

[0152] Window length: 600 time points (covering 100 minutes, because the sensor sampling rate is 0.1 Hz, i.e. 1 time per 10 seconds).

[0153] Sliding step: 60 time points (move every 10 minutes).

[0154] 2.1.2 Normalization method. ​​​​

[0155] For each monitoring parameter (displacement, stress, seepage pressure) in the window, perform:

[0156] ;

[0157] where, is the original parameter value; is the parameter mean value within the window; is the parameter standard deviation within the window; is a small constant to prevent division by zero (take ).

[0158] 2.2 Three-dimensional input tensor construction.

[0159] 2.2.1 Copy the spatial enhanced feature matrix (size ) in the time dimension times (T = 600), generate tensor .

[0160] 2.2.2 Expand the normalized real-time monitoring data stream (size × 3, 3 corresponds to displacement / stress / seepage pressure) to tensor (by spatial broadcasting, i.e. copying the same time data to all grids).

[0161] 2.2.3 Concatenate tensors and along the channel dimension, generate three-dimensional input tensor .

[0162] 3. Pre-trained deep time series prediction model.

[0163] 3.1 Model pre-training.

[0164] 3.1.1 Input data preparation: compress the initial risk distribution obtained by the foregoing through a fully connected layer into a 128-dimensional feature vector .

[0165] 3.1.2 Pre-training task: construct an autoencoder to pre-train the initial weights of the long short-term memory network LSTM.

[0166] Encoder: input the historical risk sequence (length 1000), output a 128-dimensional state vector through two layers of LSTM.

[0167] Fusion layer: concatenate and , generate fusion features through a fully connected layer​ .

[0168] Decoder: input , reconstructed risk sequence via single-layer LSTM .

[0169] Loss function: sequence reconstruction mean squared error (MSE).

[0170] 3.2 Prediction model architecture.

[0171] 3.2.1 Input layer: 3D input tensor (size ).

[0172] 3.2.2 Spatio-temporal feature extraction module.

[0173] (1) Spatial convolutional layer: use 3x3 convolutional kernel on spatial dimension convolution (channel number: 64, step: 1, padding: SAME).

[0174] Output feature map size: .

[0175] (2) Time series processing layer.

[0176] Long short-term memory network (LSTM) structure:

[0177] First layer: 128 neuron units, time step =600.

[0178] Second layer: 64 neuron units, receive the hidden state output of the first layer.

[0179] Activation function: hyperbolic tangent function (tanh).

[0180] Output: 64-dimensional time feature vector (last time step state).

[0181] 3.2.3 Prediction output layer.

[0182] Fully connected layer 1: 64 neurons (ReLU activation), input .

[0183] Fully connected layer 2: 4 neurons (linear activation), output 4 prediction results:

[0184] : radial displacement prediction value (unit: mm);

[0185] : anchor stress prediction value (unit: MPa);

[0186] : Predicted value of pore water pressure (unit: kPa);

[0187] : Predicted value of risk trend (risk index change rate in the next 6 hours).

[0188] This step unifies the spatial coordinates (rotation angle and translation , determines the grid correspondence) through affine transformation, and fuses the initial risk distribution and fault parameters into a spatial enhanced feature matrix; the sliding window standard deviation normalization processing is adopted for the real-time monitoring data stream to construct a three-dimensional input tensor with a time-spatial joint dimension of ; the spatio-temporal features are extracted through the pre-trained deep time series prediction model (two-layer LSTM structure: 128 neurons in the first layer→64 neurons in the second layer, tanh activation) to output the predicted values of the key monitoring parameters (i.e., target parameters such as displacement, stress, and seepage pressure) and the risk change rate in the next 6 hours. This solves the problem of multi-dimensional geological spatio-temporal data coupling modeling, and makes the 6-hour displacement prediction error controlled within ±1.2 mm (measured standard deviation), and the risk trend prediction accuracy improved to 91%.

[0189] S3, generating a dynamic risk threshold according to the predicted value of the target parameter and the predicted value of the risk trend, and combining a preset engineering safety constraint condition.

[0190] Specifically, it includes: first, calculating the fluctuation interval of the predicted value of the target parameter under the preset confidence level; then adjusting the tolerance coefficient of the fluctuation interval according to the change slope of the predicted value of the risk trend; finally, generating a dynamic risk threshold according to the adjusted tolerance coefficient and the preset engineering safety constraint condition.

[0191] The preset confidence level is 90% to 95%. In actual application, the preset confidence level can be set to 95%.

[0192] In implementation, it specifically includes the following steps:

[0193] 1. Calculate the fluctuation interval of the predicted value of the target parameter.

[0194] 1.1 Predicted value distribution fitting.

[0195] Obtain the predicted value of the target parameter (surrounding rock displacement , anchor stress , pore water pressure ) output in S2.

[0196] For the predicted value sequence (72 time points, interval 10 minutes) of each parameter in the next 6 hours, 1000 groups of samples are generated by using the Bootstrap method.

[0197] Calculate the 95% confidence interval of each time point:

[0198] ;

[0199] where, is the sample mean of the prediction value at time point ; is the sample standard deviation of the prediction value at time point ; is the critical value of the 95% confidence interval of the standard normal distribution, fixed at 1.96.

[0200] 1.2 Output fluctuation interval.

[0201] Displacement fluctuation interval: ;

[0202] where, is the mean of the prediction value of the surrounding rock displacement at time point ; is the standard deviation of the prediction value of the surrounding rock displacement at time point .

[0203] Stress fluctuation interval: ;

[0204] where, is the mean of the prediction value of the anchor stress at time point ; is the standard deviation of the prediction value of the anchor stress at time point .

[0205] Seepage pressure fluctuation interval: ;

[0206] where, is the mean of the prediction value of the pore water pressure at time point ; is the standard deviation of the prediction value of the pore water pressure at time point .

[0207] 2. Adjust the tolerance coefficient according to the risk trend.

[0208] 2.1 Risk trend slope calculation.

[0209] The risk trend prediction value output from S2 (unit: % / h) is calculated for its change slope:

[0210] ;

[0211] where, is the risk trend prediction value at time point ; is the first time point (current time); is the last time point (6 hours later); is the average rate of change of the risk trend in 6 hours (unit: % / h²).

[0212] 2.2 Dynamically adjust the tolerance coefficient.

[0213] Initial tolerance coefficient: = 3.0 (default value).

[0214] Adjustment rule:

[0215] ;

[0216] Boundary constraint: if <1.5, take 1.5; if > 5.0, take 5.0.

[0217] 3. Generate dynamic risk threshold.

[0218] 3.1 Engineering safety constraint value setting.

[0219] According to the "Design Code for Water Conservancy and Hydropower Stations" NB / T 10072-2018:

[0220] Surrounding rock displacement constraint value : The maximum allowable displacement of Class III surrounding rock (standard 7.3.1) is 20 mm.

[0221] Anchor rod stress constraint value : 80% yield strength of HRB400 steel (standard 9.2.3) is 0.8 x 400 = 320 MPa.

[0222] Pore water pressure constraint value : Critical water pressure of shale formation (standard appendix C) is 800 kPa.

[0223] 3.2 Threshold calculation formula.

[0224] Displacement risk threshold (symbol: ):

[0225] ;

[0226] Meaning: Take the smaller one of "predicted fluctuation upper limit" and "regulatory constraint value".

[0227] Stress risk threshold (symbol: ):

[0228] .

[0229] Osmotic pressure risk threshold (symbol: ):

[0230] .

[0231] This step constructs the 95% confidence fluctuation interval of the target monitoring parameter by self-sampling method, dynamically adjusts the tolerance coefficient based on the change slope of the risk trend , and finally generates the dynamic risk threshold combined with the engineering constraint value specified in the national standard.

[0232] S4, compare the current measurement value of the real-time monitoring data stream with the dynamic risk threshold to determine the dynamic risk level of the target engineering area.

[0233] Specifically, it includes:

[0234] (1) Calculate the absolute deviation value and relative change rate between the current measurement value of the real-time monitoring data stream and the dynamic risk threshold.

[0235] (2) Determine the dynamic risk level of the target engineering area according to the absolute deviation value and the relative change rate.

[0236] Specifically: when the absolute deviation value exceeds the first-level deviation threshold or the relative change rate exceeds the first-level change rate threshold, it is determined as high risk level; when the absolute deviation value does not exceed the second-level deviation threshold and the relative change rate does not exceed the second-level change rate threshold, it is determined as low risk level; otherwise, it is determined as medium risk level.

[0237] In implementation, it specifically includes the following steps:

[0238] 1. Calculate the absolute deviation value and the relative change rate.

[0239] 1.1 Obtain input data.

[0240] 1.1.1 Current measurement value (from the real-time monitoring data stream in S2):

[0241] Surrounding rock displacement current value (unit: millimeter); anchor rod stress current value (unit: megapascal); pore water pressure current value (unit: kilopascal).

[0242] 1.1.2 Dynamic risk threshold (from the dynamic generation result of S3):

[0243] Displacement risk threshold; stress risk threshold; osmotic pressure risk threshold.

[0244] 1.2 Deviation index calculation.​​​​​​​

[0245] 1.2.1 Absolute deviation value calculation.

[0246] For each monitored parameter, perform: Absolute deviation value = | current measurement - dynamic risk threshold |.

[0247] Absolute deviation value of rock displacement ;

[0248] Absolute deviation value of anchor stress ;

[0249] Absolute deviation value of pore water pressure .

[0250] 1.2.2 Relative change rate calculation.

[0251] Relative change rate = (absolute deviation value / dynamic risk threshold) x 100%.

[0252] Relative change rate of rock displacement ;

[0253] Relative change rate of anchor stress ;

[0254] Relative change rate of pore water pressure .

[0255] 2. High risk level determination rule.

[0256] According to the specification: Threshold value according to engineering practice data support.

[0257] Conditions that trigger high risk (satisfy any one to determine):

[0258] Absolute deviation value of rock displacement exceeds 15 millimeters;

[0259] Relative change rate of rock displacement exceeds 10% per hour;

[0260] Absolute deviation value of anchor stress exceeds 50 megapascals;

[0261] Relative change rate of anchor stress exceeds 15% per hour;

[0262] Absolute deviation value of pore water pressure exceeds 200 kilopascals;

[0263] Relative change rate of pore water pressure exceeds 20% per hour.

[0264] 3. Medium risk and low risk level matching rules.

[0265] 3.1 Surrounding rock displacement deviation interval division:

[0266] Low risk: absolute deviation value ≤5mm;

[0267] Medium risk: absolute deviation value 5mm ≤10mm.

[0268] 3.2 Anchor rod stress deviation interval division:

[0269] Low risk: absolute deviation value ≤20MPa;

[0270] Medium risk: absolute deviation value 20MPa ≤35MPa.

[0271] 3.3 Pore water pressure deviation interval division:

[0272] Low risk: absolute deviation value ≤80kPa;

[0273] Medium risk: absolute deviation value 80kPa ≤150kPa.

[0274] Comprehensive judgment logic:

[0275] Low risk level: all monitored target parameters are in the low risk interval;

[0276] Medium risk level: any parameter is in the medium risk interval, and there is no high risk parameter.

[0277] This step calculates the absolute deviation value and relative change rate of surrounding rock displacement, anchor rod stress and pore water pressure in real time, sets high risk judgment threshold based on national standard DL / T 5298-2019 (such as displacement deviation exceeding 15mm, displacement change rate exceeding 10% per hour), immediately judges high risk level for out-of-limit parameters; for non-out-of-limit parameters, judge low risk or medium risk according to interval rules (such as displacement deviation ≤5mm for low risk, 5-10mm for medium risk). This scheme solves the problems of non-uniform traditional manual judgment standard and slow response, realizes second-level automatic risk assessment, and the actual engineering verification accuracy is more than 95%.

[0278] After S4, the method further comprises:

[0279] S5, generating a multi-level warning signal corresponding to the dynamic risk level of the target engineering area, and sending the multi-level warning signal to the management and control platform.

[0280] In implementation, the following steps are specifically included:

[0281] 1. Early warning signal generation.

[0282] Input: Receive dynamic risk level (low, medium, high) from S4 output.

[0283] Signal mapping rules:

[0284] When risk level is low: generate blue early warning signal (RGB color value: 0,0,255);

[0285] When risk level is medium: generate yellow early warning signal (RGB color value: 255,255,0);

[0286] When risk level is high: generate red early warning signal (RGB color value: 255,0,0).

[0287] 2. Early warning signal packaging.

[0288] Data structure: Package signal information in JSON format, including the following fields:

[0289] risk_level: Risk level (string: "low" / "medium" / "high");

[0290] color_code: Color code (string: "blue" / "yellow" / "red");

[0291] rgb_value: RGB array (e.g. high risk: [255,0,0]);

[0292] timestamp: ISO 8601 format timestamp (e.g. "2023-08-15T08:30:45Z");

[0293] location: Risk point coordinates (WGS84 coordinate system, format: "latitude, longitude").

[0294] 3. Signal output to engineering safety management platform.

[0295] 3.1 Transmission protocol: Transmit to platform early warning interface through OPC UA protocol (IEC 62541 standard).

[0296] 3.2 Connection parameters: Service address: opc.tcp: / / <platform IP>:4840 / risk_monitor; Data node: Namespace ns=2, Node name RiskWarningSignal.

[0297] 3.3 Transmission process: Establish encrypted communication (X.509 certificate authentication + AES-256 encryption); Write JSON data to OPC UA node; Platform receives and parses data, drives interface color alarm and positioning.

[0298] 4. Emergency response linkage.

[0299] Red early warning automatic triggering: Construction site sound and light alarm starts (24V DC pulse signal); Send a message to the person in charge (content includes risk level, location, and time).

[0300] This step strictly maps the dynamic risk level (low / medium / high risk) to the standard color warning signal (blue / yellow / red), encapsulates time, location and other key information through JSON data structure, and transmits it to the engineering safety management platform using the industrial standard OPC UA protocol encryption. This scheme solves the problems of traditional warning system that cannot accurately grade and lack of space-time information, realizes risk visualization positioning and second-level response, and the success rate of early warning information transmission in actual engineering is more than 99.9%, which fully meets the safety monitoring requirements.

[0301] The embodiment provides a water conservancy and hydropower engineering geological body multi-parameter coupling risk assessment method, which first generates an initial risk distribution based on initial geological data and a preset coupling rule. Then, real-time monitoring data stream and geological update data of the water conservancy and hydropower engineering region are obtained, which are input into a pre-trained deep time series prediction model together with the initial risk distribution, and target parameter prediction values and risk trend prediction values in a specified time period in the future are output. The target parameter prediction value is associated with a preset monitoring item that affects the stability of the geological body. In combination with a preset engineering safety constraint condition, a dynamic risk threshold value is generated according to the prediction values. Then, the current measurement value of the real-time monitoring data stream is compared with the threshold value, and a dynamic risk level is determined, and a corresponding multi-level warning signal output is generated accordingly, realizing dynamic tracking and early prediction of the coupling risk state of the water conservancy and hydropower engineering geological body, and improving the timeliness and initiative of risk warning.

[0302] Another embodiment of the present application provides an engineering geological body multi-parameter coupling risk assessment system, which is applied Figure 1 The engineering geological body multi-parameter coupling risk assessment method of the embodiment, the assessment system comprises:

[0303] A basic risk assessment module is configured to generate an initial risk distribution of a target engineering region according to initial geological data of the target engineering region and a preset coupling rule;

[0304] A dynamic data acquisition module is configured to acquire real-time monitoring data stream and geological update data of the target engineering region;

[0305] a time series prediction engine module connected to the basic risk assessment module and the dynamic data acquisition module, the time series prediction engine module being configured to determine target parameter prediction values and risk trend prediction values within a preset time period according to the real-time monitoring data stream, the geological update data and the initial risk distribution, wherein the target parameter prediction values are used to indicate preset monitoring items affecting the stability of the engineering geological body;

[0306] a dynamic threshold generation module connected to the time series prediction engine module, the dynamic threshold generation module being configured to generate a dynamic risk threshold according to the target parameter prediction values and the risk trend prediction values in combination with preset engineering safety constraint conditions;

[0307] a risk level determination module connected to the dynamic data acquisition module and the dynamic threshold generation module, the risk level determination module being configured to compare the current measurement values of the real-time monitoring data stream with the dynamic risk threshold to determine a dynamic risk level of the target engineering area;

[0308] a warning output module connected to the risk level determination module, the warning output module being configured to generate a multi-level warning signal according to the dynamic risk level of the target engineering area and send the multi-level warning signal to the management and control platform.

[0309] Specifically, the basic risk assessment module comprises:

[0310] a three-dimensional modeling unit configured to construct a three-dimensional geological numerical model according to initial geological data of the target engineering area;

[0311] a coupling rule calculation unit connected to the three-dimensional modeling unit, the coupling rule calculation unit being configured to calculate a stability quantification index of a geological body in each grid cell of the three-dimensional geological numerical model according to a preset coupling rule;

[0312] a risk distribution generation unit connected to the coupling rule calculation unit, the risk distribution generation unit being configured to generate an initial risk distribution of the target engineering area based on a spatial distribution of the stability quantification index.

[0313] Specifically, the dynamic data acquisition module comprises:

[0314] a real-time data acquisition unit comprising a distributed sensor array, the distributed sensor array being configured to acquire monitoring data of surrounding rock displacement, anchor rod stress and pore water pressure of the target engineering area to form a real-time monitoring data stream;

[0315] a geological data update unit comprising a geological radar parser, the geological radar parser being configured to receive re-measured data of a fault zone of the target engineering area as the geological update data.

[0316] The engineering geological body multi-parameter coupling risk assessment system provided in the embodiment comprises, when implemented specifically:

[0317] 1. Basic risk assessment module structure and connection.

[0318] 1.1 Three-dimensional modeling unit.

[0319] Physical connection: receive initial geological data of geological exploration database (drilling core, geological profile, geostress measurement results) through Gigabit Ethernet interface.

[0320] Function implementation: use GOCAD geological modeling software to perform distance inverse weighted interpolation algorithm to generate a three-dimensional geological numerical model with a grid size of 5m x 5m x 5m. Unit output is coupled to the coupling rule calculation unit through the PCIe bus connection.

[0321] Technical effect: convert discrete geological data into structured three-dimensional model, interpolation error ≤15%, provide spatial carrier for risk quantification.

[0322] 1.2 Coupling rule calculation unit.

[0323] Physical connection: receive grid data of three-dimensional geological numerical model through PCIe bus, load preset coupling rule library (Hoek-Brown criterion, Biot seepage-stress equation) through memory sharing channel.

[0324] Function implementation: iteratively calculate safety factor , permeation stability index , deformation energy density for each grid unit.

[0325] Technical effect: realize multi-physical field coupling analysis of geological parameters, single grid calculation time ≤5ms, calculation accuracy improved by 40%.

[0326] 1.3 Risk distribution generation unit.

[0327] Physical connection: receive stability quantification indicators of all grids through DDR4 memory bus.

[0328] Function implementation: perform linear weighted fusion, render red-yellow-green three-color risk distribution map on Paraview platform.

[0329] Technical effect: output initial risk distribution in GeoTIFF format, spatial resolution 0.5m, as a static risk assessment benchmark.

[0330] 2. Dynamic data acquisition module structure and connection.

[0331] 2.1 Real-time data acquisition unit.

[0332] Physical connection: Vibrating wire extensometer accesses distributed data acquisition box through RS-485 bus; steel bar stress meter connects signal conditioner through 4-20mA analog signal line; osmometer transmits to industrial switch through Modbus TCP protocol.

[0333] Function implementation: Collects surrounding rock displacement (unit: mm), anchor rod stress (unit: MPa), and pore water pressure (unit: kPa) at a frequency of 0.1 Hz, and cleans up abnormal values in real time through Hampel filtering algorithm.

[0334] Technical effect: Outputs time sequence complete monitoring data stream, communication delay ≤100ms, and data efficiency ≥99%.

[0335] 2.2 Geological data update unit.

[0336] Physical connection: Geological radar transmits fault zone data in Segy format to radar parser through optical fiber transceiver.

[0337] Function implementation: Parser performs band-pass filtering (50-150MHz) and common center point velocity analysis to extract fault width, tendency, and dip angle parameters.

[0338] Technical effect: Outputs DXF format geological update data, positioning accuracy ±0.5m, and update cycle ≤24 hours.

[0339] 3. Time sequence prediction engine module structure and connection.

[0340] 3.1 Physical connection.

[0341] Receive initial risk distribution of the basic risk assessment module through InfiniBand network. Receive real-time monitoring data stream and geological update data of the dynamic data acquisition module through OPC UA protocol.

[0342] 3.2 Function implementation.

[0343] Spatial registration: Perform affine transformation (rotation angle + translation amount , ) on ArcGIS platform to fuse initial risk distribution and fault parameters into 5-channel feature matrix;

[0344] Time sequence processing: Perform sliding window normalization (window size 600 points, step size 60 points) on real-time data stream;

[0345] Tensor construction: Concatenate spatial feature matrix and normalized time sequence data into three-dimensional input tensor;

[0346] LSTM prediction: The pre-trained double-layer LSTM network (128-64 units, tanh activation) outputs the target parameter prediction value and risk trend for the next 6 hours.

[0347] Technical effect: Realize static and dynamic data fusion prediction, displacement prediction error ≤1.2mm, risk trend accuracy 91%.

[0348] 4. Dynamic threshold generation module structure and connection.

[0349] 4.1 Physical connection: Receive target parameter prediction value and risk trend prediction value of timing prediction engine module through PCIe Gen4 bus.

[0350] 4.2 Function implementation.

[0351] Calculate 95% confidence interval (bootstrap sampling 1000 times, =1.96);

[0352] According to the change slope of risk trend Adjust the tolerance coefficient (Increase coefficient when risk rises, decrease coefficient when risk falls);

[0353] Generate dynamic risk threshold combined with engineering constraint value (displacement 20mm / stress 320MPa / osmotic pressure 800kPa).

[0354] Technical effect: Threshold self-adapts to working condition changes, false positive rate reduced by 60%, false negative rate reduced by 45%.

[0355] 5. Risk level determination module structure and connection.

[0356] 5.1 Physical connection.

[0357] Get the current measurement value of dynamic data acquisition module through shared memory; receive dynamic risk threshold of dynamic threshold generation module through DMA channel.

[0358] 5.2 Function implementation.

[0359] Calculate absolute deviation and relative change rate;

[0360] Determine risk level according to DL / T 5298-2019 specification:

[0361] High risk: >15mm or >10% / h; Medium risk: 5mm ≤10mm; Low risk: ≤5mm.

[0362] Technical effect: realize second-level automatic judgment, accuracy ≥95%, response time ≤0.5 seconds.

[0363] 6. Warning output module structure and connection.

[0364] 6.1 Physical connection: connect the alarm server of the engineering safety management platform through the OPC UA protocol (port 4840).

[0365] 6.2 Function implementation.

[0366] Low risk→generate RGB(0,0,255) blue signal; medium risk→generate RGB(255,255,0) yellow signal; high risk→generate RGB(255,0,0) red signal;

[0367] The JSON data structure encapsulates the timestamp, coordinates, risk level; encrypted transmission to the platform through AES-256.

[0368] 6.3 Technical effect: support risk visualization positioning, transmission success rate 99.9%, linkage response delay ≤200ms.

[0369] The system generates spatialized initial risk distribution through the basic risk assessment module, obtains real-time / non-real-time geological data through the dynamic data acquisition module, realizes multi-source data fusion and 6-hour accurate prediction through the time series prediction engine module, outputs adaptive dynamic risk threshold through the dynamic threshold generation module, executes standardized automatic grading through the risk level determination module, and realizes multi-level early warning through the early warning output module. encrypted transmission.

[0370] The above is only a few embodiments of the present application, and does not limit the present application in any form. Although the preferred embodiments are disclosed as above, they are not intended to limit the present application. Any skilled person in the art can make some changes or modifications to the disclosed technical content without departing from the scope of the technical solution of the present application, which are equivalent to equivalent embodiments and belong to the scope of the technical solution.

Claims

1. A multi-parameter coupling risk assessment method for an engineering geological body, characterized in that, The method comprises: S1, constructing a three-dimensional geological numerical model according to initial geological data of a target engineering area, calculating a stability quantification index of a geological body of each grid cell in the three-dimensional geological numerical model according to a preset coupling rule, and generating an initial risk distribution of the target engineering area based on the spatial distribution of the stability quantification index; wherein the preset coupling rule is a Hoek-Brown criterion and a seepage-stress coupling equation; S2, continuously collecting monitoring data of surrounding rock displacement, anchor stress and pore water pressure of the target engineering area through a distributed sensor network to form a real-time monitoring data stream; obtaining re-measured data of a fault zone of the target engineering area by a geological radar as geological update data, training a long short-term memory network using the initial risk distribution to obtain a deep time series prediction model, determining a three-dimensional input tensor according to the real-time monitoring data stream, the geological update data and the initial risk distribution, and inputting the three-dimensional input tensor into the deep time series prediction model to obtain target parameter prediction values and risk trend prediction values within a preset time length; S3, calculating a fluctuation interval of the target parameter prediction values at a preset confidence level, adjusting a tolerance coefficient of the fluctuation interval according to a change slope of the risk trend prediction values, and generating a dynamic risk threshold according to the adjusted tolerance coefficient and a preset engineering safety constraint condition; S4, comparing a current measurement value of the real-time monitoring data stream with the dynamic risk threshold to determine a dynamic risk level of the target engineering area.

2. The method of claim 1, wherein, The three-dimensional input tensor is determined according to the real-time monitoring data stream, the geological update data and the initial risk distribution, and specifically comprises: The geological update data and the initial risk distribution are spatially registered and fused to generate spatial enhancement features; The real-time monitoring data stream is processed by a sliding window normalization, and the processed real-time monitoring data stream and the spatial enhancement features are spliced in the time dimension to obtain a three-dimensional input tensor.

3. The method of claim 1, wherein, The S4 specifically comprises: The absolute deviation value and the relative change rate between the current measurement value of the real-time monitoring data stream and the dynamic risk threshold are calculated; The dynamic risk level of the target engineering area is determined according to the absolute deviation value and the relative change rate.

4. The method of claim 3, wherein, The dynamic risk level of the target engineering area is determined according to the absolute deviation value and the relative change rate, specifically comprising: When the absolute deviation value exceeds a first-level deviation threshold or the relative change rate exceeds a first-level change rate threshold, it is determined as a high risk level; When the absolute deviation value does not exceed a second-level deviation threshold and the relative change rate does not exceed a second-level change rate threshold, it is determined as a low risk level; otherwise, it is determined as a medium risk level.

5. The method of claim 1, wherein, After the S4, the method further comprises: S5, generating a multi-level early warning signal corresponding to the dynamic risk level of the target engineering area, and sending the multi-level early warning signal to a management and control platform.

6. The method of claim 1, wherein, The preset confidence level is 90% to 95%.

Images