Pre-warning method and system for rock-soil body fracture

Abstract

The invention relates to the field of rock-soil body fracture early warning, in particular to a pre-warning method and system for rock-soil body fracture. The pre-warning system for rock-soil body fracture comprises a static geological data acquisition module, a representative monitoring parameter selection module and a rock-soil body monitoring module. According to the invention, only a single monitoring parameter of a rock-soil body is monitored during daily monitoring, so that overuse of monitoring equipment in a traditional comprehensive monitoring mode is effectively avoided, and equipment loss and maintenance cost are remarkably reduced; meanwhile, processing and analysis of a large amount of redundant data are reduced, precious computing resources are saved, especially when the state of the rock-soil body is stable, resource utilization is more reasonable through optimization of the monitoring mode, waste of computing power is avoided, and the overall monitoring efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of rock and soil body fracture early warning, and in particular to a pre-warning method and system for rock and soil body fracture. Background Art

[0002] Early warning technology for rock and soil rupture has always relied on a combination of fixed monitoring parameters and static thresholds, but its applicability in complex geological environments has significant bottlenecks. Existing systems usually pre-set a set of general parameters and determine a unified alarm threshold based on historical experience. However, the rupture mechanisms of different rock and soil bodies vary greatly. For example, the rupture of granite areas may be based on the intensity of microseismic signals, 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 sandstone slopes with developed joints and the principal strain accumulation value is ignored, the system may miss the optimal warning window due to parameter selection mismatch. Conversely, for gravel layers with high permeability, if the monitoring weight of the seepage velocity is not adjusted according to hydrological conditions, the instantaneous pore water pressure fluctuations caused by heavy rainfall may be misjudged as a precursor to rupture. This "one-size-fits-all" parameter and threshold setting mode essentially separates geological characteristics from parameter sensitivity, resulting in a simultaneous increase in 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-data collection strategy, that is, high-frequency sampling of more than ten parameters such as displacement, seepage, and acoustic emission at the same time, resulting in a large proportion of invalid data in the total transmission. Taking the slope monitoring of a mine as an example, among the large amount of data generated by the system on a daily basis, only the surface displacement and temperature during the rock mass stability period have practical monitoring significance. The continuous collection of other parameters not only increases the server load, but may also cover up the real abnormal signals due to noise interference. This rigid design not only wastes resources, but also directly weakens the system's generalization ability in different scenarios.

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

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

[0006] The pre-warning method of rock and soil rupture includes: Obtain static geological data sets for geotechnical areas, including rock type, compressive strength, shear strength, permeability, joint density, and historical failure modes; Based on the static geological data set, representative monitoring parameters are selected from the dynamic monitoring parameter set, which includes surface displacement rate, acoustic emission event frequency, pore water pressure, principal strain value, synthetic displacement, acoustic emission energy release rate, seepage velocity, surrounding rock pressure and rock temperature change rate; among them, surface displacement rate, acoustic emission event frequency, pore water pressure and principal strain value are all key monitoring parameters; At any monitoring time point, representative monitoring parameters are collected to obtain representative monitoring data; it is determined whether the obtained representative monitoring data has any anomalies. If so, a comprehensive monitoring mode is executed, that is, all monitoring parameters in the dynamic monitoring parameter set are collected at any monitoring time point to obtain a dynamic monitoring data set, and the dynamic monitoring data sets obtained at the most recent monitoring time points are sorted in time to form a rock and soil monitoring time series set; the obtained rock and soil monitoring time series set is used as the input of the rupture prediction model, and the rupture prediction result of the rock and soil area is output. If the rupture prediction result is higher than the preset rupture probability threshold, early warning measures are executed; if not, no operation is performed.

[0007] Preferably, selecting representative monitoring parameters from a dynamic monitoring parameter set based on a static geological data set specifically includes the following steps: Step 1: Obtain several rock and soil samples that have experienced rupture events, and calculate the abnormality scores of each key monitoring parameter in each rock and soil sample; Step 2: Traverse each rock and soil sample. If the abnormal score of the surface displacement rate among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used as the first screening sample; if the abnormal score of the acoustic emission event frequency among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used as the second screening sample; if the abnormal score of the pore water pressure among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used as the third screening sample; if the abnormal score of the principal strain value among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used 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 mode of the rock and soil corresponding to the surface displacement rate, the frequency of acoustic emission events, the pore water pressure and the principal strain value; Step 4: Calculate the cumulative deviation rate between the static geological data of the current rock and soil area and the universal patterns of each rock and soil body, and use the key monitoring parameters corresponding to the universal pattern of the rock and soil body with the lowest cumulative deviation rate as the representative monitoring parameters of the current rock and soil area.

[0008] Preferably, the abnormality score of each key monitoring parameter in each rock and soil sample is calculated, and the specific operation is as follows: For any rock sample, based on its Key monitoring data obtained at each monitoring time point , =1, 2, …, ; =1, 2, 3, 4; to It represents the surface displacement rate, acoustic emission event frequency, pore water pressure and principal strain value in turn; For any key monitoring data , based on preset limit thresholds , using the formula Calculate key monitoring data First Rating ; Using the formula Calculate and obtain key monitoring data Second rating of ,in, For key monitoring data The standard deviation of Calculate and obtain key monitoring data The third rating ; Finally, using the formula The anomaly scores of surface displacement rate, acoustic emission event frequency, pore water pressure and principal strain value are calculated in sequence. .

[0009] Preferably, the first screening sample, the second screening sample, the third screening sample and the fourth screening sample are used to obtain the general mode of the rock and soil corresponding to the surface displacement rate, the frequency of acoustic emission events, the pore water pressure and the principal strain value, respectively. The specific operation is as follows: Obtain a static geological dataset of all first screening samples and separate the most common rock types , average compressive strength , average shear strength , average permeability , average joint density and the most historical destruction mode Universal mode of rock mass as a function of surface displacement rate , =1, 2, ..., 6; for the second, third and fourth screening samples, perform the same operation as the first screening sample to obtain the universal mode of the rock and soil corresponding to the acoustic emission event frequency, pore water pressure and principal strain value in turn , and .

[0010] Preferably, the accumulated deviation rate between the static geological data of the current rock and soil area and the universal mode of each rock and soil body is calculated, and the specific operation is as follows: Based on a static geological dataset of the current geotechnical area and the general patterns of various rock and soil masses , to They correspond to the surface displacement rate, the frequency of acoustic emission events, the pore water pressure and the principal strain value, respectively; Using the formula Calculate and obtain the static geological data set of the current rock and soil area in sequence Common patterns with various rock and soil masses The cumulative deviation rate .

[0011] Preferably, the rupture prediction model is established based on an 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.

[0012] Preferably, the specific operations for training the rupture prediction model are as follows: A number of model training samples with annotated rupture prediction results are obtained, each of which contains a historical rock and soil monitoring time series set of a rock and soil area; all model training samples are divided into a training set and a validation set, and the rupture prediction model with initialized parameters is trained using the training set, and then the validation set is input into the rupture prediction model for verification to obtain the verification result; a first training condition is set to determine whether the obtained verification result meets the first training condition, and if so, the trained rupture prediction model is output; if not, the rupture prediction model is continuously trained using the training set.

[0013] The early warning system for rock and soil rupture includes: Static geological data acquisition module, used to obtain static geological data sets of rock and soil area; A representative monitoring parameter selection module is used to select representative monitoring parameters of the current rock and soil area from the dynamic monitoring parameter set using the acquired static geological data set; The rock and soil monitoring module includes a pattern discrimination unit and a rupture prediction unit; the pattern discrimination unit is used to obtain representative monitoring data at any monitoring time point, and determine whether the obtained representative monitoring data has an abnormality. If so, a comprehensive monitoring mode is executed; if not, no operation is performed; the rupture prediction unit is used to obtain the rock and soil monitoring time series set and use it as the input of the rupture prediction model, and output the rupture prediction result of the rock and soil area.

[0014] The present invention has the following advantages: 1. The present invention only monitors a single monitoring parameter of the rock and soil mass during daily monitoring, which effectively avoids excessive use of monitoring equipment in the traditional comprehensive monitoring mode and significantly reduces equipment loss and maintenance costs; at the same time, it reduces the processing and analysis of a large amount of redundant data and saves valuable computing resources. Especially when the state of the rock and soil mass is stable, the optimization of this monitoring method makes resource utilization more reasonable, avoids waste of computing power, and improves overall monitoring efficiency.

[0015] 2. After the present invention finds that the representative monitoring data is abnormal, it can accurately capture the long-term dependencies of the rock and soil monitoring data by collecting all the monitoring parameters and using the advanced rupture prediction model for analysis, and dynamically allocate weights through the attention mechanism to further improve the accuracy of feature extraction; after feature fusion and normalization, the model can output high-precision rock and soil rupture probability values, thereby realizing accurate early warning of rock and soil rupture, providing strong technical support for disaster prevention and mitigation work, and effectively protecting the safety of people and property. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a schematic structural diagram of a pre-warning system for rock and soil mass rupture used in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to enable persons 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.

[0018] Embodiment 1, a pre-warning method for rock and soil mass fracture, comprising: Obtain static geological data sets of the rock mass area, including rock type, compressive strength, shear strength, permeability, joint density and historical failure mode. These data are important descriptions of the basic characteristics of the rock mass. Rock type is one of the key parameters in the static geological data set, which directly determines the mineral composition, structure and mechanical properties of the rock mass. In actual operation, geologists determine the types of rocks in the rock mass through on-site investigation 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. The compressive strength reflects the rock's ability to withstand pressure. The maximum pressure that a rock can withstand when subjected to shrinkage is the core indicator for evaluating the stability of a rock mass. 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 indicates the ability of rock to resist shear failure, which is of great significance for analyzing the stability of rock mass in projects such as slopes and underground caverns. Similarly, the shear strength of rock samples can be determined through laboratory direct shear tests or triaxial shear tests. These data will provide early warning models with information about the rock mass's resistance to sliding and Key information on collapse capacity; Permeability coefficient describes the ability of rock to allow water to flow through, and this parameter is crucial to understanding the mechanical behavior and stability changes of rock and soil under the action of water; Using the laboratory's permeability test equipment, in accordance with relevant standard test procedures, rock samples are subjected to permeability tests to accurately determine their permeability coefficients, providing data support for subsequent analysis of the stability of rock and soil under water pressure; Joint density refers to the number of joints (cracks) in rock per unit volume or unit area, which directly affects the overall stability and mechanical properties of rock and soil; Geological survey personnel record the distribution of joints in rock and soil through on-site observation and measurement, Including parameters such as the spacing, length, and width of joints, and then the joint density is calculated based on these data and digitized for use in the early warning model; historical damage patterns are a summary and induction of the forms of damage that have occurred in the rock and soil in the past, such as landslides, collapses, creep, etc. This information helps to identify potential weaknesses and vulnerable areas of the rock and soil; through the review of historical geological data in the area where the rock and soil are located, on-site investigations, and interviews with local residents, the records of historical damage events of the rock and soil are collected, and different types of damage patterns are classified and coded to form numerical data, which provides an important reference for the early warning model on the laws of rock and soil damage; Select representative monitoring parameters of the current geotechnical body area from the dynamic monitoring parameter set based on the acquired static geological data set. This process aims to select, according to the specific geological characteristics of the geotechnical body, the dynamic monitoring parameters that can best reflect its stability changes, so as to improve 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, composite 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, provide rich and accurate data support for the early warning model, and thus achieve early and accurate early warning of the rupture of the geotechnical body. 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. 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 body area. If the rupture prediction result is higher than the preset rupture probability threshold, execute the early warning measure. 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 the 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.

[0019] Based on the acquired static geological data set, representative monitoring parameters of the current rock and soil area are selected from the dynamic monitoring parameter set, which specifically includes the following steps: Step 1: Obtain several rock and soil samples that have experienced rupture events, and calculate the abnormality scores of each key monitoring parameter in each rock and soil sample; Step 2: Traverse each rock and soil sample. If the abnormal score of the surface displacement rate among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used as the first screening sample; if the abnormal score of the acoustic emission event frequency among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used as the second screening sample; if the abnormal score of the pore water pressure among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used as the third screening sample; if the abnormal score of the principal strain value among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used 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 mode of the rock and soil corresponding to the surface displacement rate, the frequency of acoustic emission events, the pore water pressure and the principal strain value; Step 4: Calculate the cumulative deviation rate between the static geological data of the current rock and soil area and the universal patterns of each rock and soil body, and use the key monitoring parameters corresponding to the universal pattern of the rock and soil body with the lowest cumulative deviation rate as the representative monitoring parameters of the current rock and soil area.

[0020] Calculate the abnormal score of each key monitoring parameter in each rock and soil sample. The specific operation is as follows: For any rock sample, based on its Key monitoring data obtained at each monitoring time point , =1, 2, …, ; =1, 2, 3, 4; to It represents the surface displacement rate, acoustic emission event frequency, pore water pressure and principal strain value in turn; For any key monitoring data , based on preset limit thresholds , using the formula Calculate key monitoring data First Rating ; Using the formula Calculate and obtain key monitoring data Second rating of ,in, For key monitoring data The standard deviation of Calculate and obtain key monitoring data The third rating ; Finally, using the formula The anomaly scores of surface displacement rate, acoustic emission event frequency, pore water pressure and principal strain value are calculated in sequence. .

[0021] The first screening sample, the second screening sample, the third screening sample and the fourth screening sample are used to obtain the general mode of the rock and soil corresponding to the surface displacement rate, the frequency of acoustic emission events, the pore water pressure and the principal strain value. The specific operations are as follows: Obtain a static geological dataset of all first screening samples and separate the most common rock types , average compressive strength , average shear strength , average permeability , average joint density and the most historical destruction mode Universal mode of rock mass as a function of surface displacement rate , =1, 2, ..., 6; for the second, third and fourth screening samples, perform the same operation as the first screening sample to obtain the universal mode of the rock and soil corresponding to the acoustic emission event frequency, pore water pressure and principal strain value in turn , and .

[0022] Calculate the cumulative deviation rate between the static geological data of the current rock and soil area and the general mode of each rock and soil body. The specific operation is as follows: Based on a static geological dataset of the current geotechnical area and the general patterns of various rock and soil masses , to They correspond to the surface displacement rate, the frequency of acoustic emission events, the pore water pressure and the principal strain value, respectively; Using the formula Calculate and obtain the static geological data set of the current rock and soil area in sequence Common patterns with various rock and soil masses The cumulative deviation rate .

[0023] The rupture prediction model is established based on the LSTM model, including: input layer, LSTM time series feature extraction layer, attention mechanism layer, feature fusion fully connected layer and output normalization layer; The input layer is used to receive the time series set of rock and soil monitoring and realize the time window standardization and multi-dimensional feature normalization; The LSTM time series feature extraction layer is used to capture the long-term dependencies of the rock and soil monitoring time series set and output the time series feature vector; The attention mechanism layer is used to dynamically assign weights and output weighted time series feature vectors; The feature fusion fully connected layer is used to further extract features from the weighted time series feature vector and output the final feature vector; The output normalization layer is used to convert the final feature vector into a rupture probability value through a Sigmoid function.

[0024] The specific operations for training the rupture prediction model are as follows: A number of model training samples with annotated rupture prediction results are obtained, each of which contains a historical rock and soil monitoring time series set of a rock and soil area; all model training samples are divided into a training set and a validation set, and the rupture prediction model with initialized parameters is trained using the training set, and then the validation set is input into the rupture prediction model for verification to obtain the verification result; a first training condition is set to determine whether the obtained verification result meets the first training condition, and if so, the trained rupture prediction model is output; if not, the rupture prediction model is continuously trained using the training set.

[0025] Embodiment 2, a pre-warning system for rock and soil fracture, such as Figure 1 As shown, including: Static geological data acquisition module, used to obtain static geological data sets of rock and soil area, including rock type, compressive strength, shear strength, permeability, joint density and historical failure mode; rock type and historical failure mode are both represented by numerical codes; The representative monitoring parameter selection module is used to select representative monitoring parameters of the current rock mass area from the dynamic monitoring parameter set 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, synthetic displacement, acoustic emission energy release rate, seepage velocity, surrounding rock pressure and rock mass temperature change rate; among them, surface displacement rate, acoustic emission event frequency, pore water pressure and principal strain value are all key monitoring parameters; The rock and soil monitoring module includes a mode discrimination unit and a rupture prediction unit; the mode 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 parameter, it is judged whether the obtained representative monitoring data has an abnormality. If so, the comprehensive monitoring mode is executed; if not, no operation is performed; the rupture 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, obtain the dynamic monitoring data set, and sort the dynamic monitoring data sets obtained at the most recent monitoring time points in time to form a rock and soil monitoring time series set; the obtained rock and soil monitoring time series set is used as the input of the rupture prediction model, and the rupture prediction result of the rock and soil area is output. If the rupture prediction result is higher than the preset rupture probability threshold, the early warning measures are executed; if there is no abnormality in the dynamic monitoring data sets obtained at three consecutive monitoring time points in the comprehensive monitoring mode and the rupture prediction result is lower than the rupture probability threshold, the comprehensive monitoring mode is cancelled.

[0026] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A pre-warning method for rock and soil mass rupture, characterized in that: include: Obtain static geological data sets for geotechnical areas, including rock type, compressive strength, shear strength, permeability, joint density, and historical failure modes; Based on the static geological data set, representative monitoring parameters are selected from the dynamic monitoring parameter set, which includes surface displacement rate, acoustic emission event frequency, pore water pressure, principal strain value, synthetic displacement, acoustic emission energy release rate, seepage velocity, surrounding rock pressure and rock temperature change rate; among them, surface displacement rate, acoustic emission event frequency, pore water pressure and principal strain value are all key monitoring parameters; At any monitoring time point, representative monitoring parameters are collected to obtain representative monitoring data; it is determined whether the obtained representative monitoring data has any anomalies. If so, a comprehensive monitoring mode is executed, that is, all monitoring parameters in the dynamic monitoring parameter set are collected at any monitoring time point to obtain a dynamic monitoring data set, and the dynamic monitoring data sets obtained at the most recent monitoring time points are sorted in time to form a rock and soil monitoring time series set; the obtained rock and soil monitoring time series set is used as the input of the rupture prediction model, and the rupture prediction result of the rock and soil area is output. If the rupture prediction result is higher than the preset rupture probability threshold, early warning measures are executed; if not, no operation is performed.

2. The method for early warning of rock and soil mass rupture according to claim 1, characterized in that: Selecting representative monitoring parameters from the dynamic monitoring parameter set based on the static geological data set specifically includes the following steps: Step 1: Obtain several rock and soil samples that have experienced rupture events, and calculate the abnormality scores of each key monitoring parameter in each rock and soil sample; Step 2: Traverse each rock and soil sample. If the abnormal score of the surface displacement rate among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used as the first screening sample; if the abnormal score of the acoustic emission event frequency among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used as the second screening sample; if the abnormal score of the pore water pressure among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used as the third screening sample; if the abnormal score of the principal strain value among the key monitoring parameters of the rock and soil sample is the highest, then the rock and soil sample is used 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 mode of the rock and soil corresponding to the surface displacement rate, the frequency of acoustic emission events, the pore water pressure and the principal strain value; Step 4: Calculate the cumulative deviation rate between the static geological data of the current rock and soil area and the universal patterns of each rock and soil body, and use the key monitoring parameters corresponding to the universal pattern of the rock and soil body with the lowest cumulative deviation rate as the representative monitoring parameters of the current rock and soil area.

3. The method for early warning of rock and soil mass rupture according to claim 2, characterized in that: Calculate the abnormal score of each key monitoring parameter in each rock and soil sample. The specific operation is as follows: For any rock sample, based on its Key monitoring data obtained at each monitoring time point , =1, 2, …, ; =1, 2, 3, 4; to It represents the surface displacement rate, acoustic emission event frequency, pore water pressure and principal strain value in turn; For any key monitoring data , based on preset limit thresholds , using the formula Calculate key monitoring data First Rating ; Using the formula Calculate and obtain key monitoring data Second rating of ,in, For key monitoring data The standard deviation of Calculate and obtain key monitoring data The third rating ; Finally, using the formula The anomaly scores of surface displacement rate, acoustic emission event frequency, pore water pressure and principal strain value are calculated in sequence. .

4. The method for early warning of rock and soil mass rupture according to claim 3, characterized in that: The first screening sample, the second screening sample, the third screening sample and the fourth screening sample are used to obtain the general mode of the rock and soil corresponding to the surface displacement rate, the frequency of acoustic emission events, the pore water pressure and the principal strain value. The specific operations are as follows: Obtain a static geological dataset of all first screening samples and separate the most common rock types , average compressive strength , average shear strength , average permeability , average joint density and the most historical destruction mode Universal mode of rock mass as a function of surface displacement rate , =1, 2, …, 6; For the second, third and fourth screening samples, the same operation as the first screening sample was performed to obtain the universal modes of the rock and soil corresponding to the acoustic emission event frequency, pore water pressure and principal strain value. , and .

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

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

7. The method for early warning of rock and soil mass rupture according to claim 6, characterized in that: The specific operations for training the rupture prediction model are as follows: Obtain a number of model training samples with annotated rupture prediction results, each of which contains a historical rock and soil monitoring time series set of a rock and soil area; divide all model training samples into a training set and a validation set, use the training set to train the rupture prediction model with initialized parameters, and then input the validation set into the rupture prediction model for validation to obtain the validation results; A first training condition is set to determine whether the obtained verification result meets the first training condition. If so, the trained rupture prediction model is output; if not, the rupture prediction model is continuously trained using the training set.

8. The pre-warning system for rock and soil rupture is characterized by: The system is applied to the pre-warning method for rock and soil fracture according to any one of claims 1 to 7, comprising: Static geological data acquisition module, used to obtain static geological data sets of rock and soil area; A representative monitoring parameter selection module is used to select representative monitoring parameters of the current rock and soil area from the dynamic monitoring parameter set using the acquired static geological data set; The rock and soil monitoring module includes a pattern discrimination unit and a rupture prediction unit; the pattern discrimination unit is used to obtain representative monitoring data at any monitoring time point, and determine whether the obtained representative monitoring data has an abnormality. If so, a comprehensive monitoring mode is executed; if not, no operation is performed; the rupture prediction unit is used to obtain the rock and soil monitoring time series set and use it as the input of the rupture prediction model, and output the rupture prediction result of the rock and soil area.

Images