Bedding rock slope monitoring and early warning method based on micro-seismic and acoustic emission collaborative perception

Abstract

The invention discloses a bedding rock slope monitoring and early warning method based on micro-seismic and acoustic emission cooperative perception, and relates to the technical field of slope monitoring and early warning. Micro-seismic signals and acoustic emission signals in a micro-seismic and acoustic emission sensing array are collected in real time according to preset sampling frequencies corresponding to different sensor types; the collected signals are preprocessed, and the preprocessed signals are divided into a natural fracture event, a construction disturbance event and an environmental noise event based on a pre-trained noise recognition model. By capturing precursor behaviors such as microcrack initiation and interlayer friction slippage, effective early warning can be given out several hours to several days earlier than traditional displacement monitoring; full-scale damage monitoring from millimeter-scale cracks to meter-scale fracture surfaces is realized through cooperation of micro-shock and acoustic emission; meanwhile, the method can provide a spatial migration criterion for layer direction control, and improves the special recognition capability for bedding slopes.

Description

Technical Field

[0001] This invention relates to the field of slope monitoring and early warning technology, specifically to a method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing. Background Technology

[0002] Bedding rock slopes refer to slope types where the dip direction of the rock strata is basically consistent with and similar to the dip angle of the slope surface. Due to the high coupling between the rock strata attitude and the slope attitude, this type of slope is extremely prone to planar sliding instability along bedding planes or weak interlayers, making it one of the most dangerous and difficult slope types to treat in engineering construction. The instability and failure of bedding rock slopes not only exhibits typical gradualism, undergoing a multi-stage evolution process including crack initiation, interlayer displacement, fissure connection, and accelerated instability, but also significant suddenness, meaning that under the triggering of external rainfall, excavation, or seismic disturbances, it can rapidly transform from a creeping state to a rapid collapse in a short period of time. The coexistence of these two characteristics makes the monitoring and early warning of bedding rock slopes a key focus and challenge in the fields of engineering geology and geotechnical engineering.

[0003] Currently, the most widely used monitoring and early warning technologies fall into the following categories: First, surface displacement monitoring methods. These are represented by total stations, GNSS, inclinometers, and displacement gauges, monitoring the deformation of the slope surface and interior. While technically mature and highly operable in engineering, these methods are essentially passive, post-hoc monitoring, only generating identifiable signals after the slope's macroscopic deformation reaches a certain magnitude. They struggle to capture precursory behaviors such as deep crack initiation and interlayer slip, resulting in insufficient early warning. Second, single microseismic monitoring methods. These monitor elastic waves generated by internal rock fractures, recording the spatiotemporal distribution, energy evolution, and source mechanism of micro-fracture events in real time, reflecting the internal damage state of the slope. Numerous studies have shown that microseismic technology can be used for stability evaluation of high slopes in hydropower projects, achieving good results. However, single microseismic systems lack sensitivity to near-field small-scale cracks (especially millimeter- to centimeter-scale interlayer frictional slip), limiting their ability to capture early interlayer shear slip in bedding slopes. Third, single acoustic emission monitoring methods. Acoustic emission is highly sensitive to microcrack initiation and near-field damage, and can identify high-frequency, low-energy microcrack activity. However, acoustic emission signals attenuate rapidly, have a small effective propagation range, making it difficult to construct a monitoring network covering the entire bedding rock slope independently, and suffer from a high false alarm rate under complex noise conditions on site. The fourth category is multi-parameter integrated monitoring methods. Some projects use multi-parameter integrated monitoring of displacement, rainfall, groundwater, etc., with threshold exceeding the limit as the early warning trigger condition. However, this type of method is usually based on statistical empirical criteria, lacking a reflection of the physical essence of the damage mechanism, and the threshold setting often relies on prior experience, resulting in insufficient generalization ability and difficulty in adapting to bedding slopes with different engineering geological conditions. Therefore, existing methods have significant shortcomings in monitoring and early warning of bedding rock slopes, and there is an urgent need to develop a multi-source collaborative monitoring and early warning method based on physical field information that can organically integrate microseismic monitoring and acoustic emission monitoring, specifically targeting the entire process of progressive damage of bedding rock slopes. Summary of the Invention

[0004] The purpose of this invention is to provide a method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing, in order to solve the problems mentioned in the background art.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a method for monitoring and early warning of bedding rock slopes using a combined microseismic and acoustic emission sensing approach, comprising: S1. Obtain the engineering geological information of the target bedding rock slope, and divide the target bedding rock slope into different monitoring functional zones based on the obtained engineering geological information. The monitoring functional zones include the slope top tension cracking zone, the slope middle bedding slip extension zone, and the slope toe locking compression shear zone. S2. Based on the monitoring functional zones obtained after dividing the target bedding rock slope, microseismic sensors and acoustic emission sensors are deployed in different monitoring functional zones according to the preset sensor deployment principle to construct a microseismic and acoustic emission sensor array for the target bedding rock slope. S3. Collect micro-vibration signals and acoustic emission signals in the micro-vibration and acoustic emission sensor array in real time at preset sampling frequencies corresponding to different sensor types. Preprocess the collected signals and classify the preprocessed signals into natural rupture events, construction disturbance events and environmental noise events based on a pre-trained noise recognition model. S4. Extract acoustic emission characteristic parameters, microseismic characteristic parameters and event spatial migration characteristics corresponding to the selected natural rupture events, and construct a multi-dimensional characteristic parameter combination corresponding to the natural rupture events. S5. Based on the combination of multi-dimensional feature parameters corresponding to the obtained natural rupture events, the internal state of the target bedding rock slope is identified by the constructed progressive failure stage identification model of the bedding rock slope, and the progressive failure stage corresponding to the internal state of the target bedding rock slope is obtained. S6. Based on the progressive failure stage corresponding to the internal state of the target bedding rock slope, the weights of the preset evaluation indicators are dynamically adjusted, and the comprehensive slope hazard index of the target bedding rock slope is calculated. S7. Based on the calculation results of the comprehensive slope hazard index of the target bedding rock slope and the progressive failure stage corresponding to the current internal state of the target bedding rock slope, output the graded early warning results.

[0006] Furthermore, the engineering geological information of the target bedding rock slope in S1 includes one or more of the following: slope height, slope angle, slope direction, rock stratum dip, rock stratum dip angle, rock stratum thickness, location and thickness of weak interlayers, location of potential sliding surface, distribution of tensile cracks at the top of the slope, range of unloading relaxation zone at the toe of the slope, groundwater level, and distribution of seepage channels. In the process of delineating different monitoring functional zones within the target bedding rock slope, the boundaries of each monitoring functional zone are determined based on a joint assessment of geological structure and deformation monitoring. The specific criteria are as follows: The boundary of the tensile crack zone at the top of the slope is determined based on the rear crack development zone, the maximum tensile strain zone, or the displacement gradient abrupt change zone; the crack opening and closing degree at the top of the slope is monitored to exceed 10-20cm. The boundary of the bedding slip extension zone in the slope is determined based on the exposure range of the bedding structural plane, the deep displacement concentration zone, or the active interlayer faulting zone; areas with shear characteristics along the bedding plane are detected in the deep displacement curve; The boundary of the slope toe locking compression-shear zone is determined based on the leading anti-slip section, the compressive strain concentration zone, the shear strain abrupt change zone, or the potential shear exit range; the area where compressive strain, microseismic event density, and shear strain concentration are monitored.

[0007] Furthermore, the specific implementation method for deploying micro-vibration sensors and acoustic emission sensors in different monitoring functional areas based on preset sensor deployment principles in S2 includes: Acoustic emission sensors are deployed in the tension crack area at the top of the slope to monitor high-frequency tensile acoustic emission events around the tension crack. In the slope-parallel slip extension zone, acoustic emission sensors are deployed in boreholes along the strike and dip direction of weak interlayers or potential slip zones to form an in-hole array for monitoring interlayer friction slip type acoustic emission signals. Microseismic detectors are installed in the slope toe locking shear zone to monitor shear-type microseismic events; Multiple microseismic sensors are deployed on the slope to form a microseismic monitoring network with three-dimensional positioning capabilities. The positioning accuracy meets the requirements for attributing events to the corresponding geological zones. The S2 also includes auxiliary sensors such as displacement gauges, crack gauges, rain gauges, pore pressure gauges, and inclinometers that are simultaneously deployed in different monitoring functional areas to form a comprehensive monitoring system; all sensors are connected to the data acquisition host through waterproof cables or wireless modules to meet the requirements of real-time transmission.

[0008] Furthermore, the preprocessing methods for the acquired signals in S3 include bandpass filtering, threshold triggering, and waveform truncation. The specific method for pre-training the noise recognition model is as follows: Geotechnical engineers manually labeled historical monitoring signals, classifying them into different event categories, including natural fracturing events, construction disturbance events, and environmental noise events. A training sample set was then constructed based on these event categories. The criteria for manual labeling included: The waveform of the natural rupture event has a clear P-wave initial arrival and S-wave subsequent arrival, which gradually decays; multiple sensors are triggered sequentially according to distance and the apparent wave velocity is consistent with the longitudinal wave velocity of the rock mass, and there is no construction record during the period of occurrence; The waveforms of construction disturbance events have a long duration, exhibiting continuous vibration or repetitive impacts. The dominant frequency is related to the frequency of the construction machinery, and the timing of the events coincides with the construction log. The waveform amplitude of environmental noise events is low, with no obvious first arrival, and the number of triggering sensors is small (only 1 to 2), and it is temporally correlated with rainfall or strong winds.

[0009] The waveform features of the monitoring signal corresponding to each sample in the constructed training sample set are extracted. The waveform features include the signal’s main frequency, duration, rise time and waveform symmetry. The range of waveform features to be extracted for the monitoring signal corresponding to the sample is filtered by combining the construction log time window, and a joint feature vector corresponding to each sample is constructed. By combining the manually labeled information of each sample in the training sample set with the joint feature vector corresponding to the sample, a machine learning classification model is trained to obtain a noise recognition model for recognizing three preset event categories.

[0010] The training model used in this invention is a machine learning classification model, support vector machine and random forest model, to compare and obtain the classification model with the highest recognition accuracy. The input is the labeled training samples, which are divided into training set and test set in an 8:2 ratio. The output of each sample is the probability of the above three events.

[0011] When training the machine learning classification model, this invention establishes classification rules for the above three types of events based on the manually labeled information of each sample in the training sample set and the joint feature vector corresponding to the sample. The slice data samples of natural rupture events, construction disturbance events, and environmental noise events have the following characteristics: Natural fracture events: The rising edge is relatively clear; there are certain transient impact characteristics; the waveform duration is moderate; the frequency band is relatively wide; there is a certain propagation time difference between different sensors; the spatial response is closer to the local response of the tension zone, slip propagation zone, and locking compression shear zone.

[0012] Construction disturbance events include blasting, drilling rig and vehicle vibration. Specifically, blasting is characterized by high energy, suddenness, frequent occurrence in clusters, and obvious synchronous occurrence at multiple stations; drilling rig vibration is characterized by continuous, periodic, and narrow-band or quasi-steady-state frequency components; vehicle vibration is mainly low-frequency, repetitive, and its energy fluctuates smoothly with changes in the route.

[0013] Environmental noise events include rainfall impact, wind vibration, and animal disturbance. Specifically, rainfall impact involves a large number of random small pulses superimposed, resulting in unclear event boundaries; wind vibration is characterized by low-frequency, slow-changing, and continuous noise; and animal disturbance is characterized by short duration, localized noise, low energy, and small propagation range.

[0014] Furthermore, the acoustic emission characteristic parameters in S4 include ring count, cumulative energy, amplitude, rise time, duration, average frequency, and rise angle; The microseismic characteristic parameters include the number of events, cumulative energy, apparent stress, b-value, event location coordinates, dominant frequency, and spatiotemporal clustering degree. The b-value of the microseismic characteristic parameter is a key parameter derived from the Gutenberg-Richter law, used to describe the relative proportion between small and large earthquakes in microseismic events. The calculation formulas involved are as follows: ; Where M is the magnitude of the microseismic event; N is the total number of microseismic events with a magnitude greater than M; a is a constant term characterizing the overall level of microseismic activity in the region (the higher the a value, the greater the total number of microseismic events); and b is the slope (negative value) of the straight line characterizing the ratio of events of different magnitudes. The event space migration characteristics are obtained by calculating the migration speed of the center of gravity of the dense acoustic emission zone and the microseismic event cloud along the dip direction of the bedding plane and the rate of change of the distribution range along the bedding structure plane per unit time.

[0015] Furthermore, the progressive failure stage identification model in S5 is a Markov state model. The input of the progressive failure stage identification model is the combination of multi-dimensional feature parameters corresponding to the obtained natural fracturing events, and the output is the probability of the target bedding rock slope based on different progressive failure stages and the corresponding warning level. The progressive failure stages corresponding to the internal state of the target bedding rock slope include the basic stable period, the initial damage initiation period, the interlayer slip activation period, the fracture penetration and propagation period, and the pre-slip instability period. The specific manifestations of the above five progressive failure stages are as follows: Basic stable period (stage I): Microseismic events are rare and low in energy, acoustic emission ringing counts are at background levels, there is no obvious spatial accumulation, and the slope is in a stable state; Initial damage initiation stage (stage II): Acoustic emission ring count begins to increase, tensile acoustic emission events (low rise angle value, high average frequency value) appear, and microseismic activity is still relatively weak; combined with the slight change in the slope top crack gauge reading, it is determined that microcracks have begun to initiate. Interlayer slip activation period (stage III): Acoustic emission shear events (high rise angle value, low average frequency value) increase significantly, and the signal of the acoustic emission sensor in the borehole is significantly stronger than that of the surface sensor, indicating that interlayer frictional slip is occurring; the number of microseismic events increases, the event location begins to concentrate on the weak interlayer, and the b value decreases. During the fracture penetration and propagation stage (stage IV), the microseismic energy increases rapidly, the proportion of high-energy events rises, the b-value continues to decrease, the event location shows a clear zonal distribution along the bedding plane, the spatial migration speed accelerates, the displacement acceleration begins to rise, and the pore pressure gauge shows anomalies. Precipitation instability stage (stage V): High-energy microseismic events occur, dense rupture events occur in the locked section, acoustic emission ringing count and microseismic energy both show pulse-like surges, and the displacement rate of the locked zone at the toe of the slope suddenly increases. Based on the comprehensive assessment, it is determined that instability is imminent.

[0016] Furthermore, the preset evaluation indicators in S6 include acoustic emission activity index, microseismic energy release index, event space migration index, b-value anomaly index, displacement acceleration index, and rainfall-induced correction index. The calculation formula for the comprehensive slope hazard index of the target bedding rock slope is as follows: ; in, The acoustic emission activity index, The microseismic energy release index. The event space migration index, The b-value is an anomaly index. The displacement acceleration index, Rainfall-induced correction index; , , , , and These are the weights corresponding to different preset evaluation indicators.

[0017] Furthermore, when dynamically adjusting the weights of the preset evaluation indicators in step 6, the progressive failure stage corresponding to the internal state of the target bedding rock slope is obtained, and the weights of each preset evaluation indicator bound to the corresponding progressive failure stage are retrieved according to the progressive failure stage corresponding to the internal state of the target bedding rock slope; the weight values ​​of the same preset evaluation indicator are different in different progressive failure stages.

[0018] The specific implementation methods for dynamically adjusting the weights of preset evaluation indicators include: When the slope is in the initial damage initiation stage of Stage II, increase the weight of the acoustic emission activity index. ; When the slope is in the activation phase of interlayer slippage in stage III, the weight of the acoustic emission activity index should be increased. and the weight of the event space migration index ; When the slope is in the stage IV crack penetration and propagation phase, increase the weight of the microseismic energy release index. Weights of the b-value anomaly index ; When the slope is in the stage V pre-slip instability period, increase the weight of the microseismic energy release index. The weight of the abnormal index I_b of the b-value Weights of displacement acceleration exponent .

[0019] Furthermore, the graded early warning result in S7 is one of multiple preset early warning levels. Each early warning level corresponds to a preset comprehensive hazard index range and one or more progressive damage stages, and different early warning levels are bound to different recommended measures. The comprehensive hazard index range corresponding to the output graded early warning result includes the slope hazard comprehensive index calculation result of the target bedding rock slope, and the progressive damage stage corresponding to the output graded early warning result includes the progressive damage stage corresponding to the current internal state of the target bedding rock slope.

[0020] Compared with the prior art, the beneficial effects achieved by the present invention are: (1) By capturing precursor behaviors such as microcrack initiation and interlayer friction slip, this invention can issue an effective early warning several hours to several days earlier than traditional displacement monitoring. (2) This invention achieves full-scale damage monitoring from millimeter-level cracks to meter-level fracture surfaces through the synergy of micro-vibration and acoustic emission; (3) The present invention can provide spatial migration criteria for control of bedding direction, which improves the special identification capability of bedding slope. Attached Figure Description

[0021] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart illustrating a method for monitoring and early warning of bedding rock slopes using a combined microseismic and acoustic emission sensing approach according to the present invention. Figure 2 This is a schematic diagram of the acoustic emission monitoring deployment and data acquisition and processing of a method for monitoring and early warning of bedding rock slopes using microseismic and acoustic emission co-sensing according to the present invention. Figure 3 This is a schematic diagram of the microseismic monitoring deployment and data acquisition and processing of a method for monitoring and early warning of bedding rock slopes using a combined microseismic and acoustic emission sensing approach according to the present invention. Figure 4 This is a schematic diagram of the graded early warning results output by the co-sensing method of microseismic and acoustic emission for monitoring and early warning of bedding rock slopes according to the present invention. Figure 5 This invention provides a schematic diagram of characteristic information of the slope in the embodiment of a method for monitoring and early warning of bedding rock slopes using microseismic and acoustic emission co-sensing. Figure 6 The present invention provides a schematic diagram of a bedding rock slope in a hydropower station reservoir area based on a method for monitoring and early warning of bedding rock slopes using a combined microseismic and acoustic emission sensing approach. Figure 7 This invention provides a schematic diagram summarizing monitoring data and early warning records in an embodiment of a method for monitoring and early warning of bedding rock slopes using a combined microseismic and acoustic emission sensing approach. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figure 1 This embodiment provides a method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing, including: S1. Obtain the engineering geological information of the target bedding rock slope, and divide the target bedding rock slope into different monitoring functional zones based on the obtained engineering geological information. The monitoring functional zones include the slope top tension cracking zone, the slope middle bedding slip extension zone, and the slope toe locking compression shear zone. The engineering geological information of the target bedding rock slope in S1 includes one or more of the following: slope height, slope angle, slope direction, rock layer dip, rock layer dip angle, rock layer thickness, location and thickness of weak interlayers, location of potential sliding surface, distribution of tensile cracks at the top of the slope, range of unloading relaxation zone at the toe of the slope, groundwater level, and distribution of seepage channels. In the process of delineating different monitoring functional zones within the target bedding rock slope, the boundaries of each monitoring functional zone are determined based on a joint assessment of geological structure and deformation monitoring. The specific criteria are as follows: The boundary of the tensile crack zone at the top of the slope is determined based on the rear crack development zone, the maximum tensile strain zone, or the displacement gradient abrupt change zone; the crack opening and closing degree at the top of the slope is monitored to exceed 10-20cm. The boundary of the bedding slip extension zone in the slope is determined based on the exposure range of the bedding structural plane, the deep displacement concentration zone, or the active interlayer faulting zone; areas with shear characteristics along the bedding plane are detected in the deep displacement curve; The boundary of the slope toe locking compression-shear zone is determined based on the leading anti-slip section, the compressive strain concentration zone, the shear strain abrupt change zone, or the potential shear exit range; the area where compressive strain, microseismic event density, and shear strain concentration are monitored.

[0024] S2. Based on the monitoring functional zones obtained after dividing the target bedding rock slope, microseismic sensors and acoustic emission sensors are deployed in different monitoring functional zones according to the preset sensor deployment principle to construct a microseismic and acoustic emission sensor array for the target bedding rock slope. like Figure 2 and Figure 3 As shown, the specific implementation method of deploying microseismic sensors and acoustic emission sensors in different monitoring functional areas based on preset sensor deployment principles in S2 includes: Acoustic emission sensors (surface mounted) are deployed in the tension crack area at the top of the slope to monitor high-frequency tensile acoustic emission events around the tension crack. In the slope-parallel slip extension zone, acoustic emission sensors are deployed in boreholes along the strike and dip direction of weak interlayers or potential slip zones to form an in-hole array for monitoring interlayer friction slip type acoustic emission signals. Microseismic detectors (installed inside holes or on the surface) are deployed in the slope toe locking shear zone to monitor shear-type microseismic events; Multiple microseismic sensors are deployed on the slope to form a microseismic monitoring network with three-dimensional positioning capabilities. The positioning accuracy meets the requirements for attributing events to the corresponding geological zones. The S2 also includes auxiliary sensors such as displacement gauges, crack gauges, rain gauges, pore pressure gauges, and inclinometers that are simultaneously deployed in different monitoring functional areas to form a comprehensive monitoring system; all sensors are connected to the data acquisition host through waterproof cables or wireless modules to meet the requirements of real-time transmission.

[0025] S3. Collect micro-vibration signals and acoustic emission signals in the micro-vibration and acoustic emission sensor array in real time at preset sampling frequencies corresponding to different sensor types. Preprocess the collected signals and classify the preprocessed signals into natural rupture events, construction disturbance events and environmental noise events based on a pre-trained noise recognition model. The preprocessing methods for the acquired signals in S3 include bandpass filtering, threshold triggering, and waveform truncation. The specific method for pre-training the noise recognition model is as follows: Geotechnical engineers manually labeled historical monitoring signals, classifying them into different event categories, including natural fracturing events, construction disturbance events, and environmental noise events. A training sample set was then constructed based on these event categories. The criteria for manual labeling included: The waveform of the natural rupture event has a clear P-wave arrival and subsequent S-wave arrival, which gradually decays; the dominant frequency is 50~2000Hz; multiple sensors are triggered sequentially according to distance and the apparent wave velocity is consistent with the longitudinal wave velocity of the rock mass (2000~5000m / s), and there is no construction record during the occurrence period; The waveforms of construction disturbance events have a long duration, exhibiting continuous vibration or repetitive impacts. The dominant frequency is related to the frequency of the construction machinery (e.g., drilling operations are concentrated in the range of 10-200 Hz), and the timing of the events coincides with the construction log. The waveform amplitude of environmental noise events is low, with no obvious first arrival, a wide spectral distribution or concentrated in the low frequency band below 50 Hz, a small number of triggering sensors (only 1 to 2), and time-related to rainfall or strong winds.

[0026] The waveform features of the monitoring signal corresponding to each sample in the constructed training sample set are extracted. The waveform features include the signal's main frequency, duration, rise time, and waveform symmetry. The range of waveform feature extraction for the monitoring signal corresponding to the sample is filtered by combining the construction log time window, and a joint feature vector corresponding to each sample is constructed. The construction log time window is introduced in this invention to automatically shield signal acquisition during known blasting or strong disturbance construction periods to avoid construction noise pollution of the database. By combining the manually labeled information of each sample in the training sample set with the joint feature vector corresponding to the sample, a machine learning classification model is trained to obtain a noise recognition model for recognizing three preset event categories.

[0027] The training model used in this invention is a machine learning classification model, support vector machine and random forest model, to compare and obtain the classification model with the highest recognition accuracy. The input is the labeled training samples, which are divided into training set and test set in an 8:2 ratio. The output of each sample is the probability of the above three events.

[0028] When training the machine learning classification model, this invention establishes classification rules for the above three types of events based on the manually labeled information of each sample in the training sample set and the joint feature vector corresponding to the sample. The slice data samples of natural rupture events, construction disturbance events, and environmental noise events have the following characteristics: Natural fracture events: The rising edge is relatively clear; there are certain transient impact characteristics; the waveform duration is moderate; the frequency band is relatively wide; there is a certain propagation time difference between different sensors; the spatial response is closer to the local response of the tension zone, slip propagation zone, and locking compression shear zone.

[0029] Construction disturbance events include blasting, drilling rig and vehicle vibration. Specifically, blasting is characterized by high energy, suddenness, frequent occurrence in clusters, and obvious synchronous occurrence at multiple stations; drilling rig vibration is characterized by continuous, periodic, and narrow-band or quasi-steady-state frequency components; vehicle vibration is mainly low-frequency, repetitive, and its energy fluctuates smoothly with changes in the route.

[0030] Environmental noise events include rainfall impact, wind vibration, and animal disturbance. Specifically, rainfall impact involves a large number of random small pulses superimposed, resulting in unclear event boundaries; wind vibration is characterized by low-frequency, slow-changing, and continuous noise; and animal disturbance is characterized by short duration, localized noise, low energy, and small propagation range.

[0031] S4. Extract acoustic emission characteristic parameters, microseismic characteristic parameters and event spatial migration characteristics corresponding to the selected natural rupture events, and construct a multi-dimensional characteristic parameter combination corresponding to the natural rupture events. The acoustic emission characteristic parameters in S4 include ring count, cumulative energy, amplitude, rise time, duration, average frequency, and rise angle. The microseismic characteristic parameters include the number of events, cumulative energy, apparent stress, b-value, event location coordinates, dominant frequency, and spatiotemporal clustering degree. The b-value of the microseismic characteristic parameter is a key parameter derived from the Gutenberg-Richter law, used to describe the relative proportion between small and large earthquakes in microseismic events. The calculation formulas involved are as follows: ; Where M is the magnitude of the microseismic event; N is the total number of microseismic events with a magnitude greater than M; a is a constant term characterizing the overall level of microseismic activity in the region (the higher the a value, the greater the total number of microseismic events); and b is the slope (negative value) of the straight line characterizing the ratio of events of different magnitudes. The event space migration characteristics are obtained by calculating the migration speed of the center of gravity of the dense acoustic emission zone and the microseismic event cloud along the dip direction of the bedding plane and the rate of change of the distribution range along the bedding structure plane per unit time.

[0032] S5. Based on the combination of multi-dimensional feature parameters corresponding to the obtained natural rupture events, the internal state of the target bedding rock slope is identified by the constructed progressive failure stage identification model of the bedding rock slope, and the progressive failure stage corresponding to the internal state of the target bedding rock slope is obtained. The progressive failure stage identification model in S5 is a Markov state model. The input to this model is a combination of multi-dimensional feature parameters corresponding to the obtained natural fracturing events, and the output is the probability and corresponding warning level of the target bedding rock slope based on different progressive failure stages. The progressive failure stages corresponding to the internal state of the target bedding rock slope include the basic stable period, the initial damage initiation period, the interlayer slip activation period, the fracture penetration and propagation period, and the pre-slip instability period. The specific manifestations of the above five progressive failure stages are as follows: Basic stable period (stage I): Microseismic events are rare and low in energy, acoustic emission ringing counts are at background levels, there is no obvious spatial accumulation, and the slope is in an elastic state; Initial damage initiation stage (stage II): Acoustic emission ring count begins to increase, tensile acoustic emission events (low rise angle value, high average frequency value) appear, and microseismic activity is still relatively weak; combined with the slight change in the slope top crack gauge reading, it is determined that microcracks have begun to initiate. Interlayer slip activation period (stage III): Acoustic emission shear events (high rise angle value, low average frequency value) increase significantly, and the signal of the acoustic emission sensor in the borehole is significantly stronger than that of the surface sensor, indicating that interlayer frictional slip is occurring; the number of microseismic events increases, the event location begins to concentrate on the weak interlayer, and the b value decreases. During the fracture penetration and propagation stage (stage IV), the microseismic energy increases rapidly, the proportion of high-energy events rises, the b-value continues to decrease, the event location shows a clear zonal distribution along the bedding plane, the spatial migration speed accelerates, the displacement acceleration begins to rise, and the pore pressure gauge shows anomalies. Precipitation instability stage (stage V): High-energy microseismic events occur, dense rupture events occur in the locked section, acoustic emission ringing count and microseismic energy both show pulse-like surges, and the displacement rate of the locked zone at the toe of the slope suddenly increases. Based on the comprehensive assessment, it is determined that instability is imminent.

[0033] S6. Based on the progressive failure stage corresponding to the internal state of the target bedding rock slope, the weights of the preset evaluation indicators are dynamically adjusted, and the comprehensive slope hazard index of the target bedding rock slope is calculated. The preset evaluation indicators in S6 include acoustic emission activity index, microseismic energy release index, event space migration index, b-value anomaly index, displacement acceleration index, and rainfall-induced correction index. Specifically, the acoustic emission activity index is calculated by normalizing the acoustic emission ringing count growth rate and the proportion of shear-type events; The microseismic energy release index is calculated by normalizing the cumulative microseismic energy growth rate per unit time. The event space migration index is calculated by normalizing the migration velocity of the acoustic emission / microseismic centroid along the plane dip direction. The b-value anomaly index is calculated by normalizing the decrease in the current b-value relative to the background b-value. The displacement acceleration exponent is calculated by normalizing the tangent angle or displacement rate increment; The rainfall-induced correction index is calculated by normalizing the ratio of the current rainfall intensity to the historical trigger threshold. The calculation formula for the comprehensive slope hazard index of the target bedding rock slope is as follows: ; in, The acoustic emission activity index, The microseismic energy release index. The event space migration index, The b-value is an anomaly index. The displacement acceleration index, Rainfall-induced correction index; , , , , and These are the weights corresponding to different preset evaluation indicators.

[0034] When dynamically adjusting the weights of the preset evaluation indicators in step 6, the progressive failure stage corresponding to the internal state of the target bedding rock slope is obtained, and the weights of each preset evaluation indicator bound to the corresponding progressive failure stage are retrieved according to the progressive failure stage corresponding to the internal state of the target bedding rock slope; the weight values ​​of the same preset evaluation indicator are different in different progressive failure stages.

[0035] The specific implementation methods for dynamically adjusting the weights of preset evaluation indicators include: When the slope is in the initial damage initiation stage of Stage II, increase the weight of the acoustic emission activity index. ; When the slope is in the activation phase of interlayer slippage in stage III, the weight of the acoustic emission activity index should be increased. and the weight of the event space migration index ; When the slope is in the stage IV crack penetration and propagation phase, increase the weight of the microseismic energy release index. Weights of the b-value anomaly index ; When the slope is in the stage V pre-slip instability period, increase the weight of the microseismic energy release index. The weight of the abnormal index I_b of the b-value Weights of displacement acceleration exponent .

[0036] S7. Based on the calculated comprehensive slope hazard index of the target bedding rock slope and the progressive failure stage corresponding to the current internal state of the target bedding rock slope, output graded early warning results. These early warning results are released in real time through the monitoring platform, app / SMS, and audible and visual alarm devices, and a complete event log is recorded for post-event inspection and feedback analysis; such as... Figure 4 As shown, the output graded early warning results include the relationship between the early warning level and the calculated results of the slope hazard comprehensive index, the progressive failure stage corresponding to the current internal state of the target bedding rock slope, and the recommended measures.

[0037] The graded early warning result in S7 is one of multiple preset early warning levels. Each early warning level corresponds to a preset comprehensive hazard index range and one or more progressive damage stages, and different early warning levels are bound to different recommended measures. The comprehensive hazard index range corresponding to the output graded early warning result includes the slope hazard comprehensive index calculation result of the target bedding rock slope, and the progressive damage stage corresponding to the output graded early warning result includes the progressive damage stage corresponding to the current internal state of the target bedding rock slope.

[0038] This example uses the monitoring and early warning of bedding rock slopes in a hydropower station reservoir area as a case study. A typical bedding rock slope in a hydropower station reservoir area in Southwest China is selected as the verification object. This slope has the following characteristics: Figure 5 , Figure 6 As shown, the rock strata dip nearly parallel to the slope (bedding), with clearly defined weak interlayers and signs of initial deformation. Small-scale slope slippage has occurred after historical rainfall, exhibiting typical characteristics of progressive failure of bedding slopes, making it suitable for verification using a full-process method.

[0039] Based on the combined results of UAV oblique photography, three exploration boreholes (ZK1-ZK3), and slope geological mapping, the slope is divided into three functional zones: ①Tensile crack zone at the top of the slope (elevation 760-820m): Three tensile cracks are densely developed, which are the main cause of tensile cracking failure; ② Slope-interlayer slip extension zone (elevation 640-760m): weak shale interlayers are developed, and interlayer friction slip is the main control. ③ Slope toe locking compression-shear zone (elevation 590~640m): unloading relaxation, stress concentration, main control compression-shear shearing.

[0040] Based on the functional area division, sensors should be deployed in the following orientations: Tension crack zone at the top of the slope: Eight surface acoustic emission sensors (frequency response 50-400 kHz) are deployed along both sides of the three tensile cracks to monitor high-frequency tensile acoustic emission events; Slope slip extension zone: Four inclined holes are constructed near the weak interlayer, with two in-hole acoustic emission sensors installed in each hole, for a total of eight, forming an in-hole array to specifically monitor the acoustic emission signals of interlayer slip. Slope toe locking zone: Construct 3 vertical holes, each with 1 three-component in-hole microseismic detector, for a total of 3, to capture high-energy shear-type microseismic events; Upper and middle parts of the slope: Six surface microseismic detectors (microseismic 1 to microseismic 6) are installed in a spatial triangular layout, with a plane spacing of 30 to 50 m, forming a three-dimensional positioning network with a positioning accuracy of ≤5 m; Auxiliary sensors: 4 sets of multi-point displacement gauges, 6 crack gauges, 2 rain gauges, 4 pore pressure gauges, and 3 inclinometers were deployed to form a comprehensive monitoring system.

[0041] All sensors are connected to the slope foot data acquisition host via armored waterproof cables. The acoustic emission system has a sampling frequency of 1MHz, the micro-vibration system has a sampling frequency of 4kHz, and the data is uploaded in real time via a dedicated fiber optic line with 5G wireless backup.

[0042] The acoustic emission system has a bandpass filter range of 20–500 kHz and a trigger threshold of 45 dB; the microseismic system has a bandpass filter range of 1–800 Hz. Based on the signal's dominant frequency, duration, rise time, and waveform symmetry, the acquired signals are categorized into three types: natural rupture (C1), construction disturbance (C2), and environmental noise (C3). Types C2 and C3 are filtered, retaining only type C1 for subsequent analysis. Before system deployment, a known-scale blasting test was conducted to verify the accuracy of C2 noise identification, achieving a measured accuracy of 97%.

[0043] For C1 type events, RA values ​​(rise time / amplitude) and AF values ​​(ring count / duration) are extracted to determine the failure mode; shear failure is defined as RA > 100 μs / V and AF < 50 kHz, while tensile failure is defined as RA < 50 μs / V and AF > 100 kHz. Microseismic parameters extracted include b-value (sliding window maximum likelihood method, window length 30 days), cumulative energy, apparent stress, and event cloud centroid migration velocity.

[0044] The slope underwent continuous monitoring for 180 days (covering the entire flood season). The evolution of characteristic parameters and early warning triggering in five stages of damage are as follows: Figure 7As shown. Verification confirmed that the system issued a red alert on day 162, completing the evacuation of personnel; on day 165, a localized landslide occurred on the slope, with a volume of approximately 3.2 × 10⁻⁶ cubic meters per second. 4 m³, no casualties. The main verification indicators are as follows: Warning lead time: Red warnings are issued 72 hours in advance, and orange warnings are issued approximately 9 days in advance; Location accuracy: The consistency rate between the spatial location of the event and the damage surface in the subsequent geological investigation is ≥87%; Stage identification accuracy: The deviation between the transition time node of each stage and the inflection point of the displacement curve is ≤3 days; Synergistic advantages: Single acoustic emission monitoring cannot complete three-dimensional positioning, and single microseismic monitoring cannot identify early tensile initiation. The synergy between the two achieves full five-stage coverage, mutually corroborates each other, and effectively reduces the risk of missed and false alarms.

[0045] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0046] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing, characterized in that, include: S1. Obtain the engineering geological information of the target bedding rock slope, and divide the target bedding rock slope into different monitoring functional zones based on the obtained engineering geological information. The monitoring functional zones include the slope top tension cracking zone, the slope middle bedding slip extension zone, and the slope toe locking compression shear zone. S2. Based on the monitoring functional zones obtained after dividing the target bedding rock slope, microseismic sensors and acoustic emission sensors are deployed in different monitoring functional zones according to the preset sensor deployment principle to construct a microseismic and acoustic emission sensor array for the target bedding rock slope. S3. Collect micro-vibration signals and acoustic emission signals in the micro-vibration and acoustic emission sensor array in real time at preset sampling frequencies corresponding to different sensor types. Preprocess the collected signals and classify the preprocessed signals into natural rupture events, construction disturbance events and environmental noise events based on a pre-trained noise recognition model. S4. Extract acoustic emission characteristic parameters, microseismic characteristic parameters and event spatial migration characteristics corresponding to the selected natural rupture events, and construct a multi-dimensional characteristic parameter combination corresponding to the natural rupture events. S5. Based on the combination of multi-dimensional feature parameters corresponding to the obtained natural rupture events, the internal state of the target bedding rock slope is identified by the constructed progressive failure stage identification model of the bedding rock slope, and the progressive failure stage corresponding to the internal state of the target bedding rock slope is obtained. S6. Based on the progressive failure stage corresponding to the internal state of the target bedding rock slope, the weights of the preset evaluation indicators are dynamically adjusted, and the comprehensive slope hazard index of the target bedding rock slope is calculated. S7. Based on the calculation results of the comprehensive slope hazard index of the target bedding rock slope and the progressive failure stage corresponding to the current internal state of the target bedding rock slope, output the graded early warning results.

2. The method for monitoring and early warning of bedding rock slopes using a combined microseismic and acoustic emission sensing approach according to claim 1, characterized in that, The engineering geological information of the target bedding rock slope in S1 includes one or more of the following: slope height, slope angle, slope direction, rock layer dip, rock layer dip angle, rock layer thickness, location and thickness of weak interlayers, location of potential sliding surface, distribution of tensile cracks at the top of the slope, range of unloading relaxation zone at the toe of the slope, groundwater level, and distribution of seepage channels. In the process of delineating different monitoring functional zones within the target bedding rock slope, the boundaries of each monitoring functional zone are determined based on a joint assessment of geological structure and deformation monitoring. The specific criteria are as follows: The boundary of the tensile crack zone at the top of the slope is determined based on the rear crack development zone, the maximum tensile strain zone, or the displacement gradient abrupt change zone. The boundary of the bedding slip extension zone in the slope is determined based on the exposure range of the bedding structural plane, the deep displacement concentration zone, or the active interlayer fault zone. The boundary of the slope toe locking compression-shear zone is determined based on the leading edge anti-slip section, the compressive strain concentration zone, the shear strain abrupt change zone, or the potential shear exit range.

3. The method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing according to claim 1, characterized in that, The specific implementation method for deploying microseismic sensors and acoustic emission sensors in different monitoring functional areas based on preset sensor deployment principles in S2 includes: Acoustic emission sensors are deployed in the tension crack area at the top of the slope to monitor high-frequency tensile acoustic emission events around the tension crack. In the slope-parallel slip extension zone, acoustic emission sensors are deployed in boreholes along the strike and dip direction of weak interlayers or potential slip zones to form an in-hole array for monitoring interlayer friction slip type acoustic emission signals. Microseismic detectors are installed in the slope toe locking shear zone to monitor shear-type microseismic events; Multiple microseismic sensors are deployed on the slope to form a microseismic monitoring network with three-dimensional positioning capabilities.

4. The method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing according to claim 1, characterized in that, The preprocessing methods for the acquired signals in S3 include bandpass filtering, threshold triggering, and waveform truncation. The specific method for pre-training the noise recognition model is as follows: Geotechnical engineers manually annotate historical monitoring signals and classify them into different event categories, including natural rupture events, construction disturbance events, and environmental noise events. A training sample set is then constructed based on the historical monitoring signals classified into different event categories. The waveform features of the monitoring signal corresponding to each sample in the constructed training sample set are extracted. The waveform features include the signal’s main frequency, duration, rise time and waveform symmetry. The range of waveform features to be extracted for the monitoring signal corresponding to the sample is filtered by combining the construction log time window, and a joint feature vector corresponding to each sample is constructed. By combining the manually labeled information of each sample in the training sample set with the joint feature vector corresponding to the sample, a machine learning classification model is trained to obtain a noise recognition model for recognizing three preset event categories.

5. The method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing according to claim 1, characterized in that, The acoustic emission characteristic parameters in S4 include ring count, cumulative energy, amplitude, rise time, duration, average frequency, and rise angle. The microseismic characteristic parameters include the number of events, cumulative energy, apparent stress, b-value, event location coordinates, dominant frequency, and spatiotemporal clustering degree. The event space migration characteristics are obtained by calculating the migration speed of the center of gravity of the dense acoustic emission zone and the microseismic event cloud along the dip direction of the bedding plane and the rate of change of the distribution range along the bedding structure plane per unit time.

6. The method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing according to claim 1, characterized in that, The progressive failure stage identification model in S5 is a Markov state model. The input of the progressive failure stage identification model is the combination of multi-dimensional feature parameters corresponding to the obtained natural fracture events, and the output is the probability of the target bedding rock slope based on different progressive failure stages and the corresponding warning level. The progressive failure stages corresponding to the internal state of the target bedding rock slope include the basic stable period, the initial damage initiation period, the interlayer slip activation period, the fracture penetration and propagation period, and the pre-slip instability period.

7. The method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing according to claim 1, characterized in that, The preset evaluation indicators in S6 include acoustic emission activity index, microseismic energy release index, event space migration index, b-value anomaly index, displacement acceleration index, and rainfall-induced correction index. The calculation formula for the comprehensive slope hazard index of the target bedding rock slope is as follows: ； in, The acoustic emission activity index, The microseismic energy release index. The event space migration index, The b-value is an anomaly index. The displacement acceleration index, Rainfall-induced correction index; , , , , and These are the weights corresponding to different preset evaluation indicators.

8. The method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing according to claim 1, characterized in that, When dynamically adjusting the weights of the preset evaluation indicators in step 6, the progressive failure stage corresponding to the internal state of the target bedding rock slope is obtained, and the weights of each preset evaluation indicator bound to the corresponding progressive failure stage are retrieved according to the progressive failure stage corresponding to the internal state of the target bedding rock slope; the weight values ​​of the same preset evaluation indicator are different in different progressive failure stages.

9. The method for monitoring and early warning of bedding rock slopes using a combination of microseismic and acoustic emission sensing according to claim 1, characterized in that, The graded early warning result in S7 is one of multiple preset early warning levels. Each early warning level corresponds to a preset comprehensive danger index range and one or more progressive destruction stages, and different early warning levels are associated with different recommended measures. The output graded early warning results correspond to a comprehensive hazard index range that includes the calculated comprehensive hazard index of the target bedding rock slope, and the output graded early warning results correspond to a progressive failure stage that includes the progressive failure stage corresponding to the current internal state of the target bedding rock slope.

Images