Pre-warning method and system for rock and soil mass rupture

Abstract

The present invention relates to the field of rock and soil mass fracture early warning, and particularly to a method and system for early warning of rock and soil mass fracture. The early warning system for rock and soil mass fracture includes: a static geological data acquisition module, a representative monitoring parameter selection module, and a rock and soil mass monitoring module. When conducting daily monitoring, the present invention only monitors a single monitoring parameter of the rock and soil mass, effectively avoiding the overuse of monitoring equipment in the traditional comprehensive monitoring mode, significantly reducing equipment wear and maintenance costs; at the same time, reducing the processing and analysis of a large amount of redundant data, saving valuable computing resources. Especially when the state of the rock and soil mass is stable, the optimization of this monitoring method makes the resource utilization more reasonable, avoiding the waste of computing power and improving the overall monitoring efficiency.

Description

Technical Field

[0001] The present invention relates to the field of rock and soil mass rupture warning, and particularly to a pre-warning method and system for rock and soil mass rupture. Background Art

[0002] The warning technology for rock and soil mass rupture has always relied on the combined judgment of fixed monitoring parameters and static thresholds, but there are significant bottlenecks in its applicability in complex geological environments; existing systems usually preset a set of general parameters and determine a unified alarm threshold based on historical experience; however, the rupture mechanisms of different rock and soil masses vary greatly. For example, the rupture in granite areas may take the microseismic signal intensity as a key indicator, while the instability of clay layers is often triggered by a sudden drop in pore water pressure; if the surrounding rock pressure is continuously monitored in a sandstone slope with developed joints while ignoring the cumulative value of the principal strain, the system may miss the best warning window due to parameter selection mismatch; conversely, for gravelly soil strata with strong permeability, if the monitoring weight of the seepage velocity is not adjusted according to the hydrogeological conditions, the instantaneous pore water pressure fluctuation caused by heavy rainfall may be misjudged as a rupture precursor; this "one-size-fits-all" parameter and threshold setting mode essentially separates geological characteristics from parameter sensitivity, resulting in an increase in both false alarm rate and missed alarm rate.

[0003] The contradiction between resource efficiency and prediction accuracy further limits the practicality of existing technologies; most systems adopt a continuous full-volume data acquisition strategy, that is, high-frequency sampling of more than a dozen parameters such as displacement, seepage, and acoustic emission at the same time, resulting in a large proportion of invalid data in the total transmitted data; taking the monitoring of a certain mine slope as an example, among the large amount of data generated by the system daily, only surface displacement and temperature have practical monitoring significance during the stable period of the rock mass, and the continuous acquisition of the remaining parameters not only increases the server load but may also mask real abnormal signals due to noise interference; this rigid design not only wastes resources but also directly weakens the generalization ability of the system in different scenarios.

[0004] The above defects highlight the core contradiction that the current technology urgently needs to break through: how to establish a parameter selection mechanism adaptable to geological characteristics and construct a matching abnormal determination model without increasing hardware costs; based on the above technical gap, this patent proposes a pre-warning method and system for rock and soil mass rupture, which significantly improves the warning accuracy and robustness in complex geological scenarios while ensuring resource efficiency. Summary of the Invention

[0005] During daily monitoring, the present invention only monitors a single monitoring parameter of the rock and soil mass, effectively avoiding the overuse of monitoring equipment in the traditional comprehensive monitoring mode, significantly reducing equipment wear and maintenance costs; at the same time, reducing the processing and analysis of a large amount of redundant data, saving valuable computing resources. Especially when the state of the rock and soil mass is stable, this optimization of the monitoring method makes the resource utilization more reasonable, avoids the waste of computing power, and improves the overall monitoring efficiency.

[0006] A method for early warning of rock and soil mass rupture, comprising:

[0007] Obtain a static geological data set of the rock and soil mass area, including rock type, compressive strength, shear strength, permeability coefficient, joint density, and historical failure mode;

[0008] Select representative monitoring parameters from the dynamic monitoring parameter set based on the static geological data set. The dynamic monitoring parameter set includes surface displacement rate, acoustic emission event frequency, pore water pressure, principal strain value, synthetic displacement amount, acoustic emission energy release rate, seepage velocity, surrounding rock pressure, and rock mass temperature change rate; among them, the surface displacement rate, acoustic emission event frequency, pore water pressure, and principal strain value are all key monitoring parameters;

[0009] Collect representative monitoring parameters at any monitoring time point to obtain representative monitoring data; determine whether the obtained representative monitoring data is abnormal. If so, execute the comprehensive monitoring mode, that is, collect all monitoring parameters in the dynamic monitoring parameter set at any monitoring time point to obtain the dynamic monitoring data set, and sort the dynamic monitoring data sets obtained at the nearest several monitoring time points in time to form the rock and soil mass monitoring time series set; use the obtained rock and soil mass monitoring time series set as the input of the rupture prediction model, output the rupture prediction result of the rock and soil mass area. If the rupture prediction result is higher than the preset rupture probability threshold, execute the warning measure; if not, do nothing.

[0010] Preferably, selecting representative monitoring parameters from the dynamic monitoring parameter set based on the static geological data set specifically includes the following steps:

[0011] Step 1: Obtain several rock and soil mass samples that have experienced rupture events, and calculate the abnormal scores of each key monitoring parameter in each rock and soil mass sample;

[0012] Step 2: Traverse each geotechnical sample. If the abnormal score of the surface displacement rate in the key monitoring parameters of the geotechnical sample is the highest, then take this geotechnical sample as the first screening sample; if the abnormal score of the acoustic emission event frequency in the key monitoring parameters of the geotechnical sample is the highest, then take this geotechnical sample as the second screening sample; if the abnormal score of the pore water pressure in the key monitoring parameters of the geotechnical sample is the highest, then take this geotechnical sample as the third screening sample; if the abnormal score of the principal strain value in the key monitoring parameters of the geotechnical sample is the highest, then take this geotechnical sample as the fourth screening sample;

[0013] Step 3: Use the first screening sample, the second screening sample, the third screening sample, and the fourth screening sample to obtain the general patterns of the geotechnical body corresponding to the surface displacement rate, the acoustic emission event frequency, the pore water pressure, and the principal strain value respectively;

[0014] Step 4: Calculate the cumulative deviation rate between the static geological data of the current geotechnical body area and each general pattern of the geotechnical body, and take the key monitoring parameter corresponding to the general pattern of the geotechnical body with the lowest cumulative deviation rate as the representative monitoring parameter of the current geotechnical body area.

[0015] Preferably, calculate the abnormal score of each key monitoring parameter in each geotechnical sample. The specific operation is as follows:

[0016] For any geotechnical sample, based on the key monitoring data obtained at the previous , = 1, 2,..., ; = 1, 2, 3, 4; to successively represent the surface displacement rate, the acoustic emission event frequency, the pore water pressure, and the principal strain value;

[0017] For any key monitoring data , based on the preset limit threshold , use the formula to calculate the first score of the key monitoring data ; use the formula to calculate and obtain the second score of the key monitoring data , where is the standard deviation of the key monitoring data ; use the formula to calculate and obtain the third score of the key monitoring data ;

[0018] Finally, use the formula Calculate the anomaly scores of surface displacement rate, AE event frequency, pore water pressure, and principal strain value in sequence 。

[0019] Preferably, use the first screening sample, the second screening sample, the third screening sample, and the fourth screening sample to obtain the general patterns of the rock and soil mass corresponding to the surface displacement rate, AE event frequency, pore water pressure, and principal strain value respectively. The specific operations are as follows:

[0020] Obtain the static geological data set of all the first screening samples, and take the most rock types 、average compressive strength 、average shear strength 、average permeability coefficient 、average joint density and the most historical failure modes as the general pattern of the rock and soil mass corresponding to the surface displacement rate , i = 1, 2, …, 6; For the second screening sample, the third screening sample, and the fourth screening sample, perform the same operations as the first screening sample, and obtain the general patterns of the rock and soil mass corresponding to the AE event frequency, pore water pressure, and principal strain value in sequence 、 and 。

[0021] Preferably, calculate the cumulative deviation rate of the static geological data of the current rock and soil mass area from each general pattern of the rock and soil mass. The specific operations are as follows:

[0022] Based on the static geological data set of the current rock and soil mass area and each general pattern of the rock and soil mass , to corresponding to the surface displacement rate, AE event frequency, pore water pressure, and principal strain value in sequence;

[0023] Use the formula to calculate and obtain the cumulative deviation rate of the static geological data set of the current rock and soil mass area from each general pattern of the rock and soil mass 。

[0024] Preferably, the fracture prediction model is established based on the LSTM model, including: an input layer, an LSTM time series feature extraction layer, an attention mechanism layer, a feature fusion fully connected layer, and an output normalization layer.

[0025] Preferably, for the training of the fracture prediction model, the specific operations are as follows:

[0026] Obtain a number of model training samples with marked fracture prediction results. Each model training sample contains a historical geotechnical monitoring time series set of a geotechnical body area. Divide all model training samples into a training set and a validation set. Use the training set to train a fracture prediction model with initialized parameters. Subsequently, input the validation set into the fracture prediction model for validation to obtain a validation result. Set a first training condition, and determine whether the obtained validation result meets the first training condition. If so, output the trained fracture prediction model. If not, continue to train the fracture prediction model using the training set.

[0027] A pre-warning system for geotechnical body fracture, comprising:

[0028] A static geological data acquisition module for acquiring a static geological data set of a geotechnical body area;

[0029] A representative monitoring parameter selection module for selecting representative monitoring parameters of the current geotechnical body area from a dynamic monitoring parameter set by using the acquired static geological data set;

[0030] A geotechnical body monitoring module, including a mode discrimination unit and a fracture prediction unit; the mode discrimination unit is used to obtain representative monitoring data at any monitoring time point, and determine whether the obtained representative monitoring data is abnormal. If so, execute a comprehensive monitoring mode; if not, do nothing; the fracture prediction unit is used to obtain a geotechnical body monitoring time series set and use it as the input of the fracture prediction model, and output the fracture prediction result of the geotechnical body area.

[0031] The present invention has the following advantages:

[0032] 1. During daily monitoring, the present invention only monitors a single monitoring parameter of the geotechnical body, effectively avoiding the overuse of monitoring equipment in the traditional comprehensive monitoring mode, significantly reducing equipment loss and maintenance costs; at the same time, reducing the processing and analysis of a large amount of redundant data, saving valuable computing resources. Especially when the geotechnical body is in a stable state, the optimization of this monitoring method makes the resource utilization more reasonable, avoiding the waste of computing power and improving the overall monitoring efficiency.

[0033] 2. After the present invention discovers that the representative monitoring data is abnormal, by collecting all monitoring parameters and using an advanced fracture prediction model for analysis, it can accurately capture the long-term dependence relationship of the geotechnical body monitoring data, and dynamically allocate weights through an attention mechanism to further improve the accuracy of feature extraction; after feature fusion and normalization processing, the model can output a high-precision geotechnical body fracture probability value, thereby realizing accurate early warning of geotechnical body fracture, providing strong technical support for disaster prevention and mitigation work, and effectively protecting the safety of personnel and property. Description of the Drawings

[0034] Figure 1The structural schematic diagram of the pre-warning system for rock and soil mass rupture adopted in the embodiment of the present invention. Specific implementation mode

[0035] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0036] Embodiment 1, a pre-warning method for rock and soil mass rupture, including:

[0037] Obtain the static geological dataset of the geotechnical body area, including rock type, compressive strength, shear strength, permeability coefficient, joint density, and historical failure mode. These data are important descriptions of the basic characteristics of the geotechnical body. Rock type is one of the key parameters in the static geological dataset, which directly determines the mineral composition, structure, and mechanical properties of the geotechnical body. In actual operation, geological personnel determine the types of rocks in the geotechnical body through on-site exploration and laboratory analysis, such as granite, sandstone, claystone, etc., and convert the rock type into a numerical code according to the pre-set classification standard for computer processing and model recognition. Compressive strength reflects the maximum pressure that a rock can withstand when subjected to compressive force and is a core indicator for evaluating the stability of the geotechnical body. Geological engineers usually use standard rock compressive strength test methods, such as those specified by the International Organization for Standardization or the American Society for Testing and Materials, to test the collected rock samples in the laboratory to obtain accurate compressive strength values. Shear strength represents the ability of a rock to resist shear failure and is of great significance for analyzing the stability of geotechnical bodies in projects such as slopes and underground cavities. Similarly, the shear strength of rock samples can be determined through methods such as direct shear tests or triaxial shear tests in the laboratory, and these data will provide key information for the early warning model on the ability of the geotechnical body to resist sliding and collapse. The permeability coefficient describes the ability of a rock to allow water flow through, and this parameter is crucial for understanding the mechanical behavior and stability changes of the geotechnical body under the action of water. Using the laboratory permeability test device and following the relevant standard test procedures, conduct permeability tests on rock samples to accurately determine their permeability coefficient, providing data support for subsequent analysis of the stability of the geotechnical body under water pressure. Joint density refers to the number of joints (fractures) in a rock per unit volume or unit area, which directly affects the overall stability and mechanical properties of the geotechnical body. Geological surveyors record the distribution of joints in the geotechnical body through on-site observation and measurement, including parameters such as joint spacing, length, and width, and then calculate the joint density based on these data and numericalize it for use in the early warning model. Historical failure mode is a summary and induction of the past failure forms of the geotechnical body, such as landslides, collapses, creep, etc. This information helps to identify the potential weaknesses and easily damaged areas of the geotechnical body. By consulting historical geological data, on-site investigation, and interviews with local residents in the area where the geotechnical body is located, collect records of historical failure events of the geotechnical body and classify and code different types of failure modes to form numerical data, providing important references for the early warning model on the failure law of the geotechnical body.

[0038] Select representative monitoring parameters of the current geotechnical region from the dynamic monitoring parameter set based on the acquired static geological data set. This process aims to select the dynamic monitoring parameters that can best reflect the stability changes of the geotechnical body according to its specific geological characteristics, thereby improving the efficiency of early warning. The dynamic monitoring parameter set includes surface displacement rate, frequency of acoustic emission events, pore water pressure, principal strain value, synthetic displacement amount, acoustic emission energy release rate, seepage velocity, surrounding rock pressure, and rock mass temperature change rate. Among them, surface displacement rate, frequency of acoustic emission events, pore water pressure, and principal strain value are all key monitoring parameters. The selection and application of these key monitoring parameters are based on a large number of experimental studies and engineering practice experiences. They can capture the abnormal changes of the geotechnical body before rupture from different angles, providing rich and accurate data support for the early warning model, so as to achieve early and accurate early warning of geotechnical body rupture.

[0039] Collect representative monitoring parameters at any monitoring time point to obtain representative monitoring data. Judge whether the obtained representative monitoring data is abnormal based on the limit threshold corresponding to the representative monitoring parameters. If so, execute the comprehensive monitoring mode; if not, do nothing. The advantage of this monitoring mechanism is that it can ensure the timeliness of early warning while avoiding unnecessary resource waste.

[0040] The specific content of the comprehensive monitoring mode is as follows: Collect all monitoring parameters in the dynamic monitoring parameter set at any monitoring time point to obtain the dynamic monitoring data set, and sort the dynamic monitoring data sets obtained at the recent several monitoring time points according to time to form the geotechnical body monitoring time series set. Use the obtained geotechnical body monitoring time series set as the input of the rupture prediction model, and output the rupture prediction result of the geotechnical region. If the rupture prediction result is higher than the preset rupture probability threshold, execute the early warning measure.

[0041] If the dynamic monitoring data sets obtained at three consecutive monitoring time points in the comprehensive monitoring mode are all normal and the rupture prediction result is lower than the rupture probability threshold, stop executing the comprehensive monitoring mode. When the system is in the comprehensive monitoring mode, it will continuously and closely monitor various parameters of the geotechnical body. If, at three consecutive monitoring time points, all monitoring parameters fluctuate within the normal range, and at the same time, the rupture prediction result output by the rupture prediction model also continuously remains lower than the preset rupture probability threshold, this indicates that the stability of the geotechnical body has been restored and the risk of rupture has been significantly reduced. In this case, it is no longer necessary to continue executing the comprehensive monitoring mode, because it will not only increase the wear of monitoring equipment and the burden of data processing, but also may lead to the fatigue of monitoring personnel and the unreasonable allocation of resources. Therefore, the system will automatically stop the comprehensive monitoring mode and return to the daily state of only monitoring representative monitoring parameters. This can optimize the use of monitoring resources, improve the efficiency and sustainability of the entire early warning system while ensuring safety.

[0042] Select representative monitoring parameters of the current geotechnical area from the dynamic monitoring parameter set based on the acquired static geological data set. The specific steps are as follows:

[0043] Step 1: Obtain several geotechnical samples that have experienced rupture events, and calculate the anomaly scores of each key monitoring parameter in each geotechnical sample;

[0044] Step 2: Traverse each geotechnical sample. If the anomaly score of the surface displacement rate is the highest among the key monitoring parameters of the geotechnical sample, then take this geotechnical sample as the first screening sample; if the anomaly score of the acoustic emission event frequency is the highest among the key monitoring parameters of the geotechnical sample, then take this geotechnical sample as the second screening sample; if the anomaly score of the pore water pressure is the highest among the key monitoring parameters of the geotechnical sample, then take this geotechnical sample as the third screening sample; if the anomaly score of the principal strain value is the highest among the key monitoring parameters of the geotechnical sample, then take this geotechnical sample as the fourth screening sample;

[0045] Step 3: Use the first screening sample, the second screening sample, the third screening sample, and the fourth screening sample to obtain the general patterns of geotechnical bodies corresponding to the surface displacement rate, acoustic emission event frequency, pore water pressure, and principal strain value respectively;

[0046] Step 4: Calculate the cumulative deviation rate between the static geological data of the current geotechnical area and each general pattern of the geotechnical body, and take the key monitoring parameter corresponding to the general pattern of the geotechnical body with the lowest cumulative deviation rate as the representative monitoring parameter of the current geotechnical area.

[0047] Calculate the anomaly scores of each key monitoring parameter in each geotechnical sample. The specific operation is as follows:

[0048] For any geotechnical sample, based on the key monitoring data obtained at the previous monitoring time points before rupture t = 1, 2, …, n; where n

[0049] represents the number of monitoring time points before rupture; and for any key monitoring data x i at time point t i and based on a preset limit threshold x 0 is the standard deviation of the key monitoring data ; Using the formula calculate and obtain the third score of the key monitoring data ; ;

[0050] Finally, use the formula to calculate and obtain the anomaly scores of the surface displacement rate, acoustic emission event frequency, pore water pressure, and principal strain value in sequence .

[0051] Respectively use the first screening sample, the second screening sample, the third screening sample, and the fourth screening sample to obtain the general patterns of the rock and soil mass corresponding to the surface displacement rate, acoustic emission event frequency, pore water pressure, and principal strain value. The specific operations are as follows:

[0052] Obtain the static geological data set of all the first screening samples, and take the most rock types , average compressive strength , average shear strength , average permeability coefficient , average joint density and the most historical failure modes as the general pattern of the rock and soil mass corresponding to the surface displacement rate , i = 1, 2,..., 6; For the second screening sample, the third screening sample, and the fourth screening sample, perform the same operations as the first screening sample to obtain the general patterns of the rock and soil mass corresponding to the acoustic emission event frequency, pore water pressure, and principal strain value in sequence , and .

[0053] Calculate the cumulative deviation rate of the static geological data of the current rock and soil mass area from each general pattern of the rock and soil mass. The specific operations are as follows:

[0054] Based on the static geological data set of the current rock and soil mass area and each general pattern of the rock and soil mass , to corresponding to the surface displacement rate, acoustic emission event frequency, pore water pressure, and principal strain value in sequence;

[0055] Use the formula to calculate and obtain the cumulative deviation rate of the static geological data set of the current rock and soil mass area from each general pattern of the rock and soil mass .

[0056] The fracture prediction model is established based on the LSTM model and includes: an input layer, an LSTM time series feature extraction layer, an attention mechanism layer, a feature fusion fully connected layer, and an output normalization layer;

[0057] The input layer is used to receive the geotechnical monitoring time series set and implement time window standardization and multi-dimensional feature normalization;

[0058] The LSTM time series feature extraction layer is used to capture the long-term dependence relationship of the geotechnical monitoring time series set and output a time series feature vector;

[0059] The attention mechanism layer is used to dynamically allocate weights and output a weighted time series feature vector;

[0060] The feature fusion fully connected layer is used to further extract features from the weighted time series feature vector and output a final feature vector;

[0061] The output normalization layer is used to convert the final feature vector into a fracture probability value through the Sigmoid function.

[0062] For the training of the fracture prediction model, the specific operations are as follows:

[0063] Obtain a number of model training samples with labeled fracture prediction results. Each model training sample contains the historical geotechnical monitoring time series set of a geotechnical area; divide all model training samples into a training set and a validation set, use the training set to train the fracture prediction model with initialized parameters, and then input the validation set into the fracture prediction model for verification to obtain a verification result; set a first training condition, and judge whether the obtained verification result meets the first training condition. If so, output the trained fracture prediction model; if not, continue to use the training set to train the fracture prediction model.

[0064] Example 2, a pre-warning system for geotechnical fracture, as Figure 1 shown, includes:

[0065] A static geological data acquisition module, used to acquire the static geological data set of the geotechnical area, including rock type, compressive strength, shear strength, permeability coefficient, joint density, and historical failure mode; among them, both the rock type and the historical failure mode are represented by numerical codes;

[0066] A representative monitoring parameter selection module, used to select the representative monitoring parameters of the current geotechnical area from the dynamic monitoring parameter set by using the acquired static geological data set. The dynamic monitoring parameter set includes surface displacement rate, acoustic emission event frequency, pore water pressure, principal strain value, composite displacement amount, acoustic emission energy release rate, seepage velocity, surrounding rock pressure, and rock mass temperature change rate; among them, the surface displacement rate, acoustic emission event frequency, pore water pressure, and principal strain value are all key monitoring parameters;

[0067] The geotechnical body monitoring module includes a pattern discrimination unit and a fracture prediction unit; the pattern discrimination unit is used to collect representative monitoring parameters at any monitoring time point to obtain representative monitoring data; based on the limit threshold corresponding to the representative monitoring parameters, it is judged whether the obtained representative monitoring data is abnormal. If so, the comprehensive monitoring mode is executed; if not, no operation is performed; the fracture prediction unit is used to collect all monitoring parameters in the dynamic monitoring parameter set at any monitoring time point after executing the comprehensive monitoring mode to obtain a dynamic monitoring data set, and sort the dynamic monitoring data sets obtained at the most recent several monitoring time points according to time to form a geotechnical body monitoring time series set; use the obtained geotechnical body monitoring time series set as the input of the fracture prediction model to output the fracture prediction result of the geotechnical body area. If the fracture prediction result is higher than the preset fracture probability threshold, warning measures are executed; if the dynamic monitoring data sets obtained at three consecutive monitoring time points in the comprehensive monitoring mode are all normal and the fracture prediction result is lower than the fracture probability threshold, the comprehensive monitoring mode is cancelled.

[0068] It should be understood that those of ordinary skill in the art can make improvements or transformations according to the above description, and all such improvements and transformations should fall within the protection scope of the appended claims of the present invention. The parts not described in detail in this specification belong to the prior art well-known to those of ordinary skill in the art.

Claims

1. A pre-warning method for the rupture of rock and soil masses, characterized in that, Including: Obtain the static geological data set of the geotechnical body area, including rock type, compressive strength, shear strength, permeability coefficient, joint density, and historical failure mode; Select representative monitoring parameters from the dynamic monitoring parameter set based on the static geological data set. The dynamic monitoring parameter set includes surface displacement rate, frequency of acoustic emission events, pore water pressure, principal strain value, synthetic displacement amount, acoustic emission energy release rate, seepage velocity, surrounding rock pressure, and rock mass temperature change rate. Among them, the surface displacement rate, frequency of acoustic emission events, pore water pressure, and principal strain value are all key monitoring parameters; Collect representative monitoring parameters at any monitoring time point to obtain representative monitoring data. Determine whether the obtained representative monitoring data is abnormal. If so, execute the comprehensive monitoring mode, that is, collect all monitoring parameters in the dynamic monitoring parameter set at any monitoring time point to obtain the dynamic monitoring data set, and sort the dynamic monitoring data sets obtained at the recent several monitoring time points according to time to form the geotechnical body monitoring time series set. Use the obtained geotechnical body monitoring time series set as the input of the fracture prediction model, output the fracture prediction result of the geotechnical body area. If the fracture prediction result is higher than the preset fracture probability threshold, execute the warning measure; if not, do nothing.

2. The pre-warning method for rock and soil mass rupture according to claim 1, characterized in that Select representative monitoring parameters from the dynamic monitoring parameter set based on the static geological data set, specifically including the following steps: Step 1: Obtain several geotechnical body samples that have experienced fracture events, and calculate the anomaly scores of each key monitoring parameter in each geotechnical body sample; Step 2: Traverse each geotechnical body sample. If the anomaly score of the surface displacement rate in the key monitoring parameters of the geotechnical body sample is the highest, then use this geotechnical body sample as the first screening sample; if the anomaly score of the frequency of acoustic emission events in the key monitoring parameters of the geotechnical body sample is the highest, then use this geotechnical body sample as the second screening sample; if the anomaly score of the pore water pressure in the key monitoring parameters of the geotechnical body sample is the highest, then use this geotechnical body sample as the third screening sample; if the anomaly score of the principal strain value in the key monitoring parameters of the geotechnical body sample is the highest, then use this geotechnical body sample as the fourth screening sample; Step 3: Use the first screening sample, the second screening sample, the third screening sample, and the fourth screening sample to obtain the general patterns of the geotechnical body corresponding to the surface displacement rate, frequency of acoustic emission events, pore water pressure, and principal strain value respectively; Step 4: Calculate the cumulative deviation rate between the static geological data of the current geotechnical body area and each geotechnical body general pattern, and use the key monitoring parameter corresponding to the geotechnical body general pattern with the lowest cumulative deviation rate as the representative monitoring parameter of the current geotechnical body area.

3. The pre-warning method for rock and soil mass rupture according to claim 2, characterized in that, Calculate the anomaly scores of each key monitoring parameter in each geotechnical body sample. The specific operation is as follows: For any geotechnical sample, based on the key monitoring data obtained at monitoring time points before rupture , where \(i = 1, 2, \cdots\), ; \(j = 1, 2, 3, 4\); to successively represent the surface displacement rate, the frequency of acoustic emission events, the pore water pressure, and the principal strain value; For any key monitoring data , based on a preset limit threshold , use the formula to calculate the first score of the key monitoring data ; use the formula to calculate and obtain the second score of the key monitoring data , where is the standard deviation of the key monitoring data ; use the formula to calculate and obtain the third score of the key monitoring data ; Finally, use the formula to calculate the anomaly scores of the surface displacement rate, acoustic emission event frequency, pore water pressure, and principal strain value in sequence .

4. The pre-warning method for rock and soil mass rupture according to claim 3, characterized in that, Use the first screening sample, the second screening sample, the third screening sample, and the fourth screening sample to obtain the general patterns of the geotechnical body corresponding to the surface displacement rate, frequency of acoustic emission events, pore water pressure, and principal strain value respectively. The specific operation is as follows: Obtain the static geological datasets of all the first screened samples, and take the most rock types , average compressive strength , average shear strength , average permeability coefficient , average joint density and the most historical failure modes as the general patterns of rock and soil bodies corresponding to the surface displacement rate , = 1, 2, …, 6; For the second screening sample, the third screening sample, and the fourth screening sample, perform the same operations as for the first screening sample to sequentially obtain the general patterns of the rock and soil mass corresponding to the acoustic emission event frequency, pore water pressure, and principal strain value , and .

5. The pre-warning method for the rupture of rock and soil masses according to claim 4, characterized in that Calculate the cumulative deviation rate between the static geological data of the current geotechnical body area and each geotechnical body general pattern. The specific operation is as follows: Based on the static geological data set of the current rock and soil mass area and the general patterns of each rock and soil mass , to correspond to the surface displacement rate, the frequency of acoustic emission events, the pore water pressure, and the principal strain value in sequence; Using the formula successively calculate to obtain the static geological data set of the current geotechnical area and the cumulative deviation rate of each general geotechnical pattern . .

6. The pre-warning method for rock and soil mass rupture according to claim 5, characterized in that The fracture prediction model is established based on the LSTM model and includes: an input layer, an LSTM time series feature extraction layer, an attention mechanism layer, a feature fusion fully connected layer, and an output normalization layer.

7. The pre-warning method for rock and soil mass rupture according to claim 6, characterized in that For the training of the fracture prediction model, the specific operations are as follows: Obtain a number of model training samples with labeled fracture prediction results. Each model training sample contains a historical geotechnical monitoring time series set of a geotechnical body area; divide all model training samples into a training set and a validation set, use the training set to train the fracture prediction model with initialized parameters, and then input the validation set into the fracture prediction model for verification to obtain verification results; Set the first training condition, and judge whether the obtained verification result meets the first training condition. If so, output the trained fracture prediction model; if not, continue to train the fracture prediction model using the training set.

8. Pre-warning system for rock and soil mass rupture, characterized in that, The system is applied to the pre-warning method for geotechnical body fracture described in any one of claims 1-7 above and includes: A static geological data acquisition module for acquiring a static geological data set of a geotechnical body area; A representative monitoring parameter selection module for selecting representative monitoring parameters of the current geotechnical body area from the dynamic monitoring parameter set by using the acquired static geological data set; A geotechnical body monitoring module, including a pattern discrimination unit and a fracture prediction unit; the pattern discrimination unit is used to obtain representative monitoring data at any monitoring time point, judge whether the obtained representative monitoring data is abnormal, and if so, execute the comprehensive monitoring mode; if not, do nothing; the fracture prediction unit is used to obtain the geotechnical body monitoring time series set and use it as the input of the fracture prediction model, and output the fracture prediction result of the geotechnical body area.

Images