Slope safety early warning method, device, system and medium

Abstract

The invention discloses a slope safety early warning method, device and system and a medium, and the method comprises the steps: obtaining the detonation time of an electronic detonator in a target slope and first micro-seismic waveform data; second microseismic waveform data and a three-dimensional model of the target slope are obtained; recognizing and eliminating collaborative residual sound signals in the second micro-seismic waveform data to obtain target second micro-seismic waveform data; removing a common-mode blasting signal in the target second micro-seismic waveform data to obtain third micro-seismic waveform data; generating a blasting signature representing the blasting event and storing the blasting signature in a blasting noise dictionary; performing event detection on the third micro-seismic waveform data to obtain a plurality of to-be-analyzed events, removing blasting residual sound events from the plurality of to-be-analyzed events to obtain to-be-alarmed events, performing seismic source positioning and closing detection, and if the seismic source positioning and closing detection is passed, determining that the to-be-alarmed events are real micro-seismic events and performing danger alarm; and if not, determining that the event is an interference micro-seismic event. According to the invention, the accuracy of slope safety early warning can be improved.

Description

Technical Field

[0001] This application relates to the field of slope safety early warning technology, specifically to a slope safety early warning method, device, system and medium. Background Technology

[0002] In large-scale geotechnical engineering projects such as mining and tunnel construction, the stability of slopes is directly related to production safety and the lives of personnel. Microseismic monitoring is a key technology for early warning of slope instability, especially after blasting operations, where real-time monitoring of rock mass response is crucial.

[0003] To effectively monitor microseismic activity in slopes, existing technologies typically deploy sensor networks on the slope and set a fixed, static energy trigger threshold based on long-term historical data. When the detected signal energy exceeds the energy trigger threshold, the system records it as a potential microseismic event.

[0004] However, blasting generates a large number of complex interference signals. These signals differ in characteristics from real microseismic events, but their energy can easily exceed the set energy trigger threshold. Because current technology relies on only a single energy dimension for judgment, it cannot effectively distinguish between real microseismic events and these complex post-blasting interference microseismic events, resulting in a large number of false events in the monitoring results, affecting the accuracy of slope safety early warning. Summary of the Invention

[0005] This application provides a slope safety early warning method, device, and system to solve the problem that the accuracy of slope safety early warning is affected by interfering microseismic events caused by blasting in the microseismic detection of slopes after blasting operations.

[0006] The first aspect of this application provides a slope safety early warning method, applied in a server. The method includes: acquiring the detonation time of an electronic detonator in a target slope and first micro-seismic waveform data of the target slope within a first preset time period before the detonation time; acquiring second micro-seismic waveform data of the target slope within a second preset time period after the detonation time and a three-dimensional model of the target slope; identifying and removing co-resonance signals from the second micro-seismic waveform data to obtain target second micro-seismic waveform data, wherein the co-resonance signals are waveform segments with waveform in-phase characteristics among multiple spatially adjacent micro-seismic monitoring nodes in the three-dimensional model; and removing common-mode blasting signals from the target second micro-seismic waveform data to obtain a third micro-seismic waveform. The system analyzes the waveform characteristics of the second microseismic waveform data within the second preset time period, generates a blasting signature representing the blasting event, and saves the blasting signature to the blasting noise dictionary. Based on the first microseismic waveform data and the blasting noise dictionary, the system performs event detection on the third microseismic waveform data to obtain multiple events to be analyzed. Blasting reverberation events are removed from the multiple events to be analyzed to obtain events to be alarmed. Within a preset monitoring window, a source location closure test is performed on each event to be alarmed. If the source location closure test is passed, the event to be alarmed is confirmed as a real microseismic event, and a danger alarm is issued.

[0007] Optionally, common-mode blasting signals are removed from the target second microseismic waveform data to obtain third microseismic waveform data. Specifically, this includes: performing energy structure analysis on the target second microseismic waveform data, calculating the three-component energy ratio of the target second microseismic waveform data, where the three-component energy ratio is the ratio of the vertical energy component to the horizontal energy component of the target second microseismic waveform data; determining the common-mode blasting signals in the target second microseismic waveform data based on the three-component energy ratio, and removing the common-mode blasting signals from the target second microseismic waveform data to obtain third microseismic waveform data.

[0008] Optionally, energy structure analysis is performed on the target second microseismic waveform data to calculate the three-component energy ratio of the target second microseismic waveform data. Specifically, this includes: performing amplitude envelope processing on the target vertical energy component in the target second microseismic waveform data to obtain a first envelope curve; summing the instantaneous energies of the two target horizontal energy components corresponding to each moment in the target second microseismic waveform data to obtain a horizontal energy sequence; performing amplitude envelope processing on the horizontal energy sequence to obtain a second envelope curve; performing morphological analysis on the first envelope curve and the second envelope curve respectively to identify the first energy reversal interval on the first envelope curve where all energy decay trends reverse; and... The second energy reversal interval is defined as the second envelope curve where all energy decay trends reverse. A reversal of the energy decay trend means that the slope of the curve changes from negative or zero to positive. Multiple vertically injected energies are obtained by integrating the first envelope curve within each first energy reversal interval, and multiple horizontally injected energies are obtained by integrating the second envelope curve within each second energy reversal interval. All vertically injected energies are summed to obtain the vertical energy component, and all horizontally injected energies are summed to obtain the horizontal energy component. The ratio of the vertical energy component to the horizontal energy component is calculated to obtain the three-component energy ratio.

[0009] Optionally, the waveform characteristics of the second microseismic waveform data within the second preset duration are analyzed to generate a blasting signature characterizing the blasting event. Specifically, this includes: decomposing the second microseismic waveform data within the second preset duration into multiple modal components based on a variational mode decomposition method, each modal component corresponding to a center frequency and a finite bandwidth; for each modal component, calculating the amplitude envelope of the modal component and performing exponential decay fitting on the amplitude envelope to obtain an energy attenuation coefficient; calculating the power spectrum of each modal component and calculating the ratio of the peak energy of the maximum power spectrum to the total energy to obtain the spectral concentration; summing the modal components whose first energy attenuation coefficient is greater than or equal to a preset attenuation threshold, or whose spectral concentration is less than or equal to a preset concentration threshold, point-by-point in the time domain to obtain a pure blasting waveform; calculating the energy attenuation coefficient and the centroid frequency of the power spectrum of the pure blasting waveform, and constructing the energy attenuation coefficient and the centroid frequency of the power spectrum into a two-dimensional feature vector, using the two-dimensional feature vector as the blasting signature characterizing the blasting event.

[0010] Optionally, event detection is performed on the third microseismic waveform data based on the first microseismic waveform data and the blasting noise dictionary to obtain multiple events to be analyzed. Explosion reverberation events are then removed from the multiple events to be analyzed to obtain events to be alarmed. Specifically, this includes: setting a dynamic trigger threshold based on the first microseismic waveform data; segmenting the third microseismic waveform data to obtain multiple events to be analyzed; analyzing the waveform characteristics of the events to be analyzed; generating event signatures for the events to be analyzed, with one event signature corresponding to one event to be analyzed; the dynamic trigger threshold including a trigger threshold and an end threshold; calculating the feature difference between the event signature and all blasting signatures in the blasting noise dictionary; and removing blasting reverberation events from the events to be analyzed to obtain events to be alarmed. The blasting reverberation events are at least one of the events to be analyzed whose feature difference is less than a preset difference threshold.

[0011] Optionally, based on the first microseismic waveform data, a dynamic trigger threshold is set, and the third microseismic waveform data is segmented to obtain multiple events to be analyzed. Specifically, this includes: performing a sliding window calculation on the first microseismic waveform data based on a preset time window length to obtain a time series of the ratio of the short-time window energy mean to the long-time window energy mean in the first microseismic waveform data; calculating the mean and standard deviation of the time series, and combining the mean and standard deviation to form a background noise reference parameter set characterizing normal environmental fluctuations; calculating the trigger threshold and the termination threshold based on the background noise reference parameter set and a preset calculation formula to obtain the dynamic trigger threshold; and identifying third microseismic waveform data segments in the third microseismic waveform data that meet preset conditions based on the dynamic trigger threshold to obtain multiple events to be analyzed.

[0012] Optionally, based on the dynamic trigger threshold, identify third microseismic waveform data segments in the third microseismic waveform data that meet preset conditions to obtain multiple events to be analyzed. Specifically, this includes: performing a rolling calculation of the long-short time window energy ratio on the third microseismic waveform data based on the preset time window length to obtain a long-short time window energy ratio sequence; traversing the time series corresponding to the third microseismic waveform data, determining the time point corresponding to when the long-short time window energy ratio in the long-short time window energy ratio sequence crosses the trigger threshold as the start time point of the event to be analyzed, determining the time point corresponding to when the long-short time window energy ratio first crosses the end threshold after the start time point as the end time point of the event to be analyzed, and taking the third microseismic waveform data segment between the start time point and the end time point as one event to be analyzed, until the entire time series has been traversed.

[0013] A second aspect of this application provides a slope safety early warning system, comprising: a first acquisition module for acquiring the detonation time of an electronic detonator in a target slope and first microseismic waveform data of the target slope within a first preset time period before the detonation time; a second acquisition module for acquiring second microseismic waveform data of the target slope within a second preset time period after the detonation time and a three-dimensional model of the target slope; a first elimination module for identifying and eliminating co-resonance signals in the second microseismic waveform data to obtain target second microseismic waveform data, wherein the co-resonance signals are waveform segments with waveform phase characteristics among multiple spatially adjacent microseismic monitoring nodes in the three-dimensional model; a second elimination module for eliminating common-mode blasting signals in the target second microseismic waveform data to obtain third microseismic waveform data; and an analysis module for analyzing the... The second microseismic waveform data is analyzed within a second preset time period to generate a blasting signature representing the blasting event, and the blasting signature is saved to a blasting noise dictionary. A detection module is used to perform event detection on the third microseismic waveform data based on the first microseismic waveform data and the blasting noise dictionary, obtaining multiple events to be analyzed. Blasting reverberation events are removed from the multiple events to be analyzed, resulting in events to be alarmed. A verification module is used to perform a source location closure test on each event to be alarmed within a preset monitoring window. If the source location closure test is passed, the event to be alarmed is confirmed as a genuine microseismic event, and a danger alarm is issued. If the source location closure test is not passed, the event to be alarmed is confirmed as an interfering microseismic event.

[0014] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any of the foregoing.

[0015] A fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described in any of the preceding descriptions.

[0016] In summary, one or more technical solutions provided in this application have at least the following technical effects or advantages: 1. By collecting first and second microseismic waveform data before and after detonation, and combining them with a three-dimensional slope model, a signal recognition foundation with spatial constraints was constructed. Based on this, co-resonant signals were first eliminated by using the in-phase characteristics of spatially adjacent points, and then common-mode blasting signals were identified using detonation synchronization. This completed multi-layer cleaning of superimposed blasting interference, significantly improving the authenticity and signal-to-noise ratio of the microseismic data. Subsequently, by analyzing typical blasting events in the second microseismic waveform data, blasting signatures were extracted and a blasting noise dictionary was constructed, enabling the system to have self-learning and memory capabilities for blasting characteristics, achieving adaptive modeling of noise under different blasting modes. By combining the blasting dictionary with the first microseismic background data for linkage analysis of target data, blasting resonant events were further identified and eliminated, ensuring that subsequent analysis objects were high-confidence events awaiting alarm. Finally, through source location closure verification, events with spatial consistency and physical interpretability were confirmed as real microseismic events and alarms were triggered, realizing the distinction between real microseismic events and these complex post-blasting interference microseismic events, and improving the accuracy of slope safety early warning.

[0017] 2. By performing energy structure analysis on the second microseismic waveform data, common-mode blasting signals are identified based on the energy ratio of the vertical energy component to the horizontal energy component (i.e., the three-component energy ratio). First, amplitude envelope processing is performed on the vertical and horizontal energy components respectively. Then, energy reversal intervals are identified through morphological analysis. The injected energy is then calculated within each interval, and finally, the total vertical and horizontal energy is obtained, and their ratio is calculated. Since the blasting signal shows a significant increase in energy in the vertical direction, the three-component energy ratio will be abnormally high in certain intervals. Therefore, this can be used as a criterion for identifying common-mode blasting interference. By removing such abnormal energy ratio intervals from the waveform, common-mode interference caused by the wide-area blasting effect is eliminated, improving the cleanliness of the microseismic data and the reliability of subsequent event identification. This overcomes the problems of traditional methods being insensitive to changes in energy structure and unable to accurately identify common-mode blasting signals.

[0018] 3. By performing variational mode decomposition on the second microseismic waveform data, the data is divided into multiple modal components with center frequencies, and the amplitude envelope and power spectrum characteristics of each component are extracted. For these components, the energy attenuation coefficient and spectral concentration are calculated to screen modal signals with typical blasting characteristics, and these are then superimposed point-by-point in the time domain to reconstruct a clean blasting waveform. Subsequently, the overall energy attenuation coefficient and power spectrum centroid frequency are extracted to construct a two-dimensional blasting signature vector. This application enables the extraction of representative blasting feature templates from complex microseismic data, avoiding the subjective errors of traditional methods that rely on manual experience or frequency domain threshold judgments. It achieves standardized modeling of blasting events, improves the expressive power and recognition accuracy of blasting events, and provides a feature basis for subsequent blasting interference modeling and removal. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the system architecture of an embodiment of a slope safety early warning method or a slope safety early warning system applied in this application. Figure 2 This is a flowchart illustrating a slope safety early warning method according to an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a slope safety early warning system according to an embodiment of this application; Figure 4 This is a schematic diagram of the structure of the electronic device in the embodiments of this application.

[0020] Explanation of reference numerals in the attached drawings: 301, First acquisition module; 302, Second acquisition module; 303, First rejection module; 304, Second rejection module; 305, Analysis module; 306, Detection module; 307, Verification module; 401, Processor; 402, Communication bus; 403, User interface; 404, Network interface; 405, Memory. Detailed Implementation

[0021] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0022] Figure 1 An exemplary system architecture 100 is shown, which can be applied to an embodiment of a slope safety early warning method or a slope safety early warning system of this application.

[0023] like Figure 1 As shown, the system architecture 100 may include multiple terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 provides a communication link between the terminal devices and the server; the network may include various connection types, such as wired connections and fiber optic networks. Terminal devices 101, 102, and 103 are typically deployed at the target slope site to collect real-time microseismic waveform data, blasting initiation signals, sensor location information, etc., and transmit the data to server 105 for analysis and processing via network 104. The terminal devices can integrate various sensor modules, including seismometers, electronic detonator monitoring modules, GPS timing modules, inclinometers, and stress gauges, to support high-precision microseismic monitoring and location analysis.

[0024] Users can interact with server 105 through terminal devices 101, 102, and 103 to receive event detection results, alarm information, historical data reports, etc. Terminal devices can be hardware devices or mobile terminals with relevant software functions installed, including but not limited to smartphones, tablets, laptops, industrial control terminals, etc.; they can integrate or connect to acquisition modules, such as seismic wave sensors, detonator initiation receivers, video acquisition devices, etc. Server 105 can be a physical server, an edge computing device, or a cloud-based software system; it can be implemented as a single device or a distributed server cluster composed of multiple nodes, supporting high-concurrency processing and multi-region deployment. It should be noted that... Figure 1 The number of terminal devices, networks, and servers shown is for illustrative purposes only. Depending on actual deployment needs, the system can be expanded to a multi-terminal collaborative, multi-region, multi-server architecture; in some scenarios, the system may also consist only of terminal devices and local servers, without the need for network connection.

[0025] Figure 2 This is a flowchart illustrating a slope safety early warning method in an embodiment of this application.

[0026] Please see Figure 2 This application embodiment provides a slope safety early warning method, which is applied in a server. The method includes: S201, acquiring the detonation time of the electronic detonator in the target slope and the first micro-vibration waveform data of the target slope within a first preset time before the detonation time; First, areas with potential structural planes, well-developed rock joints, or obvious historical sliding signs within the target slope are selected as microseismic monitoring deployment areas. These areas typically exhibit high geological activity sensitivity and therefore have significant research and early warning value in slope stability analysis. Centered on the deployment area, no fewer than six microseismic monitoring nodes are deployed along the slope face and toe. Each node is equipped with a three-component seismic detector to independently acquire seismic wave data in the vertical direction and two horizontally orthogonal directions. All nodes are time-synchronized via a Global Navigation Satellite System (GNSS) to ensure high temporal consistency of data acquired from different locations, meeting the requirements for subsequent source location and waveform feature comparison. All monitoring nodes use a unified GPS module for time synchronization, ensuring that the seismic waveform data has a consistent and high-precision time signature, facilitating subsequent multi-point joint analysis. Each monitoring node establishes a data link with a ground server via fiber optic or low-latency wireless communication networks, and the acquired waveform data is uploaded in real-time or cached locally at a fixed sampling period.

[0027] The detonation time of the electronic detonator is recorded by a digital detonator controller in the blasting control system. The digital detonator controller is a precision blasting control device with millisecond-level time control accuracy, automatically generating and recording a detonation timestamp simultaneously with the sending of the detonation pulse. This timestamp is synchronously transmitted to the server via a communication interface (such as RS-485 or TCP / IP). To avoid time drift caused by network latency, the server also enables a time synchronization protocol with the digital detonator controller to ensure that the recorded detonation time is consistent with the microseismic waveform recording system. Upon receiving the detonation timestamp, the server immediately sends a retrieval command to the data cache module to extract microseismic waveform data within a first preset time period prior to the detonation time. This first preset time period refers to a fixed time interval preceding the detonation moment, typically set between 10 and 30 seconds, with the specific value determined based on the geological characteristics of the slope and the changes in microseismic background noise during historical blasting activities. By extracting data from this time period, we can obtain the natural microseismic background state and environmental fluctuation characteristics of the slope before the blasting, which can be used to calculate key indicators such as the energy ratio sequence of long and short time windows, dynamic triggering threshold, and background noise characteristic parameters.

[0028] The microseismic waveform data is a continuous time series, stored on the server in its raw data format, including the sampling rate, time stamp, three-component amplitude values, and monitoring point number for each monitoring channel. After extracting the data, the system performs a unified formatting process to ensure that subsequent analysis modules can directly call it, and provides basic data support for processes such as blasting event identification, blasting reverberation removal, and event signature construction.

[0029] S202. Obtain the second microseismic waveform data of the target slope and the three-dimensional model of the target slope within the second preset time after the detonation time; After the detonation time is recorded by the digital detonator controller and uploaded to the server, the server immediately triggers the data scheduling module to start the data acquisition process for the second time period. This second preset duration is a fixed time interval extending from the detonation time, typically set to 20 to 60 seconds, with specific settings depending on the blasting scale, rock mass structure complexity, and microseismic propagation velocity. Within the second preset duration, all microseismic monitoring nodes deployed in the slope continue to maintain high-frequency sampling. The acquired three-component seismic waveform data is uploaded to the server in real time via the existing communication link, or cached locally and then uploaded uniformly. Based on the detonation timestamp, the server retrieves and extracts data from all monitoring nodes covering the second preset duration, forming a complete second microseismic waveform dataset.

[0030] Simultaneously, to achieve spatial correlation analysis and waveform propagation path modeling, a 3D model of the target slope was constructed during the system deployment phase. The 3D model was generated by fusing data from UAV oblique photogrammetry, laser scanning, and geological exploration. Oblique photogrammetry technology used a UAV equipped with a high-resolution camera to acquire images of the slope from multiple angles, achieving high-precision modeling through the structured beam method. Laser scanning technology constructed high-density point cloud data by measuring surface echo time, reflecting the spatial geometry of the slope. Combined with structural information obtained from geological exploration, such as lithology distribution, joint surface distribution, and fault location, a comprehensive 3D slope model was finally generated in the modeling software, containing the 3D coordinates of the monitoring points, rock mass structural properties, and the spatial location of the monitoring nodes. The 3D model not only reflects the specific spatial location of the monitoring nodes in the slope but also includes rock mass medium properties (such as wave velocity and density) and geometric structural features (such as slope angle and distribution of densely jointed areas). Therefore, it can be used in subsequent analysis to establish waveform propagation path models and identify waveform in-phase characteristics between spatially adjacent monitoring points.

[0031] S203. Identify and remove the cooperative residual signal in the second microseismic waveform data to obtain the target second microseismic waveform data. The cooperative residual signal is a waveform segment with waveform phase characteristics between multiple spatially adjacent microseismic monitoring nodes in the three-dimensional model. In this embodiment, to eliminate non-genuine microseismic disturbance signals generated at multiple microseismic monitoring nodes during the spatial propagation of the blast wave, the second microseismic waveform data obtained in step S202 needs to undergo cooperative reverberation signal identification and elimination processing. Cooperative reverberation signals refer to waveform segments with similar waveform shapes and phase characteristics that are synchronously recorded at multiple monitoring points with similar paths when the blast wave propagates through the rock mass to multiple spatially adjacent monitoring points within a short period after the blast. These signals do not belong to independent microseismic events, but rather are background responses generated when the energy of the blast itself propagates through the rock mass in the form of approximately spherical or surface waves. If they are not identified and eliminated, they can easily interfere with the subsequent identification and classification of genuine microseismic events.

[0032] To achieve the identification of coordinated residual signals, firstly, based on the three-dimensional model obtained in step S202, the three-dimensional coordinate information of all microseismic monitoring nodes is extracted, and a spatial adjacency matrix based on Euclidean distance is constructed. The spatial adjacency matrix is ​​used to quantify the spatial distance relationship between any two monitoring points. A threshold distance is set (e.g., 5 to 10 meters), and monitoring point pairs with a distance below this threshold are defined as spatially adjacent point pairs. Subsequently, in the second microseismic waveform data, a sliding window correlation analysis is performed on the waveform data corresponding to all spatially adjacent point pairs. The correlation analysis uses the normalized cross-correlation coefficient calculation method, which calculates the cross-correlation value after standardizing the waveform segments within the corresponding time window to determine the degree of phase consistency between waveforms.

[0033] If, within a given time window, waveform segments between multiple spatially adjacent monitoring nodes all exhibit cross-correlation coefficients higher than a set threshold (e.g., 0.9) and share consistent phase fluctuation trends and frequency spectrum morphology, then the waveform segment can be identified as a cooperative reverberation signal. To enhance the robustness of identification, the system further introduces a waveform principal component analysis algorithm to reduce the dimensionality of waveform data from all monitoring nodes within this time window. If, after dimensionality reduction, the principal component contribution rate is significantly concentrated in the first one or two principal components, it indicates that most of the energy originates from the same source, which can also be considered a manifestation of cooperative response.

[0034] After identification, the system uses the time window of the coordinated reverberation signal as the elimination interval to remove the corresponding segments from the second microseismic waveform data. The elimination operation processes the waveform segments identified as coordinated reverberation signals while ensuring the temporal continuity and data structure consistency of the target second microseismic waveform data. Specific processing methods include linear interpolation, zero-filling, or waveform reconstruction based on a background noise model constructed from the first microseismic waveform data. The choice of processing method is determined based on the intended use of the target second microseismic waveform data in subsequent processing: when the target second microseismic waveform data will be used to establish a feature template for a blasting event, the blasting event modeling module requires preserving the amplitude variation trend and event boundary characteristics of the real waveform; therefore, a waveform reconstruction method based on a background noise model is preferred to maintain the continuity of amplitude statistical characteristics. When the target second microseismic waveform data will be used for event initiation determination and category identification, the event identification module focuses more on the abrupt changes and energy variations of the waveform in the time domain; therefore, linear interpolation can be used to preserve the smoothness of the temporal transition. If the target second microseismic waveform data is only used for source location inversion and propagation path fitting, the source location module mainly relies on waveform arrival time information and relative time delay; therefore, a data zero-filling strategy can be adopted to avoid introducing additional phase interference. By selecting the removal method according to the different sensitivities of waveform characteristics to different data processing modules, it is possible to remove cooperative reverberation interference while preserving the key waveform information required for different analysis tasks to the greatest extent.

[0035] S204. Remove common-mode blasting signals from the target second microseismic waveform data to obtain third microseismic waveform data. Common-mode blasting signals refer to waveform responses that radiate synchronously from the blast source to each monitoring point during the blasting process and exhibit similar energy distribution characteristics in multiple channels. They typically have a dominant vertical energy component. If not removed, they will interfere with the analysis and identification of real microseismic events. To achieve the above objective, the system calculates the three-component energy ratio of the target second microseismic waveform data using energy structure analysis methods, and identifies common-mode blasting signal waveform segments by combining them with preset blasting interference intervals, thereby achieving accurate removal of these signals and finally obtaining third microseismic waveform data for subsequent event detection and analysis. This may include steps S2041-S2042: S2041. Perform energy structure analysis on the target's second microseismic waveform data, and calculate the three-component energy ratio of the target's second microseismic waveform data. The three-component energy ratio is the ratio of the vertical energy component to the horizontal energy component of the target's second microseismic waveform data. In this embodiment, to achieve accurate identification and elimination of common-mode blasting signals in the target's second microseismic waveform data, it is necessary to first perform energy structure analysis on the waveform data, extract the energy distribution characteristics in the vertical and two horizontal directions, and calculate the three-component energy ratio, thereby establishing a basis for distinguishing the dominant direction of the blasting wave from the direction of natural microseismic events. The three-component energy ratio refers to the ratio of the total energy of the vertical component to the total energy of the two horizontal components. It usually exhibits a significant vertical energy bias in blasting activities, so this ratio can be used as an effective criterion for identifying blasting interference. To ensure that the energy ratio calculation results have physical interpretability and temporal resolution, it is necessary to identify time intervals with energy rebound characteristics in the waveform data and perform energy integration calculations within these intervals. The process may include the following steps: Perform amplitude envelope processing on the target vertical energy component in the target second microseismic waveform data to obtain a first envelope curve; sum the instantaneous energies of the two target horizontal energy components corresponding to each moment in the target second microseismic waveform data to obtain a horizontal energy sequence; perform amplitude envelope processing on the horizontal energy sequence to obtain a second envelope curve; perform morphological analysis on the first and second envelope curves respectively to identify the first energy reversal intervals where all energy decay trends on the first envelope curve reverse, and the second energy reversal intervals where all energy decay trends on the second envelope curve reverse. A reversal of the energy decay trend refers to the slope of the curve changing from negative or zero to positive; perform integral calculations on the first envelope curve within each first energy reversal interval to obtain multiple vertically injected energies, and perform integral calculations on the second envelope curve within each second energy reversal interval to obtain multiple horizontally injected energies; sum all vertically injected energies to obtain a vertical energy component, and sum all horizontally injected energies to obtain a horizontal energy component; calculate the ratio of the vertical energy component to the horizontal energy component to obtain the three-component energy ratio.

[0036] In the specific implementation process, the vertical component of the target's second microseismic waveform data is first subjected to amplitude envelope processing to obtain a first envelope curve reflecting the trend of vertical energy change over time. Amplitude envelope processing is a signal envelope extraction technique based on Hilbert transform. Its principle is to obtain the envelope curve by converting the original waveform into an analytic signal and then taking the modulus value sequence. This curve has high sensitivity to the detection of energy change trends. At the same time, the two horizontal components corresponding to each time point in the target's second microseismic waveform data are squared and summed to obtain the instantaneous horizontal energy at that moment. The same amplitude envelope processing is then applied to this horizontal energy sequence to generate a second envelope curve representing the trend of horizontal energy change.

[0037] By performing morphological analysis on the first and second envelope curves, the time intervals in which all energy decay trends reverse are further identified and defined as energy reversal intervals. Energy reversal refers to the portion of the envelope curve that originally showed a downward trend (i.e., a negative or zero slope) turning into an upward trend (positive slope) after a certain time point, usually indicating a new energy injection process within that interval. The energy reversal intervals identified in the vertical component are called the first energy reversal intervals, and those identified in the horizontal component are called the second energy reversal intervals. These reversal intervals can effectively reflect the energy injection window of blasting or microseismic events, avoiding the introduction of invalid or redundant background parts in energy calculations and improving the accuracy of energy ratio calculations.

[0038] After identifying all valid energy reversal intervals, the first envelope curve within each first energy reversal interval is integrated to obtain multiple vertically injected energies. Similarly, the second envelope curve within each second energy reversal interval is integrated to obtain multiple horizontally injected energies. The integration operation uses definite integration, numerically calculating the area under the envelope curve within each interval to obtain the cumulative energy per unit time. After completing the integration calculations for all energy injection segments, all vertically injected energies are summed to obtain the vertical energy component; all horizontally injected energies are summed to obtain the horizontal energy component. The ratio of these two components is used as the final three-component energy ratio to characterize the degree of bias in the vertical and horizontal directions of the energy distribution in the current waveform.

[0039] By implementing step S2041, high-precision analysis of the vertical and horizontal energy structures in the target second microseismic waveform data was achieved. It also provided a discriminant index with physical interpretability for the identification of common-mode blasting signals, enabling the overall early warning method to have a stronger adaptive identification capability for blasting disturbances, and laying the foundation for obtaining purer third microseismic waveform data in the future.

[0040] S2042. Determine the common-mode blasting signal in the second microseismic waveform data of the target based on the three-component energy ratio, and remove the common-mode blasting signal from the second microseismic waveform data of the target to obtain the third microseismic waveform data.

[0041] The preset blasting interference range refers to a period of time after the blast, typically set between 0 and 10 seconds after detonation. During this time, the main blast wave and its multiple reflections and scattered waves are still in the high-energy propagation stage, making it the region most prone to common-mode blasting signals. In specific operation, the system segments the target's second microseismic waveform data according to a fixed sliding window length (e.g., 1 second), obtains the three-component energy ratio corresponding to each sliding window segment based on step S2042, and compares it with a preset energy ratio threshold. The preset energy ratio threshold is typically set between 2.5 and 3.0 to distinguish the energy structure differences between normal microseismic events and blasting-dominated events.

[0042] If the energy ratio of the three components within a sliding window segment is greater than a preset energy ratio threshold, and the time segment is within a preset blasting interference range, then the waveform segment can be determined to belong to a common-mode blasting signal. The system uniformly marks this waveform segment as a common-mode blasting fragment on the three component data and performs a removal operation. The removal operation is not a simple deletion of data, but rather a repair process based on the data continuity requirements of subsequent analysis modules. Generally, linear interpolation or spectral reconstruction based on background noise characteristics is used to ensure that the removed waveform maintains a smooth transition on the time axis, avoiding the introduction of unnatural phase changes or energy abrupt changes. At the same time, the system retains the timestamp and removal mark information of the original waveform during the removal operation, which is convenient for subsequent traceability and the elimination of abnormal segments in model training.

[0043] Through the above step S2042, common-mode blasting signals caused by the concentrated release of blasting energy in the target second microseismic waveform data can be identified and removed. The obtained third microseismic waveform data has higher geological event credibility and waveform independence, providing a cleaner waveform basis for subsequent blasting signature construction and real microseismic event identification, thereby significantly improving the ability of the entire slope safety early warning system to identify weak abnormal responses under complex blasting backgrounds.

[0044] S205. Analyze the waveform characteristics of the second microseismic waveform data within the second preset time period, generate a blasting signature representing the blasting event, and save the blasting signature to the blasting noise dictionary. The blasting noise dictionary is a set of blasting signatures that contains multiple blasting signatures and is dynamically updated. In this application, an event refers to a waveform segment identified from continuous microseismic waveform data that has independent start and end times, energy abrupt change characteristics, and may represent a certain physical disturbance (such as blasting, microseismic events, interference, etc.). A blasting signature refers to the feature vector of a typical disturbance signal represented by a blasting event in terms of frequency domain, energy attenuation law, etc. It can reflect the propagation characteristics and time-domain attenuation characteristics of blasting disturbances under different modes, thus forming a standard template that can be identified and excluded. To ensure that the blasting signature has both time-frequency resolution and energy description capability, step S205 uses the Variational Mode Decomposition (VMD) method to extract modes from the original waveform, and combines amplitude envelope analysis, spectrum analysis, and feature screening mechanisms to achieve multi-dimensional modeling of blasting features. This may include the following steps: Based on the variational mode decomposition method, the second microseismic waveform data within a second preset time period is decomposed into multiple modal components, each corresponding to a center frequency and a finite bandwidth. For each modal component, the amplitude envelope of the modal component is calculated, and the amplitude envelope is fitted with exponential decay to obtain the first energy attenuation coefficient. The power spectrum of each modal component is calculated, and the ratio of the peak energy of the maximum power spectrum to the total energy is calculated to obtain the spectral concentration. The modal components with the first energy attenuation coefficient greater than or equal to a preset attenuation threshold, or whose spectral concentration is less than or equal to a preset concentration threshold, are summed point-by-point in the time domain to obtain the pure blasting waveform. The second energy attenuation coefficient and the centroid frequency of the power spectrum of the pure blasting waveform are calculated, and the second energy attenuation coefficient and the centroid frequency of the power spectrum are constructed into a two-dimensional feature vector, which is used as the blasting signature characterizing the blasting event.

[0045] In practical implementation, the system uses the second microseismic waveform data as input and performs variational mode decomposition (VMD) on the entire waveform signal within a second preset time period. VMD is an adaptive time-frequency analysis technique. Its core principle is to decompose the original signal into several intrinsic mode components with finite bandwidth. Each mode component corresponds to a center frequency, effectively distinguishing the energy distribution of different frequency band components in the signal. Compared with traditional empirical mode decomposition, VMD is more reliable in terms of mathematical stability and mode separation effect, and is suitable for processing complex waveform signals with severe spectral overlap and strong non-stationarity, such as blasting disturbances. After decomposition, the system obtains multiple mode components, each representing the local energy contribution within a certain frequency range of the waveform.

[0046] The system calculates the amplitude envelope for each modal component and fits it using an exponential function to obtain the first energy attenuation coefficient characterizing the signal energy decay rate. The amplitude envelope is obtained by extracting the modal value sequence after performing a Hilbert transform on the modal components. The exponential fitting curve is in the form of A·exp(-αt), where A is the initial amplitude, representing the maximum value or initial amplitude of the signal envelope at time t=0, reflecting the initial energy level of the signal; α is the energy attenuation coefficient, reflecting the intensity of energy decay over time; and t is the time variable. Simultaneously, the system calculates the power spectrum for each modal component and calculates the spectral concentration based on the frequency domain distribution, i.e., the ratio of the maximum peak energy in the power spectrum to the total modal energy. Spectral concentration reflects whether the modal energy is concentrated near a certain dominant frequency. Explosion signals typically have a high concentration of dominant frequency energy, while microseismic signals have a more dispersed energy distribution.

[0047] After obtaining the first energy attenuation coefficient and spectral concentration of each modal component, the system uses preset attenuation and concentration thresholds as screening criteria to select modal components whose energy attenuation coefficient is greater than or equal to the attenuation threshold, or whose spectral concentration is less than or equal to the concentration threshold, as the dominant characteristic modes of blasting disturbance. The attenuation threshold can be set based on the empirical statistical average of typical energy attenuation coefficients in historical blasting events, and the concentration threshold can be set as an empirical upper limit based on the spectral distribution characteristics of normal microseismic events. These modal components are then summed point-by-point to construct a clean blasting waveform, ensuring that the blasting signature does not contain local disturbance components from non-blasting sources, thus improving the representativeness and robustness of the template. After constructing the clean blasting waveform, to achieve a quantitative expression of the blasting event characteristics and provide a standard template for the subsequent establishment of a blasting noise dictionary, it is necessary to further extract feature parameters from the clean blasting waveform to obtain a two-dimensional feature vector that can be used to describe the essential characteristics of blasting disturbance. The two-dimensional feature vector consists of the second energy decay coefficient and the centroid frequency of the power spectrum, collectively referred to as the blast signature, which is used to characterize the core characteristics of the current blast event in two dimensions: time decay and frequency distribution.

[0048] The second energy attenuation coefficient is a parameter used to measure the rate of energy decay of a pure blast waveform in the time domain. First, a Hilbert transform is performed on the pure blast waveform to extract its amplitude envelope, forming an envelope curve reflecting the signal intensity's trend over time. Since blast waveforms typically exhibit an approximately exponential decay trend, the system fits this envelope curve to a function based on an exponential model. By taking the natural logarithm of the envelope curve and performing linear regression, the slope of the fitted line can be obtained, and thus the second energy attenuation coefficient can be calculated. The second energy attenuation coefficient reflects the concentration and attenuation rate of the blast signal energy release; a larger value indicates rapid energy decay in a short time, typically a characteristic of blast events.

[0049] On the other hand, to extract the frequency characteristics of the pure explosion waveform, the system performs a Fast Fourier Transform on the waveform signal to obtain its power spectral density curve. Based on this power spectral data, its power spectral centroid frequency is further calculated. The power spectral centroid frequency refers to the weighted average frequency of the entire spectrum energy distribution, used to measure the center position of the signal frequency distribution. It is calculated by multiplying each frequency component by its corresponding power value, summing the results, and then dividing by the total power. The specific formula is as follows: Among them, f c f represents the centroid frequency of the power spectrum. i Let P(f) represent the i-th frequency component. i The value represents the power spectrum corresponding to the frequency component, and N is the total number of frequency points. The centroid frequency of the power spectrum reflects the location of energy concentration of the blasting signal in the frequency domain. It is usually located in the mid-to-high frequency range and is an important parameter that distinguishes blasting events from low-frequency natural micro-vibrations.

[0050] Finally, the system constructs a two-dimensional eigenvector V = [α, f] by combining the calculated energy attenuation coefficient with the power spectrum centroid frequency. c This serves as the blast signature for the current blasting event. This blast signature is quantifiable, comparable, and archiveable, accurately characterizing typical patterns of blasting disturbances in the time-frequency space, providing standard samples for constructing a blasting noise dictionary. Subsequently, during microseismic waveform classification, vector similarity comparison can be used to match the third-party microseismic waveform data with the blast signature, thereby achieving rapid identification and elimination of blasting interference events and improving the overall system's identification accuracy and robustness.

[0051] After constructing the blasting signature, the system writes it as a new standard template into the blasting noise dictionary. The blasting noise dictionary can be managed using a local database, structured files, or distributed storage. The blasting noise dictionary is a set of structured feature templates used in the system to store historical blasting event characteristic information, supporting rapid comparison and elimination of blasting interference in subsequent microseismic event identification. Each entry in the dictionary is a blasting signature, containing a two-dimensional feature vector composed of the energy attenuation coefficient and the centroid frequency of the power spectrum extracted from the blasting event, along with metadata such as the blasting event's initiation time, the time range of the waveform segment used, the monitoring point number, modeling parameters, and the blasting cycle number. The blasting noise dictionary supports dynamic updates and access, continuously accumulating new blasting signatures as blasting operations continue, enabling continuous learning and template expansion of blasting disturbance characteristics under different blasting modes and geological conditions.

[0052] S206. Based on the first micro-seismic waveform data and the blasting noise dictionary, perform event detection on the third micro-seismic waveform data to obtain multiple events to be analyzed. Remove the blasting reverberation events from the multiple events to be analyzed to obtain the events to be alarmed. To accurately identify real geological anomalies and effectively eliminate blasting interference, the system, after obtaining the third microseismic waveform data after removing common-mode interference, further performs event detection and blasting reverberation screening operations. Step S206 comprehensively utilizes the first microseismic waveform data and the constructed blasting noise dictionary, and triggers and segments the third microseismic waveform data by setting a dynamic trigger threshold, extracting multiple events to be analyzed and generating a corresponding event signature for each event. Subsequently, the system calculates the similarity between the event signatures and all blasting signatures in the blasting noise dictionary, identifies blasting reverberation events highly similar to typical blasting patterns, and removes them from the set of events to be analyzed, ultimately obtaining a set of events with higher credibility and practical early warning significance. This may include steps S2061-S2062: S2061. Based on the first microseismic waveform data, a dynamic trigger threshold is set, the third microseismic waveform data is segmented to obtain multiple events to be analyzed, and the waveform characteristics of the events to be analyzed are analyzed to generate event signatures for the events to be analyzed. One event to be analyzed corresponds to one event signature. The dynamic trigger threshold includes a trigger threshold and an end threshold. After constructing the third microseismic waveform data and blasting noise dictionary, to accurately identify potential abnormal events in the microseismic data, step S2061 involves segmenting the third microseismic waveform data and constructing a structured event description for each event. Step S2061 combines the statistical characteristics of the first microseismic waveform data to set a set of dynamic trigger thresholds that adapt to changes in the current monitoring environment. Based on this, multiple independent signal segments that may contain geological anomalies or blasting disturbances are extracted from the third microseismic waveform data as multiple events to be analyzed. Furthermore, the system models the waveform characteristics of each event segment, generating a corresponding event signature, providing basic data support for subsequent blasting reverberation identification and alarm event screening. This may include the following steps: A sliding window calculation is performed on the first microseismic waveform data based on a preset time window length to obtain a time series of the ratio of the short-time window energy mean to the long-time window energy mean in the first microseismic waveform data. The mean and standard deviation of the time series are calculated, and the mean and standard deviation are combined to form a background noise reference parameter set characterizing normal environmental fluctuations. Based on the background noise reference parameter set and a preset calculation formula, a trigger threshold and a termination threshold are calculated to obtain a dynamic trigger threshold. Based on the dynamic trigger threshold, third microseismic waveform data segments that meet preset conditions are identified in the third microseismic waveform data to obtain multiple events to be analyzed.

[0053] To achieve dynamic trigger threshold setting, the system first extracts continuous records within a certain time period from the first microseismic waveform data as background samples, and then performs energy characteristic analysis based on this data. To this end, the system introduces an energy ratio calculation method based on a sliding window to measure the energy abrupt change characteristics of the signal at different time scales. Specifically, at each sliding time point, two fixed window lengths are defined: a short window and a long window. The short window is used to capture the instantaneous energy change of the signal, and the long window is used to estimate the current background energy level. At each time point t, the system calculates the energy ratio from tT... s The energy within a short time window up to t is denoted as E. s (t), and simultaneously calculate from tT l The energy within a long time window up to t is denoted as E. l (t). The energy is calculated using the signal square integral, i.e.: Where s(i) is the waveform amplitude at the i-th sampling point, T s With T l These represent the lengths of the short and long time windows, respectively (in units of sampling points or time). Then, the energy ratio is calculated: Here, ε is a tiny positive number to avoid a denominator of zero. The energy ratio R(t) reflects the intensity of the change in the instantaneous energy of the signal at the current time point relative to the historical background energy. When R(t) is much greater than 1, it indicates that the short-term energy is significantly higher than the background, which may correspond to the onset of a sudden microseismic event.

[0054] The system performs sliding calculations on the entire first microseismic waveform data to obtain a complete energy ratio time series, and then performs statistical analysis based on this energy ratio time series. Specifically, the system calculates the mean μ of this series. R and standard deviation σ R and (μ) R , σ R The parameter set constituted by these parameters is defined as the background noise reference parameter set, used to characterize the statistical regularity of energy fluctuations under normal, disturbance-free conditions. Since the first microseismic waveform data typically originates from long-term monitoring records without anomalies, its statistical characteristics can serve as a reliable reference for the environmental noise state. Next, based on this reference parameter set and preset sensitivity coefficients, the system calculates the dynamic trigger threshold, which includes two parts: the trigger threshold R... trigger With the end threshold R end The trigger threshold is used to determine the start of an abnormal event, and the end threshold is used to determine the termination time of the event. The specific calculation method can be as follows: R trigger =μ R +k1·σ R R end =μ R +k2·σ RWhere k1 and k2 are empirically set sensitivity control coefficients, satisfying k1>k2 to avoid false triggering caused by event boundary oscillations. After obtaining the dynamic trigger threshold, the system applies it to the third microseismic waveform data to determine the event to be analyzed.

[0055] After setting the background noise benchmark parameter set and dynamic trigger threshold based on the first microseismic waveform data, the system further applies this dynamic trigger mechanism to the third microseismic waveform data to identify time segments that may contain anomalous disturbances and extract them as multiple events to be analyzed. To this end, the system first needs to process the third microseismic waveform data in the same way as the first waveform data, i.e., perform rolling calculations based on the same preset time window length to generate a long-short time window energy ratio sequence corresponding to the data. The main function of this step is to transform the original continuous seismic waveform signal into a feature sequence reflecting the local energy change trend, thereby providing a quantifiable basis for subsequent threshold-based determination of event start and end points. This may include the following steps: performing rolling calculations on the long-short time window energy ratio of the third microseismic waveform data based on a preset time window length to obtain a long-short time window energy ratio sequence; The time series corresponding to the third microseismic waveform data is traversed. The time point when the energy ratio of the long and short time windows in the long and short time window energy ratio series crosses the trigger threshold is determined as the start time point of the event to be analyzed. The time point when the energy ratio of the long and short time windows crosses the end threshold for the first time after the start time point is determined as the end time point of the event to be analyzed. The third microseismic waveform data segment between the start time point and the end time point is taken as an event to be analyzed. This process is repeated until the entire time series is traversed.

[0056] In its implementation, the system uses a sliding window to calculate the ratio of short-time window energy to long-time window energy point by point on the third microseismic waveform data. The window lengths of the short and long time windows remain consistent with the previous steps; for example, the short-time window length is 0.2 seconds and the long-time window length is 1 second. At each sliding position, waveform segments of both the short and long time windows ending at that time point are extracted, and their energy values ​​are calculated. The energy calculation method still uses the square integral of the signal, i.e., summing the squares of the waveform amplitudes. Subsequently, the ratio of short-time window energy to long-time window energy is calculated to obtain the energy ratio R(t) corresponding to that time point. By continuously sliding the window across the entire waveform sequence, the system ultimately forms a sequence of long and short time window energy ratios corresponding to the time axis of the third microseismic waveform data. This sequence reflects the energy abrupt change behavior of the signal in the local time domain and is a key intermediate feature for event identification.

[0057] After obtaining the energy ratio sequence of long and short time windows, the system performs event boundary identification based on a pre-set dynamic trigger threshold. This involves traversing the entire ratio sequence to find interaction points with the trigger and end thresholds. During this traversal, when the energy ratio at time point t1 changes from below the trigger threshold to above it (i.e., an upward crossing occurs), the system marks this time point as the start time point of an event to be analyzed. Subsequently, the system continues scanning the sequence. When the energy ratio at subsequent time point t2 first drops from above the end threshold to below it (i.e., a downward crossing occurs), the system marks this time point as the end time point of the event. The time range between the start and end time points constitutes a valid segment that may contain microseismic signals or blast reverberation, and the corresponding waveform data is extracted as an event to be analyzed.

[0058] The core advantage of this event boundary identification mechanism lies in its full utilization of the energy ratio's sensitive response to local anomalies. The overpass trigger threshold signifies that the signal energy is momentarily significantly higher than the background level, possessing the physical characteristics of an event initiation; while the underpass termination threshold indicates that the event has decayed and entered a stable state, possessing the characteristics of event termination. Through this overpass-underpass strategy, the system can effectively identify disturbance events with complete start-end structures while filtering out short-term, meaningless energy fluctuations.

[0059] In practical applications, for example, in the energy ratio sequence obtained after rolling calculation of a segment of third microseismic waveform data, the ratio at time point t1 suddenly rises from 1.5 to 2.3, exceeding the currently set trigger threshold of 1.9. The system marks t1 as the event start time point. Subsequently, at t1+2.5s, the ratio first drops to 1.6, below the end threshold of 1.7, and the system marks this point as the event end point. The waveform data segment from t1 to t1+2.5s is then extracted as an event to be analyzed, used for subsequent event signature generation and explosion aftershock comparison analysis.

[0060] In another feasible implementation, the event to be analyzed can be determined by the following method: When the detonation time of the electronic detonator is identified as a sequential blasting mode containing multiple sub-detonation times, the target second microseismic waveform data within a preset template duration is extracted based on the first sub-detonation time as a reference, and used as a meta-blasting waveform template characterizing a single sub-blasting event; based on the precise time information of all sub-detonation times, the meta-blasting waveform template is time-shifted and superimposed to construct a theoretical synthetic blasting chain waveform that does not contain microseismic events; the third microseismic waveform data is subtracted from the synthetic blasting chain waveform in the time domain to obtain a residual signal, which theoretically eliminates the energy generated by the sequential blasting itself; based on the first microseismic waveform data, a dynamic triggering threshold is set, and the residual signal, rather than the original third microseismic waveform data, is segmented to obtain multiple events to be analyzed.

[0061] Specifically, when the system detects that the electronic detonator's detonation mode is a sequential blasting involving multiple sub-detonation moments—that is, in a blasting cycle, each blasting point is detonated sequentially at a preset interval—a series of superimposed blasting signals are formed in the microseismic waveform. The microseismic waveform structure in this mode is often complex, and simply relying on full waveform feature matching is insufficient to effectively eliminate blasting disturbances. Therefore, the system proposes a method based on time template alignment and waveform superposition modeling to construct a theoretical blasting chain waveform, and uses this as a basis to extract residual signals from the third microseismic waveform data.

[0062] In the actual process, the system first uses the identified first sub-detonation time as the time reference and extracts a segment of corresponding target second microseismic waveform data with a preset template duration (e.g., 200ms to 300ms). This data segment typically covers the complete response process caused by a single blast point, including the explosion initiation, peak, tail, and energy decay process, and is representative, therefore defined as the meta-blast waveform template. This template is equivalent to the typical response characteristics of a single sub-blast event on the sensor and is the basic unit for subsequent blast chain reconstruction. Next, based on all sub-detonation time sequences provided by the electronic detonator control system, the system performs equal-amplitude time-shifted replication of the above meta-blast waveform template according to each sub-detonation time, and linearly superimposes all time-shifted templates in the time domain to construct the synthetic blast chain waveform generated by continuous blasting. This synthetic waveform does not contain any actual geological anomaly response and only reflects the mechanical disturbance and energy release generated by the blasting system itself, making it an ideal representation of the blasting background energy.

[0063] After constructing the synthetic waveform, the system subtracts the third microseismic waveform data from the synthetic blasting chain waveform point by point along the time axis to obtain the residual signal. Theoretically, this residual signal has eliminated the energy term generated by the sequence blasts themselves, retaining only the portion of the actual observation data not interpreted by the synthetic waveform, which may include components such as real geological anomalies, boundary reflections, or equipment response errors. Therefore, the residual signal has higher geological event sensitivity than the original waveform data and can more effectively distinguish between blasting and non-blasting events.

[0064] After obtaining the residual signal, the system no longer performs event detection based on the original third microseismic waveform data. Instead, it uses the residual signal as input and calls the aforementioned dynamic trigger threshold mechanism for event segmentation. The specific method is the same as the aforementioned implementation method, that is, it uses a sliding window to calculate the ratio sequence of short-time window energy to long-time window energy, and identifies the start and end points of events based on trigger thresholds and end thresholds. Finally, it extracts multiple microseismic segments that still have significant energy abrupt changes after the blasting energy is removed, as events to be analyzed.

[0065] S2062. Calculate the feature difference between the event signature and all blast signatures in the blast noise dictionary, and remove the blast reverberation events in the events to be analyzed to obtain the events to be alarmed. The blast reverberation events are at least one event to be analyzed whose feature difference is less than the preset difference threshold.

[0066] After extracting the events to be analyzed and generating event signatures, in order to further identify and remove non-geological disturbance signals caused by the blasting process, step S2062 introduces a feature difference comparison mechanism based on the event signature and the blasting signature in the blasting noise dictionary to construct an event discrimination model, thereby completing the identification and filtering of blasting reverberation events.

[0067] In its implementation, the system first processes the event signature of each event to be analyzed into a feature vector. The event signature contains the same two-dimensional feature parameters as the blasting signature: the energy attenuation coefficient and the power spectrum centroid frequency. The energy attenuation coefficient reflects the energy release and attenuation characteristics of the event waveform in the time domain, while the power spectrum centroid frequency describes the energy concentration location of the event in the frequency domain. This two-dimensional vector structure effectively captures the typical propagation characteristics of microseismic events and forms a quantifiable basis for comparison with the common characteristics of blasting events in the time and frequency domains. The generation principle of the event signature is the same as that of the blasting signature, and will not be described in detail here.

[0068] The system then retrieves all blasting signatures stored in the blasting noise dictionary and calculates the similarity for each signature of the event to be analyzed. Euclidean distance can be used to calculate the Euclidean distance between the signature of the event to be analyzed and each blasting signature stored in the blasting noise dictionary, obtaining the feature difference between the event signature of the event to be analyzed and each blasting signature. This feature difference is compared to a preset difference threshold. If at least one blasting signature has a feature difference less than the preset difference threshold, the event is considered highly similar to blasting characteristics, indicating that its propagation characteristics are highly likely to be caused by blasting disturbances. The system marks this as a blasting reverberation event and removes it from the event set to be analyzed. Conversely, if the feature difference is greater than the preset difference threshold, it indicates that the event differs significantly from typical blasting characteristics and is retained as a potentially real geological anomaly event, included in the set of events to be alerted.

[0069] S207. Within the preset monitoring window, perform a source location closure test on each event to be alarmed. If the source location closure test is passed, the event to be alarmed is confirmed as a real micro-seismic event, and a danger alarm is issued.

[0070] After eliminating the explosion reverberations of the events to be analyzed and obtaining the final set of events to be alarmed, in order to further ensure that the identified events have real spatial source characteristics and eliminate false signals caused by factors such as noise, equipment false alarms or boundary reflections, a source location closure verification mechanism is introduced in step S207. The source solvability of each event to be alarmed is judged within a preset monitoring window, thereby finally confirming the authenticity of the event and triggering a danger alarm.

[0071] In the specific implementation process, the system first calls up the corresponding third microseismic waveform data segment for each event to be alarmed, and then, in conjunction with multiple microseismic monitoring nodes deployed at the target slope site, extracts the initial motion time of each node receiving the event signal. The initial motion time refers to the time point at which a monitoring node first detects the energy change of the event. It can be determined by the waveform change point within the corresponding time period in the aforementioned sliding window energy ratio sequence, or by automatically identifying it by calculating the change of the first derivative of the waveform envelope. Since all monitoring nodes have undergone unified time calibration through the Global Navigation Satellite System, the slight differences in the initial motion time between different nodes can be directly used for source location calculation.

[0072] Based on the 3D model obtained in step S202, which includes information such as the slope's spatial geometry, the 3D coordinates of monitoring nodes, and the properties of the rock mass, the system establishes a set of source location equations, combined with the initial motion times extracted from each monitoring node. This set of equations uses the source location coordinates in 3D space and their initial time as the variables to be solved, and calculates the theoretical propagation time difference based on the propagation path length between each monitoring node and the assumed source location, as well as the rock mass wave velocity. The system uses the least squares method to numerically solve this set of equations, obtaining an optimal source location solution that satisfies the current observation data.

[0073] To verify the physical rationality of the source solution, the system uses this solution as a predicted source and derives the theoretical time at which each monitoring node should receive the initial motion signal in the 3D model. This theoretical time is then compared point-by-point with the actual extracted initial motion time. If the average error between the theoretical and actual arrival times of all monitoring nodes is within the set time tolerance (e.g., less than 2 milliseconds), it indicates that the current event has a clear spatial source structure, and the calculated propagation path highly matches the actual response. Based on this, the system confirms the event to be alerted as a genuine microseismic event. If the average error exceeds the tolerance, or some monitoring nodes cannot match the waveform's initial motion time, or the solution process fails to converge or becomes unstable, the current event is determined to lack a closable source structure and is highly likely to be a disturbing microseismic event. The system removes this event from the alert event set without triggering a danger alarm.

[0074] For example, after a slope blasting operation, the system identified a potential alarm event in step S206, where the waveform showed a clear initial motion response in all six microseismic monitoring nodes. Based on the 3D model and the 3D coordinates of the monitoring nodes, combined with the initial motion time of each node for this event, the system constructed and solved a source location model, determining that the source location was 14.7 meters below the slope toe. The average error between the theoretical propagation time and the actual observation time was 1.2 milliseconds, meeting the current accuracy requirements. The system confirmed this event as a genuine microseismic event and pushed it to the danger alarm module. Conversely, another event showed a low-amplitude signal only in two monitoring nodes, and the source location model had no solution; the system marked it as an interfering microseismic event.

[0075] Please see Figure 3 This is a schematic diagram of the structure of a slope safety early warning system provided in an embodiment of this application. The slope safety early warning system 300 specifically includes: The first acquisition module 301 is used to acquire the detonation time of the electronic detonator in the target slope and the first microseismic waveform data of the target slope within a first preset time period before the detonation time; the second acquisition module 302 is used to acquire the second microseismic waveform data of the target slope within a second preset time period after the detonation time and the three-dimensional model of the target slope; the first elimination module 303 is used to identify and eliminate the cooperative reverberation signal in the second microseismic waveform data to obtain the target second microseismic waveform data, wherein the cooperative reverberation signal is a waveform segment with waveform phase characteristics between multiple spatially adjacent microseismic monitoring nodes in the three-dimensional model; the second elimination module 304 is used to eliminate the common-mode blasting signal in the target second microseismic waveform data to obtain the third microseismic waveform data; analysis Module 305 is used to analyze the waveform characteristics of the second microseismic waveform data within a second preset time period, generate a blasting signature representing the blasting event, and save the blasting signature to the blasting noise dictionary; Detection module 306 is used to perform event detection on the third microseismic waveform data based on the first microseismic waveform data and the blasting noise dictionary, obtain multiple events to be analyzed, remove blasting reverberation events from the multiple events to be analyzed, and obtain events to be alarmed; Verification module 307 is used to perform source location closure verification on each event to be alarmed within a preset monitoring window. If the source location closure verification is passed, the event to be alarmed is confirmed as a real microseismic event, and a danger alarm is issued; if the source location closure verification is not passed, the event to be alarmed is confirmed as an interfering microseismic event.

[0076] Optionally, the second elimination module 304 is specifically used for: performing energy structure analysis on the target second microseismic waveform data, calculating the three-component energy ratio of the target second microseismic waveform data, the three-component energy ratio being the ratio of the vertical energy component to the horizontal energy component of the target second microseismic waveform data; eliminating the common-mode blasting signal in the target second microseismic waveform data to obtain the third microseismic waveform data, the common-mode blasting signal being a waveform segment whose three-component energy ratio is within a preset blasting interference range.

[0077] Optionally, the second elimination module 304 is further specifically used for: performing amplitude envelope processing on the target vertical energy component in the target second microseismic waveform data to obtain a first envelope curve; summing the instantaneous energies of the two target horizontal energy components corresponding to each moment in the target second microseismic waveform data to obtain a horizontal energy sequence; performing amplitude envelope processing on the horizontal energy sequence to obtain a second envelope curve; performing morphological analysis on the first envelope curve and the second envelope curve respectively to identify the first energy reversal interval where all energy decay trends on the first envelope curve reverse, and all energy decay intervals on the second envelope curve... The second energy reversal interval is where the trend reverses. A reversal of the energy decay trend means that the slope of the curve changes from negative or zero to positive. Integrating the first envelope curve within each first energy reversal interval yields multiple vertically injected energies. Integrating the second envelope curve within each second energy reversal interval yields multiple horizontally injected energies. All vertically injected energies are summed to obtain the vertical energy component, and all horizontally injected energies are summed to obtain the horizontal energy component. The ratio of the vertical energy component to the horizontal energy component is calculated to obtain the three-component energy ratio.

[0078] Optionally, the analysis module 305 is specifically used for: decomposing the second microseismic waveform data within a second preset time period into multiple modal components based on the variational mode decomposition method, each modal component corresponding to a center frequency and a finite bandwidth; for each modal component, calculating the amplitude envelope of the modal component and performing exponential decay fitting on the amplitude envelope to obtain a first energy attenuation coefficient; calculating the power spectrum of each modal component and calculating the ratio of the peak energy of the maximum power spectrum to the total energy to obtain the spectral concentration; summing the modal components whose first energy attenuation coefficient is greater than or equal to a preset attenuation threshold, or whose spectral concentration is less than or equal to a preset concentration threshold, point-by-point in the time domain to obtain a pure blasting waveform; calculating the second energy attenuation coefficient and the centroid frequency of the power spectrum of the pure blasting waveform, and constructing the second energy attenuation coefficient and the centroid frequency of the power spectrum into a two-dimensional feature vector, using the two-dimensional feature vector as a blasting signature characterizing the blasting event.

[0079] Optionally, the detection module 306 is specifically used for: setting a dynamic trigger threshold based on the first microseismic waveform data, segmenting the third microseismic waveform data to obtain multiple events to be analyzed, analyzing the waveform characteristics of the events to be analyzed, generating an event signature for each event to be analyzed, with one event signature corresponding to each event to be analyzed, and the dynamic trigger threshold including a trigger threshold and an end threshold; calculating the feature difference between the event signature and all blasting signatures in the blasting noise dictionary, and removing blasting reverberation events from the events to be analyzed to obtain events to be alarmed, wherein the blasting reverberation event is at least one event to be analyzed with a feature difference less than a preset difference threshold.

[0080] Optionally, the detection module 306 is further specifically used for: performing sliding window calculation on the first microseismic waveform data based on a preset time window length to obtain a time series of the ratio of the short-time window energy mean to the long-time window energy mean in the first microseismic waveform data; calculating the mean and standard deviation of the time series, and combining the mean and standard deviation into a background noise reference parameter set characterizing normal environmental fluctuations; calculating the trigger threshold and termination threshold based on the background noise reference parameter set and a preset calculation formula to obtain a dynamic trigger threshold; and identifying third microseismic waveform data segments in the third microseismic waveform data that meet preset conditions based on the dynamic trigger threshold to obtain multiple events to be analyzed.

[0081] Optionally, the detection module 306 is further specifically used for: performing rolling calculations on the energy ratio of long and short time windows on the third microseismic waveform data based on a preset time window length to obtain a sequence of long and short time window energy ratios; traversing the time series corresponding to the third microseismic waveform data, determining the time point corresponding to when the energy ratio of long and short time windows in the sequence crosses the trigger threshold as the start time point of the event to be analyzed, determining the time point corresponding to when the energy ratio of long and short time windows first crosses the end threshold after the start time point as the end time point of the event to be analyzed, and taking the segment of third microseismic waveform data between the start time point and the end time point as an event to be analyzed, until the entire time series has been traversed.

[0082] It should be noted that the device provided in the above embodiments is only illustrated by the division of the above functional modules when implementing its functions. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0083] This embodiment also discloses an electronic device, referring to... Figure 4The electronic device may include: at least one processor 401, at least one communication bus 402, a user interface 403, a network interface 404, and at least one memory 405. The communication bus 402 is used to enable communication between these components. The user interface 403 may include a display screen and a camera; optionally, the user interface 403 may also include a standard wired interface or a wireless interface. The network interface 404 may optionally include a standard wired interface or a wireless interface. The processor 401 may include one or more processing cores. The processor 401 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 405, and by calling data stored in the memory 405. The memory 405 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 405 may also be at least one storage device located remotely from the processor 401. Figure 4 As shown, the memory 405, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a slope safety early warning method. Figure 4 In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 401 can be used to call an application program stored in the memory 405 for a slope safety early warning method. When executed by one or more processors 401, the electronic device executes one or more methods as described in the above embodiments.

[0084] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0085] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will be readily apparent to those skilled in the art upon consideration of the disclosure in this specification. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for slope safety early warning, characterized in that, Applied in a server, the method comprises: Obtaining the initiation time of an electronic detonator in a target slope and first microseismic waveform data of the target slope within a first preset time length before the initiation time; Obtaining second microseismic waveform data of the target slope within a second preset time length after the initiation time and a three-dimensional model of the target slope; Identifying and removing a coordinated residual signal in the second microseismic waveform data to obtain target second microseismic waveform data, the coordinated residual signal being a waveform segment having a waveform in-phase feature between multiple microseismic monitoring nodes that are spatially adjacent in the three-dimensional model; Removing a common-mode blasting signal in the target second microseismic waveform data to obtain third microseismic waveform data; Analyzing waveform features of the second microseismic waveform data within the second preset time length to generate a blasting signature representing a blasting event, and saving the blasting signature to a blasting noise dictionary, the blasting noise dictionary being a dynamic updating blasting signature set containing multiple blasting signatures; Performing event detection on the third microseismic waveform data based on the first microseismic waveform data and the blasting noise dictionary to obtain multiple events to be analyzed, removing a blasting residual signal event from the multiple events to be analyzed to obtain an event to be alarmed; Within a preset monitoring window, performing a source location closed-loop test on each event to be alarmed, and if the source location closed-loop test is passed, confirming that the event to be alarmed is a real microseismic event and performing a danger alarm.

2. The method of claim 1, wherein, The removing of the common-mode blasting signal in the target second microseismic waveform data to obtain the third microseismic waveform data specifically comprises: Performing energy structure analysis on the target second microseismic waveform data to calculate a three-component energy ratio of the target second microseismic waveform data, the three-component energy ratio being a ratio of a vertical energy component to a horizontal energy component of the target second microseismic waveform data; Determining the common-mode blasting signal in the target second microseismic waveform data according to the three-component energy ratio, and removing the common-mode blasting signal from the target second microseismic waveform data to obtain the third microseismic waveform data.

3. The method of claim 2, wherein, The energy structure analysis on the target second microseismic waveform data to calculate the three-component energy ratio of the target second microseismic waveform data specifically comprises: Performing amplitude envelope processing on a target vertical energy component in the target second microseismic waveform data to obtain a first envelope curve, and performing summation on instantaneous energies of two target horizontal energy components corresponding to each time in the target second microseismic waveform data to obtain a horizontal energy sequence, and performing amplitude envelope processing on the horizontal energy sequence to obtain a second envelope curve; Performing morphological analysis on the first envelope curve and the second envelope curve respectively to identify a first energy reversal interval in which all energy attenuation trends on the first envelope curve are reversed, and a second energy reversal interval in which all energy attenuation trends on the second envelope curve are reversed, the energy attenuation trend being reversed being that a slope of the curve is changed from a negative value or zero to a positive value; respectively, to obtain a plurality of vertical injection energies, and respectively integrating each of the second envelope curves within each of the second energy reversal intervals to obtain a plurality of horizontal injection energies; accumulatively summing all the vertical injection energies to obtain the vertical energy component, and accumulatively summing all the horizontal injection energies to obtain the horizontal energy component; calculating the ratio of the vertical energy component to the horizontal energy component to obtain the three-component energy ratio.

4. The method of claim 1, wherein, The analysis of the waveform characteristics of the second microseismic waveform data within the second preset time length generates a blasting signature representing a blasting event, specifically comprising: decompose the second microseismic waveform data within the second preset time length into a plurality of modal components based on a variational mode decomposition method, each of the modal components corresponding to a center frequency and a limited bandwidth; for each of the modal components, calculate the amplitude envelope of the modal component and perform exponential decay fitting on the amplitude envelope to obtain a first energy decay coefficient; calculate the power spectrum of each of the modal components, and calculate the ratio of the maximum power spectrum peak energy to the total energy in the power spectrum to obtain a spectral concentration degree; point-by-point summing in the time domain the modal components whose first energy decay coefficients are greater than or equal to a preset decay threshold or whose spectral concentration degrees are less than or equal to a preset concentration threshold to obtain a pure blasting waveform; calculate the second energy decay coefficient and the power spectrum centroid frequency of the pure blasting waveform, and construct the second energy decay coefficient and the power spectrum centroid frequency into a two-dimensional feature vector, and take the two-dimensional feature vector as the blasting signature representing a blasting event.

5. The method of claim 1, wherein, The event detection of the third microseismic waveform data based on the first microseismic waveform data and the blasting noise dictionary obtains a plurality of events to be analyzed, and removes blasting reverberation events from the plurality of events to be analyzed to obtain events to be alarmed, specifically comprising: setting a dynamic trigger threshold based on the first microseismic waveform data, segmenting the third microseismic waveform data to obtain a plurality of events to be analyzed, and analyzing the waveform characteristics of the events to be analyzed to generate event signatures of the events to be analyzed, one of the events to be analyzed corresponding to one of the event signatures, the dynamic trigger threshold including a trigger threshold and an end threshold; calculating the feature difference degree of the event signatures and all the blasting signatures in the blasting noise dictionary, and removing blasting reverberation events from the events to be analyzed to obtain events to be alarmed, the blasting reverberation events being at least one of the events to be analyzed whose feature difference degree is less than a preset difference degree threshold.

6. The method of claim 5, wherein, The setting of a dynamic trigger threshold based on the first microseismic waveform data, the segmentation of the third microseismic waveform data to obtain a plurality of events to be analyzed, specifically comprising: performing sliding window calculation on the first microseismic waveform data based on a preset time window length to obtain a time series of long-short time window energy ratios of short-time window energy mean and long-time window energy mean in the first microseismic waveform data; calculate a mean and a standard deviation of the time series, combine the mean and the standard deviation into a background noise reference parameter set representing normal environmental fluctuations; calculate the trigger threshold and the end threshold based on the background noise reference parameter set and a pre-designed calculation formula, to obtain the dynamic trigger threshold; identify third microseismic waveform data segments meeting a preset condition in the third microseismic waveform data based on the dynamic trigger threshold, to obtain a plurality of the to-be-analyzed events.

7. The method of claim 6, wherein, The identifying third microseismic waveform data segments meeting a preset condition in the third microseismic waveform data based on the dynamic trigger threshold, to obtain a plurality of the to-be-analyzed events, specifically includes: perform rolling calculation of long-short time window energy ratios on the third microseismic waveform data based on the preset time window length, to obtain a long-short time window energy ratio sequence; traverse the time series corresponding to the third microseismic waveform data, determine a time point corresponding to when the long-short time window energy ratio in the long-short time window energy ratio sequence crosses the trigger threshold from above as a starting time point of the to-be-analyzed event, determine a time point corresponding to when the long-short time window energy ratio first crosses the end threshold from below after the starting time point as an ending time point of the to-be-analyzed event, and take the third microseismic waveform data segment between the starting time point and the ending time point as one to-be-analyzed event, until the entire time series is traversed.

8. A slope safety warning system, characterized by, including: a first acquisition module configured to acquire a detonation time of an electronic detonator in a target slope and first microseismic waveform data of the target slope within a first preset time length before the detonation time; a second acquisition module configured to acquire second microseismic waveform data of the target slope within a second preset time length after the detonation time and a three-dimensional model of the target slope; a first elimination module configured to identify and eliminate a cooperative residual signal in the second microseismic waveform data, to obtain target second microseismic waveform data, the cooperative residual signal being a waveform segment having a waveform in-phase feature between a plurality of microseismic monitoring nodes that are spatially adjacent in the three-dimensional model; a second elimination module configured to eliminate a common-mode blasting signal in the target second microseismic waveform data, to obtain third microseismic waveform data; an analysis module configured to analyze waveform features of the second microseismic waveform data within the second preset time length, to generate a blasting signature representing a blasting event, and to save the blasting signature into a blasting noise dictionary, the blasting noise dictionary being a dynamically updated blasting signature set containing a plurality of the blasting signatures; a detection module configured to perform event detection on the third microseismic waveform data based on the first microseismic waveform data and the blasting noise dictionary, to obtain a plurality of to-be-analyzed events, and to eliminate a blasting residual signal event from the plurality of to-be-analyzed events, to obtain a to-be-alarmed event; a verification module configured to perform a source location closure verification on each to-be-alarmed event within a preset monitoring window, and if the to-be-alarmed event passes the source location closure verification, to confirm that the to-be-alarmed event is a real microseismic event and to perform a danger alarm.

9. An electronic device, comprising: including: one or more processors and a memory; The memory is coupled with the one or more processors, and is configured to store computer program codes, the computer program codes comprising computer instructions, which are invoked by the one or more processors to cause the electronic device to perform the method according to any one of claims 1-7.

10. A computer-readable storage medium comprising instructions, characterized in that, The instructions, when executed on an electronic device, cause the electronic device to perform the method according to any one of claims 1-7.

Images