Rock mass blasting disturbance risk real-time early warning method

Abstract

This application provides a real-time early warning method for rock mass blasting disturbance risk, comprising: deploying multi-point displacement probes and microseismic pickup units on the rock mass surface to collect displacement data, vibration waveform signals, and surface strain distribution data, and recording the location of the first detachment point and the time of each subsequent detachment; determining the detachment distance from the first detachment point to the geometric center of the rock mass based on the location of the first detachment point, identifying the time interval between two adjacent detachments based on the time of each subsequent detachment, and using the ratio of the detachment distance to the time interval as an adjustment coefficient for the event separation threshold; when the time interval between two adjacent detachments is less than the preset event separation threshold, collecting vibration signals, identifying the overlapping parts in the vibration signals and denoising them to obtain the start and end time boundaries of a single detachment event; determining the coordinates of the boundary points based on the preliminary outline of the collapse impact range, identifying the areas with missing data in the strain concentration zone, filling the missing areas using interpolation, and generating a complete damage area distribution map.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for real-time early warning of rock mass blasting disturbance risk. Background Technology

[0002] Real-time early warning of rock mass blasting disturbance risks is one of the core areas of safety control in underground engineering projects such as mines, tunnels, and water conservancy projects. Its importance lies in its direct impact on the safety of workers and the overall stability of the engineering structure. Inaccurate early warnings can lead to large-scale collapse accidents, causing serious casualties and economic losses. Currently, most early warning methods rely on the location and distribution of debris falls to determine the subsequent damage trend. It is generally believed that the closer the initial debris fall point is to the geometric center of the rock mass, the more concentrated the subsequent damage area will be and the clearer its boundaries will be.

[0003] This static judgment based on spatial location reveals significant shortcomings in actual engineering: when multiple detachments occur consecutively in the central area within a very short period, the vibration signals generated between adjacent detachment events severely overlap, making it difficult for the monitoring system to clearly distinguish the start and end times of each independent detachment. The originally concentrated damage area becomes blurred due to signal aliasing, making it impossible to accurately delineate the true boundary of the collapse. The root cause of this phenomenon lies in the strong mutual influence between the temporal interval and spatial location of detachment events. A small offset distance at the detachment point in the central area should indicate a high concentration of damage, but when the time interval is extremely short, the vibrations caused by consecutive detachments superimpose, rendering the separation criteria for individual events ineffective. Conversely, if relying solely on increasing the time separation threshold to avoid aliasing, it is easy to incorrectly separate multiple small detachments belonging to the same damage process into unrelated events, leading to either an overestimation or an underestimation of the damage area, and consequently, deviations in the selection of support areas.

[0004] For example, after a blast at a tunnel face, monitoring revealed three separate debris falls within 1.2 seconds within a 30-centimeter radius of the rock mass center. The first fall was off-center, while the second and third followed closely behind, almost overlapping in location. Due to the continuous superposition of vibration signals, the system struggled to determine whether these three events represented three stages of the continuous expansion of the same fracture or three relatively independent failure events. If misjudged as a single event, the collapse area could be severely underestimated; if separated into three independent events, the affected area would be artificially expanded, resulting in wasted support resources or incorrect selection of key protection locations.

[0005] Therefore, how to accurately separate the true boundary of each independent detachment event in the case of continuous detachment in the central area over a short period of time, and reliably delineate the scope of the collapse based on this, so as to determine the priority area for support in a timely manner, has become a key issue for real-time early warning of rock mass blasting disturbance risk. Summary of the Invention

[0006] This invention provides a method for real-time early warning of rock mass blasting disturbance risk, including:

[0007] Displacement data, vibration waveform signals, and surface strain distribution data were collected at various points on the rock mass surface, and the location of the first detachment point and the time of subsequent detachment were recorded.

[0008] Determine the distance between the initial detachment point and the geometric center of the rock mass. Identify the time interval between two adjacent detachments based on the time of each subsequent detachment. Use the ratio of the detachment distance to the time interval as an adjustment coefficient for the event separation threshold.

[0009] When the time interval between two adjacent detachments is less than the preset event separation threshold, vibration signals are collected, overlapping parts in the vibration signals are identified and denoised to obtain the start and end time boundaries of a single detachment event.

[0010] Based on the surface strain distribution data, the geometric shape of the strain concentration zone is determined. Combined with the start and end time boundaries of the single detachment event, the continuous detachment is grouped according to the detachment location using a clustering algorithm. The distance change feature between two adjacent detachment points is extracted. Based on the distance change feature, the spatial boundary of the independent detachment event is determined, and a preliminary outline of the collapse impact range is obtained.

[0011] Based on the preliminary outline of the collapse-affected area, the coordinates of the boundary points are determined, the data-missing regions in the strain concentration zone are identified, the data-missing regions are filled using interpolation, and a complete damage area distribution map is generated.

[0012] The area overlap ratio of adjacent areas in the damage area distribution map is evaluated. When the area overlap ratio exceeds the preset overlap ratio threshold, the event separation threshold is dynamically adjusted according to the adjustment coefficient of the event separation threshold. The time boundaries of each independent event in the continuous shedding in the central area are re-identified in combination with the adjusted event separation threshold.

[0013] Based on the temporal and spatial boundaries of the independent events, and combined with the time interval between two adjacent detachments, high-risk boundary segments within the collapse impact range are determined, and a real-time early warning signal for rock mass blasting disturbance risk is generated.

[0014] Preferably, the acquisition of displacement data, vibration waveform signals, and surface strain distribution data at various points on the rock mass surface, and the recording of the location of the initial detachment and the time of subsequent detachments, includes:

[0015] Based on the geometric contour of the rock mass surface, multiple displacement probes and microseismic pickup units are deployed at fixed intervals along the strike and dip. The coordinates of the installation points of each probe are obtained by total station measurement and written into the coordinate registration table.

[0016] The displacement data, vibration waveform signal, and surface strain distribution data of each point are collected to obtain the original acquisition record with point coordinate markers;

[0017] From the displacement data of the original acquisition record, read the hourly change of displacement amplitude point by point;

[0018] When the change at a certain moment exceeds the preset mutation threshold, the corresponding moment is marked as a shedding trigger moment;

[0019] The coordinates of the point corresponding to the earliest detachment trigger time in chronological order are taken as the location of the first detachment point. The subsequent detachment trigger times are arranged in chronological order to obtain a complete record of the location of the first detachment point and the times of each subsequent detachment.

[0020] Preferably, the step of determining the detachment distance between the initial detachment point and the geometric center of the rock mass, identifying the time interval between two adjacent detachments based on the times of subsequent detachments, and using the ratio of the detachment distance to the time interval as an adjustment coefficient for the event separation threshold includes:

[0021] Read the three-dimensional coordinates of the first detachment point, calculate the Euclidean distance between the three-dimensional coordinates of the first detachment point and the three-dimensional coordinates of the geometric center of the rock mass, and obtain the detachment spacing.

[0022] Read the trigger times of two adjacent detachments in chronological order to obtain the time interval between each pair of adjacent detachments, forming a time interval sequence;

[0023] For each time interval in the time interval sequence, the shedding interval is divided by that time interval to obtain an adjustment coefficient that serves as the event separation threshold.

[0024] Preferably, when the time interval between two adjacent detachments is less than a preset event separation threshold, vibration signals are collected, overlapping portions in the vibration signals are identified and denoised to obtain the start and end time boundaries of a single detachment event, including:

[0025] Read the time interval of each pair of adjacent detachments from the time interval sequence. If the time interval is less than the product of the preset event separation threshold and the adjustment coefficient, it is marked as a candidate pair for stacking.

[0026] The vibration waveform signal within the corresponding time window of the superposition candidate pair is extracted to obtain the amplitude envelope curve;

[0027] The location where the waveform amplitude crosses zero in the vibration waveform signal is detected to determine the overlapping boundary section;

[0028] Based on the start and end times of the superimposed boundary section, the vibration waveform signal is divided into independent single vibration segments;

[0029] The single vibration segment is decomposed and reconstructed using wavelet decomposition to obtain the vibration segment after noise removal;

[0030] The start and end times of the single detachment event are obtained by extracting the start time when the amplitude of the vibration segment after noise filtering exceeds the preset start threshold and the end time when it falls back below the preset start threshold.

[0031] Preferably, the geometric shape of the strain concentration zone is determined based on the surface strain distribution data. Combined with the start and end time boundaries of the single detachment event, a clustering algorithm is used to group consecutive detachments according to their detachment locations. Distance abrupt change features between two adjacent detachment points are extracted. Based on these distance abrupt change features, the spatial boundary of independent detachment events is determined, resulting in a preliminary outline of the collapse-affected area, including:

[0032] Read the surface strain distribution data at each point, and compare the strain value differences between adjacent points along the strike and dip directions.

[0033] When the difference between the strain values ​​of two adjacent points exceeds the preset strain gradient threshold, the pair of points is marked as a high gradient point pair.

[0034] All high gradient point pairs are connected sequentially according to their point coordinates to form strain gradient boundary lines. The area enclosed by the strain gradient boundary lines is taken as the strain concentration zone. The boundary point coordinates of the strain concentration zone are extracted to obtain the geometric shape of the strain concentration zone.

[0035] Based on the start and end time boundaries of the single detachment event, only the coordinates of the detachment points that fall within the strain concentration zone are retained;

[0036] Using the directional and dip components of each detachment point as two-dimensional features, density clustering algorithm is used to group spatially adjacent detachment points into the same detachment group, resulting in several detachment groups and the set of detachment point coordinates contained therein.

[0037] For each of the aforementioned detachment groups, the coordinates of the detachment points within the group are arranged according to the order of the detachment occurrence time, and the Euclidean distance between two adjacent detachment points is calculated to form a successive detachment spacing sequence.

[0038] When the Euclidean distance between a pair of adjacent detachment points exceeds a preset multiple threshold of the mean of the successive detachment interval sequence, the corresponding position is marked as a distance mutation point. The detachment group is then divided into several subgroups based on the distance mutation points, and each subgroup corresponds to the spatial boundary of an independent detachment event.

[0039] For each subgroup, the extreme values ​​of the orientation and the extreme values ​​of the detachment point coordinates within the group are taken to form a rectangular bounding box as the spatial boundary of the independent detachment event;

[0040] The rectangular bounding boxes of all subgroups are sequentially spliced ​​along the outer boundary, and the area enclosed by the spliced ​​outer envelope is used as the preliminary outline of the collapse-affected area.

[0041] Preferably, the boundary point coordinates are determined based on the preliminary outline of the collapse-affected area, the data-missing regions in the strain concentration zone are identified, and the data-missing regions are filled using interpolation to generate a complete damage area distribution map, including:

[0042] Extract the coordinates of all boundary points on the outer envelope of the preliminary contour of the collapse-affected area;

[0043] The area defined by the coordinates of the boundary points is the region, and regular grid nodes are arranged along the direction and dip.

[0044] For each grid node, the nearest probe point within a preset search radius is found from the surface strain distribution data, the strain value of that probe point is read, and assigned to the corresponding grid node;

[0045] Grid nodes without corresponding probe locations within the search radius are marked as missing nodes;

[0046] The missing nodes were filled with strain values ​​using inverse distance weighted interpolation to obtain a complete strain mesh after filling.

[0047] Based on the filled complete strain mesh, the strain value of each mesh node is mapped to a two-dimensional plane. The outer envelope of the preliminary contour is used as the boundary clipping line. Each mesh node within the clipping line is colored according to the strain value to obtain the complete damage area distribution map.

[0048] Preferably, the area overlap ratio of adjacent areas in the damage area distribution map is evaluated. When the area overlap ratio exceeds a preset overlap ratio threshold, the event separation threshold is dynamically adjusted according to the adjustment coefficient of the event separation threshold. The time boundaries of each independent event in the continuous shedding of the central area are then re-identified based on the adjusted event separation threshold, including:

[0049] Read the spatial boundary corresponding to each independent detachment event, and use the rectangular bounding box enclosed by each spatial boundary as the coverage area;

[0050] For two spatially adjacent coverage areas, the ratio of the overlapping area formed by the intersection of the directional and inclined directions to the reference area is calculated to obtain the area overlap ratio.

[0051] When the area overlap ratio exceeds the preset overlap ratio threshold, the adjusted event separation threshold is obtained by multiplying the preset event separation threshold by the adjustment coefficient associated with the detachment event.

[0052] Based on the adjusted event separation threshold, the overlapping boundary segment identification is re-executed for the time windows corresponding to adjacent detachment events in the central region where the area overlap ratio exceeds the threshold, resulting in an updated time boundary.

[0053] Preferably, based on the temporal and spatial boundaries of the independent events, and combined with the time interval between two adjacent detachments, a high-risk boundary segment within the collapse impact range is determined, and a real-time early warning signal for rock mass blasting disturbance risk is generated, including:

[0054] Read the start and end times of each independent event, as well as the coordinates of the corresponding bounding box;

[0055] For two independent events that are spatially adjacent or overlapping, read the time interval between them;

[0056] When the time interval is less than the adjusted event separation threshold, the adjacent or overlapping boundary segments between the rectangular bounding boxes of the two independent events are marked as high-risk boundary segments.

[0057] Based on the distribution of all high-risk boundary segments, a buffer distance is extended outward from each high-risk boundary segment as the center, and grids are divided within the extended area.

[0058] Grid cells falling within the buffer zone of at least one high-risk boundary segment are marked as support candidate cells and merged into connected regions to delineate priority support areas.

[0059] The percentage of grid cells in the priority support area whose strain values ​​exceed a preset strain risk threshold is used as a risk level indicator.

[0060] When the risk level index exceeds the preset value, a real-time early warning signal for rock mass blasting disturbance risk is output for the corresponding priority support area.

[0061] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0062] This invention discloses a real-time early warning method for rock mass blasting disturbance risk. By deploying multiple displacement probes and microseismic pickup units on the rock mass surface, displacement, vibration waveforms, and surface strain distribution data are collected in real time, accurately recording the location of the first spalling point and the time of each subsequent spalling. Addressing the core problems of difficulty in accurately separating continuous spalling events under rock mass blasting disturbance, difficulty in defining the dynamic evolution of the collapse range, and difficulty in real-time delineation of high-risk areas, a dynamic early warning technology route integrating spatiotemporal features is proposed. First, the event separation threshold is adaptively adjusted based on the ratio of the initial spalling interval to the time interval of adjacent spalling events. When the time interval is less than the threshold, vibration signals are collected and denoised to accurately extract the start and end boundaries of a single spalling event. Then, by combining the morphology of the surface strain concentration zone and the spalling location clustering, distance abrupt change features are extracted to determine the spatial boundary, forming a preliminary outline of the collapse impact range. Subsequently, a complete damage distribution map is generated by interpolating and filling the strain-deficient areas. The event separation threshold is dynamically optimized based on the area overlap ratio of adjacent areas, and the time boundaries of discrete events are redefined. Finally, high-risk boundary segments are identified based on the spatiotemporal boundaries and time intervals of independent events. Grid division is used to delineate priority support areas, and a blasting disturbance risk early warning signal is generated in real time. This invention enables precise spatiotemporal separation of continuous detachment events and dynamic tracking of the collapse range, effectively improving the real-time early warning accuracy and targeted support for collapse disasters under rock mass blasting disturbance. Attached Figure Description

[0063] Figure 1 This is a flowchart of a real-time early warning method for rock mass blasting disturbance risk according to the present invention;

[0064] Figure 2 This is the strain curve corresponding to 25℃ in an embodiment of the present invention;

[0065] Figure 3 This is the strain curve corresponding to 250℃ in the embodiment of the present invention;

[0066] Figure 4 This is the strain curve corresponding to 500℃ in the embodiment of the present invention;

[0067] Figure 5 This is the strain curve corresponding to 750℃ in the embodiment of the present invention;

[0068] Figure 6 This is the strain curve corresponding to 1000℃ in the embodiment of the present invention;

[0069] Figure 7 This is failure mode 1 of rock mass specimens at different temperatures after loading in this embodiment of the invention;

[0070] Figure 8 This represents failure mode 2 of rock mass specimens subjected to loading at different temperatures in this embodiment of the invention. Detailed Implementation

[0071] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0072] like Figure 1 This embodiment of a real-time early warning method for rock mass blasting disturbance risk may specifically include:

[0073] Step S101: Deploy multi-point displacement probes and micro-vibration pickup units on the rock mass surface to collect displacement data, vibration waveform signals and surface strain distribution data, and record the location of the first detachment point and the time of each subsequent detachment.

[0074] Based on the geometric contour of the rock mass surface, multiple displacement probes and microseismic pickup units are deployed at fixed intervals along the strike and dip. The coordinates of each probe's installation point are obtained by total station measurements and written into a coordinate registration table. The sampling frequency of the microseismic pickup units is preset during the deployment phase. Displacement data, vibration waveform signals, and surface strain distribution data are collected at each point to obtain raw acquisition records with point coordinate markings. From the displacement data in the raw acquisition records, the hourly change in displacement amplitude is read point by point. If the change at a certain moment exceeds a preset abrupt change threshold, that moment is marked as a detachment trigger moment. The point coordinates corresponding to the earliest detachment trigger moment in chronological order are taken as the initial detachment point location. Subsequent detachment trigger moments are arranged in chronological order to obtain a complete record of the initial detachment point location and the times of subsequent detachments.

[0075] In one embodiment, multi-point displacement gauges and micro-vibration pickup units are installed point by point along the strike and dip directions of the rock mass surface at preset intervals, wherein the intervals are determined based on the size of the exposed rock mass surface and the density of structural surface development.

[0076] For example, at the tunnel face with densely developed joints, the spacing between adjacent probes should be smaller to increase the coverage density, while the spacing should be appropriately increased on the slope rock surface with better integrity.

[0077] Specifically, after each probe is installed, a total station is used to perform three-dimensional coordinate measurements at each installation point. The measured coordinate values ​​include the lateral position along the strike, the longitudinal position along the dip, and the normal offset perpendicular to the rock surface. These three coordinates are written into a coordinate registration table, establishing a one-to-one mapping relationship with the corresponding probe number. During the deployment phase, the sampling frequency of the microseismic pickup unit is uniformly set. The selection of this frequency refers to the main frequency range of the vibration signal generated by rock blasting disturbance, ensuring that the pickup unit has a sufficient number of sampling points within the main signal frequency band. Based on this, displacement data, vibration waveform signals, and surface strain distribution data are synchronously collected at each point. The surface strain distribution data is acquired through strain gauges or fiber optic grating sensors attached to the rock surface. The strain gauge converts the minute deformations of the rock surface into resistance change signals, which are amplified by a bridge circuit and output as strain values. The formula for calculating the strain value is ε=ΔR / (K×R0), where ε is the strain value (dimensionless), ΔR is the resistance change (unit: Ω), R0 is the initial resistance value (unit: Ω), and K is the strain gauge sensitivity coefficient. All three types of data carry corresponding location coordinates and collection time labels, which are then aggregated to form the original collection record.

[0078] It should be noted that the displacement data records the real-time displacement of each point in the normal direction in the form of a time series. When debris falls off the rock surface, the displacement at the point of falloff increases sharply in a very short time, which means that the hourly change in displacement amplitude is much greater than the background fluctuation level before the falloff.

[0079] In one embodiment, for the displacement data in the original acquisition record, the displacement amplitude difference between two adjacent sampling times is read point by point, and this difference is used as the hourly change at that time. If the hourly change at a certain time exceeds a preset abrupt change threshold, it is determined that the time corresponds to a debris shedding event, and the time is marked as the shedding trigger time. The value of the abrupt change threshold is determined based on the background displacement fluctuation amplitude collected during the deployment phase.

[0080] For example, a static data acquisition is performed on the rock mass surface before blasting operations, and the mean μ and standard deviation σ of the hourly displacement changes at each point are statistically analyzed, along with the abrupt change threshold T. th The calculation formula is: T th =μ + n × σ, where n is a multiplier coefficient, typically ranging from 2 to 3, to distinguish between normal micro-motion and abrupt changes caused by shedding. When the hourly change at a certain moment exceeds T... th When a debris detachment event occurs, it is determined that the given moment corresponds to a debris detachment event. Among all the marked detachment trigger moments, the earliest one in chronological order is selected, and the corresponding coordinates are read to determine the location of the initial detachment point. The remaining detachment trigger moments are arranged chronologically, and together with their corresponding coordinates, they form a complete record of the times of subsequent detachments.

[0081] Step S102: Determine the detachment distance from the initial detachment point to the geometric center of the rock mass based on the location of the initial detachment point; identify the time interval between two adjacent detachments based on the time of each subsequent detachment; and use the ratio of the detachment distance to the time interval as an adjustment coefficient for the event separation threshold.

[0082] The three-dimensional coordinates of the initial detachment point are read from the coordinate registration table of the initial detachment point location. The three-dimensional coordinates of the rock mass aggregation center are determined in the following way:

[0083] For a regular rock mass exposed surface, the arithmetic mean of the coordinates of all probe points in the strike, dip and normal directions is taken as the three-dimensional coordinates of the geometric center of the rock mass.

[0084] For irregular or non-uniformly distributed rock mass exposed surfaces, the coordinates of the boundary contour points of the exposed rock mass are obtained by three-dimensional laser scanning or total station measurement. Then, the boundary contour points are projected onto the strike-dip plane, and the geometric centroid coordinates of the region enclosed by the boundary contour are calculated using the polygon centroid algorithm. Finally, the average normal position of the exposed rock mass surface in the normal direction is taken as the third-dimensional coordinate, thereby obtaining the three-dimensional coordinates of the geometric center of the rock mass.

[0085] Based on the coordinates of the probes deployed on the exposed rock mass surface, the arithmetic mean of all coordinates in the strike, dip, and normal directions is taken as the three-dimensional coordinates of the rock mass's geometric center. The Euclidean distance between the three-dimensional coordinates of the initial detachment point and the three-dimensional coordinates of the geometric center is calculated, yielding the detachment distance from the initial detachment point to the rock mass's geometric center. From the complete record of subsequent detachment occurrences, the trigger times of two adjacent detachments are read sequentially in chronological order. The time interval between each pair of adjacent detachments is obtained by subtracting the trigger time of the previous detachment from the trigger time of the next detachment. All these time intervals are arranged sequentially according to their corresponding detachment order to form a time interval sequence. For each time interval in the time interval sequence, the value of the detachment distance is read. The detachment distance is divided by the time interval to obtain the corresponding ratio. If a certain time interval Δt is less than the preset lower limit threshold Δt_min, a saturation function is used to calculate the corresponding adjustment coefficient to avoid an abnormally large ratio due to division operations.

[0086] The formula for the saturation function is: α = α_max × [1 - exp(-k × d / Δt_min)] where: - α is the adjustment coefficient; - α_max is the upper cutoff value, ranging from 5 to 20, calibrated based on historical shedding event data, with a recommended value of 10; - k is the attenuation coefficient, ranging from 0.5 to 2.0, used to control the steepness of the saturation curve, with a recommended value of 1.0; - d is the shedding interval (unit: m); - Δt_min is the lower limit threshold of the interval (unit: s), determined based on the sampling frequency of the microseismic pickup unit, and calculated using the formula Δt_min = N_min / f_s, where N_min is the minimum resolvable sampling point number (usually taken as 10-20), and f_s is the sampling frequency (Hz). The characteristics of this saturation function are: when the time interval approaches zero, the adjustment coefficient smoothly approaches the upper cutoff value α_max instead of infinity; when the time interval is large, the function value is approximately equal to the linear ratio d / Δt, thus preserving the dynamic distinguishing ability.

[0087] In one implementation, the three-dimensional coordinates of the initial detachment point are directly read from the coordinate registration table, which is determined and stored point by point by the total station during the deployment phase. Each record contains the probe number and its corresponding orientation coordinates, dip coordinates, and normal coordinates.

[0088] Specifically, the three-dimensional coordinates of the geometric center of the rock mass are obtained by taking the coordinates of all the probes. The arithmetic mean of the coordinates of all the probes in the strike, dip, and normal directions is then taken, and the three average values ​​are the coordinate components of the geometric center in the three directions.

[0089] For example, in a tunnel face scenario, if several columns of probes are arranged laterally along the face and several rows of probes are arranged longitudinally, then the average value of all coordinates of the points in the lateral direction corresponds to the orientation coordinate of the geometric center, and the average value of all coordinates of the points in the longitudinal direction corresponds to the dip coordinate of the geometric center. The same applies to the normal direction.

[0090] It should be noted that the detachment spacing is measured using Euclidean distance. This involves calculating the coordinate differences between the first detachment point and the geometric center in the strike, dip, and normal directions, summing the squares of these three differences, and taking the square root of the sum. The resulting value is the detachment spacing. This detachment spacing reflects the degree to which the initial detachment point deviates from the central region of the rock mass.

[0091] In one embodiment, the extraction process of the time interval sequence is as follows: all detachment trigger times are arranged chronologically, and the difference between two adjacent times is used to obtain a time interval. This process is repeated for all adjacent time pairs to form a time interval sequence that corresponds one-to-one with the detachment sequence position. Further, the adjustment coefficient is calculated by dividing the detachment interval by the time interval.

[0092] Understandably, when the detachment spacing is large and the time interval is long, the ratio is within a moderate range, indicating that the detachment events have distinguishable intervals in both space and time. When the detachment spacing remains constant but the time interval shortens, the ratio increases, meaning that continuous detachment is more concentrated in time. At this point, the possibility of vibration signal overlap increases, and the event separation threshold should be tightened accordingly. Therefore, this ratio is directly used as an adjustment coefficient for the event separation threshold to reflect the coupling relationship between spatial offset and temporal compactness.

[0093] Preferably, when the value of a certain time interval is extremely small, the ratio obtained by division will be abnormally large, deviating from the normal adjustment range. To address this situation, a lower limit threshold for the interval is preset. If the time interval is less than this lower limit threshold, the division operation is no longer performed. Instead, a saturation function is used to calculate the ratio, making it smoothly approach the preset upper limit. This constrains the value of the adjustment coefficient within a reasonable range while retaining dynamic differentiation capability.

[0094] In one possible implementation, the lower limit threshold of the interval is determined based on the minimum resolvable time window corresponding to the sampling frequency of the micro-vibration pickup unit. That is, when the time interval between two detachments is less than the shortest duration for the pickup unit to distinguish two independent vibration waveforms, the time interval no longer has effective separation significance, and the corresponding adjustment coefficient is replaced by the upper limit cutoff value.

[0095] Step S103: When the time interval between two adjacent detachments is less than the preset event separation threshold, a vibration signal is collected, the overlapping part in the vibration signal is identified and denoised, and the start and end time boundaries of a single detachment event are obtained.

[0096] The time interval of each pair of adjacent detachments is read sequentially from the time interval sequence. If the time interval is less than the product of a preset event separation threshold and the adjustment coefficient, the pair of adjacent detachments is marked as a candidate pair for overlay. The vibration waveform signal within the time window corresponding to the candidate pair for overlay is extracted from the original acquisition record. The absolute value of the amplitude of the vibration waveform signal is taken point by point and connected in chronological order to obtain the amplitude envelope curve within the time window. In the time axis direction of the original vibration waveform signal, the position where the waveform amplitude crosses zero is detected point by point and the time corresponding to each zero crossing point is recorded. The interval between two adjacent zero crossing points is taken as a single vibration cycle. The start and end times of each vibration cycle are mapped to the corresponding time interval on the amplitude envelope curve. The maximum amplitude value is found point by point in this interval as the peak amplitude. If the amplitude envelope between two adjacent peaks falls back to below the preset noise threshold, the falling section is determined as the overlay boundary section.

[0097] Specifically, the candidate peak points are obtained by scanning the amplitude envelope curve point by point along the time axis. A sampling point is marked as a candidate peak point when its amplitude simultaneously meets the following two conditions: Specifically, the amplitude of this point is greater than the amplitudes of its N_neighbor neighboring sampling points, where N_neighbor = f_s × T_window, f_s is the sampling frequency, and T_window is the local window duration (valued at 0.005-0.01 seconds); the amplitude of this point exceeds S times the mean of the amplitude envelope curve, where S is the significance coefficient (valued at 1.5-3.0, recommended value 2.0).

[0098] Specifically, the fallback zone is defined as a segment between two adjacent candidate peaks where the amplitude of more than M consecutive sampling points is lower than a preset noise threshold. M = f_s × T_min_gap, where T_min_gap is the minimum interval duration (0.01-0.05 seconds).

[0099] In this embodiment, the starting point of the fallback section is defined as the sampling moment when the amplitude first falls below the noise threshold after dropping from the peak; the ending point of the fallback section is defined as the sampling moment when the amplitude first rises above the noise threshold after falling below it. When the duration of the fallback section exceeds T_min_gap, the fallback section is determined to be an overlay boundary section, indicating that there is a separable gap between the vibration signals of two adjacent detachment events at this point.

[0100] The preset noise threshold is determined based on the background amplitude level of the vibration waveform at each point during the static acquisition stage before blasting. The calculation formula is: A_noise = μ_bg + n_noise × σ_bg, where μ_bg and σ_bg are the mean and standard deviation of the amplitude envelope during the static stage, respectively, and n_noise is a multiplier (value 2-3).

[0101] Based on the start and end times of the superimposed boundary section, the vibration waveform signal of the superimposed candidate pair is divided into independent single vibration segments. For each single vibration segment, a preset wavelet basis function is selected to perform multi-level decomposition. The components with amplitudes lower than a preset coefficient threshold in each layer of wavelet coefficients are set to zero and the signal is reconstructed to obtain the vibration segment after noise is filtered out. The moment when the amplitude of the vibration segment first exceeds the preset oscillation threshold is extracted as the start time, and the moment when the amplitude finally falls back to below the oscillation threshold is extracted as the end time, thus obtaining the start and end time boundaries of a single detachment event.

[0102] In one implementation, the determination of the superposition candidate pair is performed one by one for each pair of adjacent detachments. Specifically, the time interval between the pair of adjacent detachments is compared with an adjusted event separation threshold, where the adjusted event separation threshold is equal to a preset event separation threshold multiplied by a corresponding adjustment coefficient. When the time interval is less than this product value, it indicates that the two detachments are too close in time, and the corresponding vibration signals may superimpose. In this case, the pair of adjacent detachments is marked as a superposition candidate pair.

[0103] Specifically, for each candidate stacking pair, vibration waveform signals within the corresponding time window are extracted from the original acquisition records according to the order of the pair's detachment trigger times. The absolute value of the amplitude is taken for each sampling point of the extracted vibration waveform signal, and then connected sequentially along the time axis to form an amplitude envelope curve. This curve reflects the profile of vibration energy change over time, where peaks correspond to periods of concentrated vibration energy, and troughs correspond to periods of vibration decay or intermittency.

[0104] It should be noted that the zero-crossing detection is performed on the captured original vibration waveform signal, not on the amplitude envelope curve. The vibration waveform signal alternates periodically between positive and negative values. A zero-crossing point is generated whenever the waveform amplitude changes from a positive value to a negative value or vice versa. The vibration waveform signal is scanned point by point along the time axis. When the amplitude signs of two adjacent sampling points are opposite, the time corresponding to that position is recorded as the zero-crossing point. The interval between two adjacent zero-crossing points corresponds to one half-vibration cycle, thus dividing the entire vibration waveform signal into segments according to the vibration cycle. Based on this, the amplitude envelope curve is returned to, and the envelope peak value within each vibration cycle interval is read. If there is a segment between two adjacent envelope peak values ​​where the envelope amplitude is continuously lower than a preset noise threshold, then the vibration energy in that segment has attenuated to the background noise level and can be identified as an overlay boundary segment.

[0105] In one embodiment, the preset noise threshold is determined based on the background amplitude level of the vibration waveform at each point during the static acquisition phase before blasting. For example, a multiple of the mean amplitude envelope value during the static phase is taken as the upper bound of the noise threshold. Further, based on the start and end times of the stacking boundary section, the vibration waveform signal of the stacking candidate pair is divided into several independent single vibration segments along the boundary position, with each segment corresponding to the vibration response generated by a detachment event.

[0106] For example, the specific operation process of wavelet denoising is as follows: For each single vibration segment, a db4 wavelet basis function is selected to decompose it layer by layer. The number of decomposition layers N is based on the sampling frequency f of the vibration segment. s The dominant frequency f0 of the rock mass detachment vibration is determined, and the calculation formula is N=log2(f s / f0)⌋。 After decomposition, each wavelet coefficient in the high-frequency detail components of each layer is compared with the coefficient threshold λ. The coefficient threshold is determined using a general thresholding method, and the calculation formula is as follows: Where σ is the estimated standard deviation of noise and M is the number of signal sampling points. Wavelet coefficients with amplitudes lower than λ are set to zero, while wavelet coefficients with amplitudes higher than λ retain their original values. Subsequently, the processed wavelet coefficients of each layer are reconstructed layer by layer according to the reverse process of decomposition to synthesize the vibration segment after high-frequency noise has been filtered out.

[0107] Understandably, the vibration segment after wavelet denoising still retains the dominant frequency vibration component of the detachment event itself, while random high-frequency interference has been suppressed. On this noise-filtered vibration segment, scanning forward along the time axis, the sampling time when the amplitude first exceeds the preset oscillation threshold is taken as the start time of this detachment event; scanning backward along the time axis, the sampling time when the amplitude last falls back below the oscillation threshold is taken as the end time, thereby determining the start and end time boundaries of a single detachment event.

[0108] Step S104: Determine the geometric shape of the strain concentration zone based on the surface strain distribution data, combine the start and end time boundaries of a single detachment event, group the continuous detachments according to their detachment locations using a clustering algorithm, extract the distance abrupt change features between two adjacent detachment points, determine the spatial boundary of an independent detachment event based on the distance abrupt change features, and obtain the preliminary outline of the collapse-affected area.

[0109] Surface strain distribution data for each point is read from the original acquisition record. The strain value difference between adjacent points is compared along the strike and dip directions. If the difference between the strain values ​​of two adjacent points exceeds a preset strain gradient threshold, the pair of points is marked as a high gradient point pair. All high gradient point pairs are connected sequentially according to their point coordinates to form a strain gradient boundary line. The area enclosed by the strain gradient boundary line is taken as the strain concentration zone. The boundary point coordinates of the strain concentration zone are extracted to obtain the geometric shape of the strain concentration zone. According to the start and end time boundaries of the single detachment event, the coordinates of the detachment point corresponding to each independent detachment event are read from the coordinate registration table. Using the geometric shape of the strain concentration zone as a spatial constraint, only the coordinates of the detachment points falling within the strain concentration zone are retained. Using the strike and dip components of each detachment point as two-dimensional features, a density clustering algorithm is used. The preset neighborhood radius and minimum number of neighborhood points are used as clustering criteria to group spatially adjacent detachment points into the same detachment group, resulting in several detachment groups and their contained detachment point coordinate sets. For each detachment group, the coordinates of the detachment points within the group are arranged chronologically according to the time of detachment. The Euclidean distance between adjacent detachment points is calculated sequentially to form a successive detachment spacing sequence. If the Euclidean distance between a pair of adjacent detachment points exceeds a preset multiple threshold of the sequence mean, the location is marked as a distance abrupt change point. The detachment group is then divided into several subgroups based on these distance abrupt change points, with each subgroup corresponding to the spatial boundary of an independent detachment event. For each subgroup, the extreme values ​​of the detachment point coordinates within the group are taken along the strike and dip directions. The strike minimum, strike maximum, dip minimum, and dip maximum are used to form a rectangular bounding box as the spatial boundary of the independent detachment event. The rectangular bounding boxes of all subgroups are sequentially spliced ​​along their outer boundaries, and the area enclosed by the spliced ​​outer envelope is used as the preliminary outline of the collapse impact range.

[0110] In one implementation, the identification of the strain concentration zone is performed based on surface strain distribution data at each point. The strain values ​​of two adjacent points are taken along both the strike and dip directions, and the absolute value of the difference is the strain gradient between the pair of points. When the strain gradient exceeds a preset strain gradient threshold, it indicates a significant abrupt change in the deformation of the rock mass surface within a short distance, and the pair of points is marked as a high-gradient point pair.

[0111] Specifically, the strain gradient threshold is determined based on the statistical distribution of strain differences between adjacent points on the rock mass surface during the static acquisition stage before blasting. For example, a threshold is set as a multiple of the average strain gradient during the static stage. All marked high-gradient point pairs are connected sequentially on the strike and dip planes according to their respective point coordinates, forming one or more strain gradient boundary lines. When the boundary lines enclose a closed region, the strain value inside the region is generally higher than that outside, thus forming a strain concentration zone. The coordinates of the boundary points of the strain concentration zone are extracted, and the extreme value ranges of the boundary are read along the strike and dip directions, respectively, thereby obtaining the geometric shape of the strain concentration zone.

[0112] Specifically, such as Figures 2 to 6 As shown, in the initial loading stage, almost all the microcracks (such as fine cracks and micropores) inside the rock mass samples underwent a gradual evolution process, and the strain curves all showed a concave shape, indicating that this stage was the compaction stage. With the increase of heat treatment temperature, the crack evolution process in the samples became more obvious. The pre-fabricated fractured rock masses after high-temperature heat treatment at 500℃ and 750℃ exhibited significant crack propagation, with some cracks expanding during loading and exhibiting rockburst phenomena (i.e., micro-crack propagation and acoustic emission). The strain of both groups of samples showed a negative value region in this stage, and the concave shape of the curves was more pronounced than that of the samples at room temperature and low temperature, indicating a significant energy release during crack propagation. As the load continued to increase, the strain curves of the samples gradually approached linearity, entering the elastic deformation stage. In this stage, the elastic modulus of the rock was relatively stable, exhibiting typical elastic behavior.

[0113] It should be noted that the geometry of the strain concentration zone plays a role as a spatial constraint in the subsequent screening of detachment points. Based on the start and end times of each independent detachment event, the corresponding detachment point coordinates are read from the coordinate register. Each detachment point is then checked to determine whether it falls within the boundary of the strain concentration zone. Only detachment point coordinates within this range are retained, while discrete detachment points outside the strain concentration zone are excluded. Furthermore, the screened detachment point coordinates are grouped using a density clustering algorithm. The core criteria of the density clustering algorithm include two preset parameters: neighborhood radius and minimum number of neighborhood points. The neighborhood radius defines the radius of the search circle drawn on a two-dimensional plane formed by the direction and dip, with a given detachment point coordinate as its center. The minimum number of neighborhood points defines the minimum number of other detachment points that should be included within this search circle. The specific operation process is as follows: Using the directional and dip components of each detachment point as two-dimensional coordinates, the number of other detachment points within its neighborhood radius is counted for each point. If this number reaches or exceeds the minimum number of neighborhood points, the detachment point is marked as a core point. Starting from any core point, all detachment points within its neighborhood are included in the same detachment group. The newly included core points are then expanded outwards until no further expansion is possible, thus obtaining a complete detachment group. The above expansion process is repeated for core points that have not yet been grouped, ultimately resulting in several detachment groups and the set of coordinates of the detachment points they contain.

[0114] In one embodiment, the value of the neighborhood radius is determined by referring to a multiple of the distance between adjacent probes, and the minimum number of neighborhood points is determined based on the ratio of the total number of detachment points within the strain concentration zone to the expected number of groups.

[0115] For example, in a tunnel face blasting disturbance scenario, if there are two relatively concentrated but clearly spaced detachment areas within the strain concentration zone, the density clustering algorithm will automatically assign the two areas to different detachment groups based on the neighborhood radius. The sparse detachment points between the two areas will not be assigned to any group due to insufficient points in their neighborhood.

[0116] It is understandable that, such as Figure 7 and Figure 8As shown, the detachment points within the same detachment group are spatially adjacent, but may still contain multiple independent detachment events in the temporal dimension. For each detachment group, the coordinates of all detachment points within the group are arranged chronologically according to the time of detachment occurrence, and the Euclidean distance between any two temporally adjacent detachment points is calculated sequentially, forming a successive detachment spacing sequence. This sequence is scanned one by one, and the arithmetic mean of all spacing values ​​in the sequence is calculated simultaneously. If the Euclidean distance between a pair of adjacent detachment points exceeds a preset multiple threshold of this mean, it indicates that a sudden spatial jump occurred around that moment, and this location is marked as a distance abrupt change point. Using these distance abrupt change points as sub-division positions within the group, the original detachment group is divided into several sub-groups. The detachment points within each sub-group remain continuous in both time and space, corresponding to the spatial boundary of an independent detachment event.

[0117] In one possible implementation, the preset multiple threshold is determined based on the dispersion of successive detachment intervals within a detachment group. For example, when the standard deviation of the detachment interval sequence is large, the multiple threshold is appropriately increased to avoid excessive splitting. For each sub-group, the coordinates of all detachment points it contains are read, and the minimum and maximum values ​​of the coordinates are taken along the strike direction and along the dip direction. The rectangular bounding box formed by the four extreme values ​​is used as the spatial boundary of this independent detachment event. For the rectangular bounding boxes of all sub-groups, they are sequentially spliced ​​along the outer boundaries of each box. During the splicing process, the overlapping or tangent edges of adjacent rectangular bounding boxes are used as common boundaries for merging. The complete area enclosed by the outer envelope formed after splicing is used as the preliminary outline of the collapse-affected area.

[0118] Step S105: Determine the coordinates of the boundary points based on the preliminary outline of the collapse-affected area, identify the areas with missing data in the strain concentration zone, fill the missing areas using interpolation, and generate a complete distribution map of the damaged area.

[0119] The coordinates of all boundary points on the outer envelope of the preliminary contour of the collapse-affected area are extracted. Using the area defined by these boundary point coordinates as the region, regular grid nodes are arranged along the strike and dip directions at a preset grid spacing. For each grid node, the nearest probe point within a preset search radius is found from the surface strain distribution data. The strain value of this probe point is read and assigned to the corresponding grid node. If no probe point corresponds to a certain grid node within the search radius, the grid node is marked as a missing node, resulting in a strain grid with missing node markings. The coordinates of each missing node are read one by one from the strain grid with missing node markings. Within the search radius, the coordinates and strain values ​​of the adjacent grid nodes around the missing node are obtained. An inverse distance weighted interpolation method is used, with the reciprocal of the Euclidean distance between the missing node and each adjacent assigned node as the weight, to perform a weighted summation of the strain values ​​of each adjacent assigned node, obtaining the interpolated strain value of the missing node. This interpolated strain value is written into the corresponding grid node, resulting in a filled complete strain grid. Based on the filled complete strain mesh, the strain value of each mesh node is mapped onto the two-dimensional plane formed by the direction and dip according to its coordinate position. The outer envelope of the preliminary contour is used as the boundary clipping line. The area outside the clipping line is not displayed. Each mesh node inside the clipping line is colored one by one according to the mapping rule that the strain value corresponds to the color level from light to dark, so as to obtain a complete distribution map of the damaged area.

[0120] In one embodiment, the preliminary outline of the collapse-affected area is defined by the outer envelope obtained in the previous processing stage. The coordinates of each turning point are read sequentially along the outer envelope, which are the boundary point coordinates. The area enclosed by the boundary point coordinates is used as the range, and regular grid nodes are arranged row by row and column by column along the direction of strike and dip at a preset grid spacing, so that the grid nodes uniformly cover the internal area of ​​the preliminary outline.

[0121] Specifically, for each grid node, the nearest probe point to that node's coordinates is searched within a preset search radius. If at least one probe point is found, its strain value is read and assigned to the corresponding grid node. If no probe point is found within the search radius, the grid node is marked as a missing node. The search radius is typically determined by a multiple of the distance between adjacent probes. After this node-by-node traversal, a strain grid with missing node markings is obtained.

[0122] It should be noted that the missing nodes are caused by the absence of probes in certain areas within the initial outline, or by probe damage after blasting disturbances, resulting in the inability to collect data. If these missing nodes are not filled, the corresponding positions in the damage area distribution map will appear blank, failing to reflect the actual strain state at that location. Furthermore, inverse distance weighted interpolation is performed on each missing node in the strain grid with missing markers. The calculation formula for the inverse distance weighted interpolation method is: ε0=Σ(w i ×ε i ) / Σw i Where ε0 is the interpolated strain value of the missing node, ε i w is the strain value of the i-th adjacent already assigned node. i This represents the corresponding weight value. The formula for calculating the weight value is: w i =1 / d i p , where d i Let be the Euclidean distance (in meters) between the missing node and its i-th adjacent assigned node, and p be the power parameter, typically set to 2. Assigned nodes closer to the missing node have greater weight and contribute more to the interpolation result. After writing the interpolated strain values ​​into each missing node, the filled complete strain mesh is obtained.

[0123] In one embodiment, if the number of assigned nodes within the search radius of a missing node is small, the search radius can be appropriately expanded until a preset minimum number of assigned nodes are obtained before performing interpolation. Based on the filled complete strain mesh, the strain values ​​of each mesh node are mapped onto a two-dimensional plane formed by the orientation and dip according to their coordinate positions. Each node is colored one by one according to the rule that the strain values ​​correspond to light to dark color levels from low to high. The outer envelope of the preliminary contour is used as the boundary clipping line, and the area outside the clipping line is not displayed, thereby obtaining a complete distribution map of the damaged area.

[0124] Step S106: Evaluate the area overlap ratio of adjacent areas in the damage area distribution map. When the area overlap ratio exceeds the preset overlap ratio threshold, dynamically adjust the event separation threshold according to the adjustment coefficient of the event separation threshold, and re-identify the time boundaries of each independent event in the continuous shedding of the central area based on the adjusted event separation threshold.

[0125] The spatial boundary corresponding to each independent detachment event is read from the damage area distribution map. The rectangular bounding box enclosed by each spatial boundary is taken as the coverage area of ​​the event. For two spatially adjacent coverage areas, the product of the length of the intersection of the two rectangular bounding box intervals in the directional direction and the width of the intersection in the directional direction is taken as the overlapping area. The overlapping area is divided by the reference area to obtain the area overlap ratio of the pair of adjacent areas. If the area overlap ratio exceeds a preset overlap ratio threshold, the adjustment coefficient associated with the detachment event corresponding to each pair of adjacent areas is read. The preset event separation threshold is multiplied by the adjustment coefficient associated with the detachment event to obtain the adjusted event separation threshold. Based on the adjusted event separation threshold, the time windows corresponding to adjacent detachment events in the central area with an area overlap ratio exceeding the threshold are re-selected from the time interval sequence. The original threshold is replaced with the adjusted event separation threshold, and the overlapping boundary segment identification is re-performed on the vibration waveform signal within the time window to obtain the updated overlapping boundary segment. Based on the start and end times of the updated overlapping boundary section, the corresponding vibration waveform signal is re-divided into independent single vibration segments. After filtering out noise using wavelet denoising method for each single vibration segment, the start and end times are extracted to obtain the time boundaries of each independent event in the re-identified continuous shedding of the central region.

[0126] In one implementation, the assessment of the area overlap ratio is performed pairwise for spatially adjacent independent detachment events on the damage area distribution map. The coverage area of ​​each independent detachment event is defined by a rectangular bounding box enclosed by its spatial boundary, and whether there is overlap between two adjacent coverage areas depends on whether their coordinate intervals in the directional and dip directions intersect.

[0127] Specifically, for any pair of adjacent rectangular bounding boxes A and B, let the traverse interval of A be [x...]. A1 ,x A2 The tendency interval is [y]. A1 ,y A2 The direction interval of B is [x]. B1 ,x B2 The tendency interval is [y]. B1 ,y B2 The overlap length L of the direction of travel. x =max(0,min(x A2 ,x B2 )-max(x A1 ,x B1 The overlap width L in the directional direction y =max(0,min(y A2 ,y B2 )-max(y A1 ,y B1Overlapping area S overlap =L x ×L y The formula for calculating the area overlap ratio R is: R = S overlap / min(S A ,S B ), where S A and S B Let A and B be the areas of the bounding rectangles, respectively. When determining the baseline area, the smaller of the areas of the two bounding rectangles is used. The overlapping area is divided by the baseline area; the quotient is the area overlap ratio of the pair of adjacent regions. The smaller area is chosen as the baseline because when a smaller coverage area is largely contained within a larger adjacent area, using the smaller area as the baseline more accurately reflects the actual degree of overlap between the two.

[0128] It should be noted that the area overlap ratio ranges from zero to one. The closer the value is to one, the higher the spatial overlap between the two adjacent coverage areas. In this case, the original event separation threshold is insufficient to clearly distinguish the two in the time dimension, and dynamic adjustment of the threshold needs to be triggered. Furthermore, when the area overlap ratio of a pair of adjacent areas exceeds the preset overlap ratio threshold, the adjustment coefficient associated with the detachment events corresponding to each pair of adjacent areas is read. The adjustment coefficient has been determined in the preprocessing stage by the ratio of detachment distance to time interval, reflecting the coupling relationship between spatial offset and temporal compactness. The larger of the two adjustment coefficients is selected, which means that the larger the adjustment coefficient, the higher the ratio of spatial offset to time interval of the corresponding detachment event, which means that the event has a more stringent requirement for temporal separation accuracy. The preset event separation threshold is multiplied by the larger adjustment coefficient, and the product is the adjusted event separation threshold. This adjusted threshold tightens the temporal criteria for event separation, so that adjacent detachment events that were originally merged and identified due to an overly wide threshold can be distinguished during re-identification.

[0129] In one embodiment, the preset overlap ratio threshold is preset based on the ratio between the spacing of rock mass structural surfaces and the average coverage area of ​​the detachment event. For example, in a tunnel face with densely developed joints, since the coverage areas of adjacent detachment events are likely to overlap, the overlap ratio threshold is set to a relatively high value to avoid triggering threshold adjustment too frequently.

[0130] For example, in a monitoring scenario following roof blasting in a mine, if the bounding boxes of two adjacent independent detachment events significantly overlap in both the strike and dip directions, and the area overlap exceeds a preset threshold, it indicates that these two events are highly spatially overlapping. The original event separation threshold may cause their vibration signals to be incorrectly merged in the time dimension. In this case, by multiplying by a larger adjustment coefficient to tighten the threshold, more refined segmentation of the vibration signals within that time window can be achieved during subsequent re-identification.

[0131] It is understood that the re-identification is performed only on adjacent detachment event pairs whose area overlap ratio exceeds a threshold, rather than reprocessing all detachment events. Based on the adjusted event separation threshold, the time window corresponding to the event pair is located from the time interval sequence. Within this time window, the vibration waveform signal is re-extracted, and the adjusted threshold replaces the original threshold as the criterion for determining the superposition candidate pair. The superposition boundary segment identification process is then re-executed to obtain the updated superposition boundary segment.

[0132] In one possible implementation, if the same detachment event overlaps with multiple adjacent events in an area exceeding a threshold, then for all the threshold-exceeding events associated with that event, an adjusted event separation threshold is calculated, and the larger value is taken as the final separation threshold for that event, thus achieving a strict segmentation criterion within the event's time window. Based on the updated superposition boundary segment, the vibration waveform signal within the corresponding time window is re-divided into independent single vibration segments. After filtering out noise using wavelet denoising for each segment, the start and end times are extracted to obtain the time boundaries of each independent event in the re-identified continuous detachment of the central region.

[0133] Step S107: Based on the time and space boundaries of independent events, and combined with the time interval between two adjacent detachments, determine the high-risk boundary segment within the collapse range, delineate the priority support area using a grid division method, and generate a real-time early warning signal for rock mass blasting disturbance risk.

[0134] The start and end times of each independent event are read from the time boundaries of the re-identified independent events. The bounding box coordinates of each independent event are read from the corresponding spatial boundaries. For two spatially adjacent or overlapping independent events, the time interval between them is read from the time interval sequence. If the time interval is less than the adjusted event separation threshold, the adjacent or overlapping boundary segments between the bounding boxes of the two independent events are marked as high-risk boundary segments. All marked high-risk boundary segments are aggregated according to their coordinate positions to determine the distribution of high-risk boundary segments within the collapse impact range. Based on the distribution of high-risk boundary segments, a preset buffer distance is extended outward along the strike and dip directions, with each high-risk boundary segment as the center. The extended area is uniformly divided into grids with a preset grid size. Grid units falling within the buffer range of at least one high-risk boundary segment are marked as support candidate units. Adjacent support candidate units are merged into connected regions using eight-neighborhoods. The outer contour of each connected region is used as the boundary to delineate the priority support area. For each grid cell within the priority support area, the strain value of the corresponding grid node in the distribution map of the damaged area within the coverage area of ​​that cell is read. The number of grid cells whose strain values ​​exceed a preset strain risk threshold is counted. The ratio of this number to the total number of grid cells within the priority support area is used as a risk level indicator. If the risk level indicator exceeds 0.7, a real-time early warning signal for rock mass blasting disturbance risk is output for that priority support area.

[0135] In one implementation, the determination of the high-risk boundary segment is based on the joint execution of the temporal and spatial boundaries of independent events. For two independent detachment events that are spatially adjacent or overlapping, the coordinates of their bounding boxes are read respectively, and it is determined whether there is an intersection or close proximity of the coordinate intervals of the two bounding boxes in the direction or tendency direction. If so, the adjacent or overlapping line segments on the boundaries of the two events are extracted.

[0136] Specifically, the time interval between the pair of independent events is read from the time interval sequence and compared with an adjusted event separation threshold. When the time interval is less than the adjusted event separation threshold, it indicates that the pair of events is too close in time, and there is a high risk of concentrated damage at the intersection of the corresponding spatial boundaries. The adjacent or overlapping boundary segments are marked as high-risk boundary segments. After performing the above judgment on all adjacent independent event pairs one by one, all marked high-risk boundary segments are aggregated according to their coordinate positions to obtain the distribution of high-risk boundary segments within the collapse impact range. Further, the grid division is performed outward based on the high-risk boundary segments. Starting from the centerline of each high-risk boundary segment, a preset buffer distance is extended outward along the strike and dip directions. The buffer distance is determined based on the larger value between the rock mass structural surface spacing and the probe spacing. Within the area defined by the buffer distance, the grid is uniformly divided along the strike and dip directions with a preset grid size to form several grid units. Each grid unit is judged to see if it falls within the buffer range of at least one high-risk boundary segment. If it does, it is marked as a support candidate unit. The four or eight adjacent support candidate units in space are merged into a connected region. That is, starting from any support candidate unit, the region is gradually expanded to its adjacent support candidate units until it can no longer be expanded. The outer contour of the resulting connected region is the boundary of the support priority region.

[0137] It should be noted that the setting of the buffer distance directly affects the coverage of the priority support area. In the tunnel face scenario, when the spacing between structural surfaces is small and the detachment events are densely distributed, a smaller buffer distance is used to avoid excessive expansion of the support area; in the mine roof scenario, if the spacing between structural surfaces is large and the detachment events are sparsely distributed, a larger buffer distance is used to cover the potential crack propagation area.

[0138] In one embodiment, the grid size is determined based on a fraction of the probe spacing, such that the area of ​​a single grid cell is smaller than the area covered by a single probe, thereby achieving a higher spatial resolution in the priority support area than the probe deployment density.

[0139] Understandably, the risk level index reflects the severity of strain concentration within the priority support area. For each priority support area, the strain values ​​of the grid nodes in the damage area distribution map corresponding to the coverage area of ​​each grid unit within the area are read. The number of grid units whose strain values ​​exceed a preset strain risk threshold is counted, and the ratio of this number to the total number of grid units in the area is used as the risk level index. If the risk level index exceeds a preset early warning trigger threshold, a real-time early warning signal for rock mass blasting disturbance risk is output for that priority support area. The early warning signal carries the boundary coordinates of the priority support area and the risk level index value.

[0140] Specifically, this application involves various preset thresholds. The following explains the value range and determination method of each preset parameter:

[0141] The preset event separation threshold is used to determine whether two adjacent detachments belong to the same destructive event. The value ranges from 0.5 to 2.0 seconds, with the specific value determined based on the sampling frequency of the microseismic pickup unit and the typical duration of the rock mass detachment vibration signal. The calculation method is: Event Separation Threshold T_sep = k × (1 / f_s) × N_min, where f_s is the sampling frequency (Hz), N_min is the minimum number of sampling points that can identify a single detachment event (usually 50-100), and k is a safety factor (value 1.5-2.0). For example, when the sampling frequency is 1000Hz, the event separation threshold is 0.1-0.2 seconds.

[0142] The preset overlap ratio threshold is used to determine whether the spatial boundaries of adjacent independent events need to be re-identified. The value range is 0.3-0.6, with a recommended value of 0.5. When the area overlap ratio exceeds 0.5, it indicates a significant spatial overlap between two adjacent events, requiring dynamic adjustment of the threshold. This threshold can be calibrated based on the ratio between the distance between rock mass structural surfaces and the average coverage area of ​​the detachment event.

[0143] The density clustering algorithm parameter is the neighborhood radius Eps, which ranges from 0.5 to 2.0 times the distance between adjacent probes. In tunnel faces with dense joint development, Eps is 0.5 to 1.0 times the probe distance; in slopes with good integrity, Eps is 1.5 to 2.0 times the probe distance. - Minimum number of neighborhood points MinPts: The value ranges from 3 to 10, and the calculation method is MinPts = max(3, N_total / N_expected), where N_total is the total number of detachment points within the strain concentration zone, and N_expected is the expected number of groups. The expected number of groups is usually estimated based on the geometry of the strain concentration zone and is taken as 2 to 5.

[0144] The preset strain risk threshold is used to determine whether a grid cell belongs to a high-risk area. The value ranges from 60% to 80% of the rock mass's ultimate strain, with the specific value determined based on the mechanical properties of the rock material. For hard rocks such as granite, the strain risk threshold is 0.002-0.003; for soft rocks such as shale, the strain risk threshold is 0.005-0.008.

[0145] The preset value of the risk level indicator, i.e., the early warning trigger threshold, ranges from 0.5 to 0.8, with a recommended value of 0.7. An early warning signal is output when the proportion of grid cells exceeding the strain limit within the priority support area exceeds 0.7.

[0146] If the technical solution of this application involves the collection, storage, use, processing, transmission, provision, disclosure, or deletion of personal information, the products using this technical solution have clearly and understandably informed the users of the personal information processing rules before processing personal information, and have obtained the individuals' voluntary consent in accordance with the law. If the technical solution of this application involves sensitive personal information (such as biometrics, religious beliefs, specific identities, medical and health information, financial accounts, and location tracking), the products using this solution have obtained the individuals' separate consent before processing sensitive personal information, and have also met the requirement of "express consent," ensuring that individuals make authorization decisions voluntarily based on full knowledge.

[0147] Specific implementation methods include, but are not limited to, the following: setting up clear and prominent signs at personal information collection devices such as cameras and sensors to inform relevant personnel that they have entered the scope of personal information collection and that their personal information will be collected and processed. If an individual voluntarily enters the collection scope after being informed, it is deemed that they have agreed to the collection of their personal information; or using obvious icons, text descriptions, or other means on the terminal device or system interface for personal information processing to inform them of the rules for personal information processing, and obtaining the individual's explicit authorization through interactive methods such as pop-up prompts, check confirmation boxes, or asking the individual to upload their personal information themselves.

[0148] The aforementioned personal information processing rules should include, but are not limited to, the name and contact information of the personal information processor, the specific purpose of personal information processing, the processing method, the types of personal information processed, the retention period, and the methods and procedures for individuals to exercise their relevant rights.

[0149] The above-disclosed embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of the invention. Those skilled in the art will understand that implementing all or part of the above-described embodiments and making equivalent changes in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A method for real-time early warning of rock mass blasting disturbance risk, characterized in that, include: Displacement data, vibration waveform signals, and surface strain distribution data were collected at various points on the rock mass surface, and the location of the first detachment point and the time of subsequent detachment were recorded. Determine the distance between the initial detachment point and the geometric center of the rock mass. Identify the time interval between two adjacent detachments based on the time of each subsequent detachment. Use the ratio of the detachment distance to the time interval as an adjustment coefficient for the event separation threshold. When the time interval between two adjacent detachments is less than the preset event separation threshold, vibration signals are collected, overlapping parts in the vibration signals are identified and denoised to obtain the start and end time boundaries of a single detachment event. Based on the surface strain distribution data, the geometric shape of the strain concentration zone is determined. Combined with the start and end time boundaries of the single detachment event, the continuous detachment is grouped according to the detachment location using a clustering algorithm. The distance change feature between two adjacent detachment points is extracted. Based on the distance change feature, the spatial boundary of the independent detachment event is determined, and a preliminary outline of the collapse impact range is obtained. Based on the preliminary outline of the collapse-affected area, the coordinates of the boundary points are determined, the data-missing regions in the strain concentration zone are identified, the data-missing regions are filled using interpolation, and a complete damage area distribution map is generated. The area overlap ratio of adjacent areas in the damage area distribution map is evaluated. When the area overlap ratio exceeds the preset overlap ratio threshold, the event separation threshold is dynamically adjusted according to the adjustment coefficient of the event separation threshold. The time boundaries of each independent event in the continuous shedding in the central area are re-identified in combination with the adjusted event separation threshold. Based on the temporal and spatial boundaries of the independent events, and combined with the time interval between two adjacent detachments, high-risk boundary segments within the collapse impact range are determined, and a real-time early warning signal for rock mass blasting disturbance risk is generated.

2. The real-time early warning method for rock mass blasting disturbance risk according to claim 1, characterized in that, The data collected include displacement data, vibration waveform signals, and surface strain distribution data at various points on the rock mass surface. The location of the initial detachment and the timing of subsequent detachments are recorded, including: Based on the geometric contour of the rock mass surface, multiple displacement probes and microseismic pickup units are deployed at fixed intervals along the strike and dip. The coordinates of the installation points of each probe are obtained by total station measurement and written into the coordinate registration table. The displacement data, vibration waveform signal, and surface strain distribution data of each point are collected to obtain the original acquisition record with point coordinate markers; From the displacement data of the original acquisition record, read the hourly change of displacement amplitude point by point; When the change at a certain moment exceeds the preset mutation threshold, the corresponding moment is marked as a shedding trigger moment; The coordinates of the point corresponding to the earliest detachment trigger time in chronological order are taken as the location of the first detachment point. The subsequent detachment trigger times are arranged in chronological order to obtain a complete record of the location of the first detachment point and the times of each subsequent detachment.

3. The real-time early warning method for rock mass blasting disturbance risk according to claim 1, characterized in that, The process of determining the detachment distance between the initial detachment point and the geometric center of the rock mass, identifying the time interval between two adjacent detachments based on the times of subsequent detachments, and using the ratio of the detachment distance to the time interval as an adjustment coefficient for the event separation threshold includes: Read the three-dimensional coordinates of the first detachment point, calculate the Euclidean distance between the three-dimensional coordinates of the first detachment point and the three-dimensional coordinates of the geometric center of the rock mass, and obtain the detachment spacing. Read the trigger times of two adjacent detachments in chronological order to obtain the time interval between each pair of adjacent detachments, forming a time interval sequence; For each time interval in the time interval sequence, the shedding interval is divided by that time interval to obtain an adjustment coefficient that serves as the event separation threshold.

4. The real-time early warning method for rock mass blasting disturbance risk according to claim 3, characterized in that, When the time interval between two adjacent detachments is less than a preset event separation threshold, vibration signals are collected, overlapping portions in the vibration signals are identified and denoised, and the start and end time boundaries of a single detachment event are obtained, including: Read the time interval of each pair of adjacent detachments from the time interval sequence. If the time interval is less than the product of the preset event separation threshold and the adjustment coefficient, it is marked as a candidate pair for stacking. The vibration waveform signal within the corresponding time window of the superposition candidate pair is extracted to obtain the amplitude envelope curve; The location where the waveform amplitude crosses zero in the vibration waveform signal is detected to determine the overlapping boundary section; Based on the start and end times of the superimposed boundary section, the vibration waveform signal is divided into independent single vibration segments; The single vibration segment is decomposed and reconstructed using wavelet decomposition to obtain the vibration segment after noise removal; The start and end times of the single detachment event are obtained by extracting the start time when the amplitude of the vibration segment after noise filtering exceeds the preset start threshold and the end time when it falls back below the preset start threshold.

5. The real-time early warning method for rock mass blasting disturbance risk according to claim 1, characterized in that, Based on the surface strain distribution data, the geometric shape of the strain concentration zone is determined. Combined with the start and end time boundaries of each individual detachment event, a clustering algorithm is used to group consecutive detachments according to their location. Distance abrupt change features between two adjacent detachment points are extracted. Based on these distance abrupt change features, the spatial boundaries of independent detachment events are determined, resulting in a preliminary outline of the collapse-affected area, including: Read the surface strain distribution data at each point, and compare the strain value differences between adjacent points along the strike and dip directions. When the difference between the strain values ​​of two adjacent points exceeds the preset strain gradient threshold, the pair of points is marked as a high gradient point pair. All high gradient point pairs are connected sequentially according to their point coordinates to form strain gradient boundary lines. The area enclosed by the strain gradient boundary lines is taken as the strain concentration zone. The boundary point coordinates of the strain concentration zone are extracted to obtain the geometric shape of the strain concentration zone. Based on the start and end time boundaries of the single detachment event, only the coordinates of the detachment points that fall within the strain concentration zone are retained; Using the directional and dip components of each detachment point as two-dimensional features, density clustering algorithm is used to group spatially adjacent detachment points into the same detachment group, resulting in several detachment groups and the set of detachment point coordinates contained therein. For each of the aforementioned detachment groups, the coordinates of the detachment points within the group are arranged according to the order of the detachment occurrence time, and the Euclidean distance between two adjacent detachment points is calculated to form a successive detachment spacing sequence. When the Euclidean distance between a pair of adjacent detachment points exceeds a preset multiple threshold of the mean of the successive detachment interval sequence, the corresponding position is marked as a distance mutation point. The detachment group is then divided into several subgroups based on the distance mutation points, and each subgroup corresponds to the spatial boundary of an independent detachment event. For each subgroup, the extreme values ​​of the orientation and the extreme values ​​of the detachment point coordinates within the group are taken to form a rectangular bounding box as the spatial boundary of the independent detachment event; The rectangular bounding boxes of all subgroups are sequentially spliced ​​along their outer boundaries, and the area enclosed by the spliced ​​outer envelope is used as the preliminary outline of the collapse-affected area.

6. The real-time early warning method for rock mass blasting disturbance risk according to claim 1, characterized in that, Based on the preliminary outline of the collapse-affected area, the coordinates of boundary points are determined, and data-missing regions within the strain concentration zone are identified. Interpolation is then used to fill these data-missing regions, generating a complete damage area distribution map, including: Extract the coordinates of all boundary points on the outer envelope of the preliminary contour of the collapse-affected area; The area defined by the coordinates of the boundary points is the region, and regular grid nodes are arranged along the direction and dip. For each grid node, the nearest probe point within a preset search radius is found from the surface strain distribution data, the strain value of that probe point is read, and assigned to the corresponding grid node; Grid nodes without corresponding probe locations within the search radius are marked as missing nodes; The missing nodes were filled with strain values ​​using inverse distance weighted interpolation to obtain a complete strain mesh after filling. Based on the filled complete strain mesh, the strain value of each mesh node is mapped to a two-dimensional plane. The outer envelope of the preliminary contour is used as the boundary clipping line. Each mesh node within the clipping line is colored according to the strain value to obtain the complete damage area distribution map.

7. The real-time early warning method for rock mass blasting disturbance risk according to claim 1, characterized in that, Assess the area overlap ratio of adjacent areas in the damage area distribution map. When the area overlap ratio exceeds a preset overlap ratio threshold, dynamically adjust the event separation threshold according to the adjustment coefficient of the event separation threshold. Combine the adjusted event separation threshold with the re-identification of the time boundaries of each independent event in the continuous shedding of the central area, including: Read the spatial boundary corresponding to each independent detachment event, and use the rectangular bounding box enclosed by each spatial boundary as the coverage area; For two spatially adjacent coverage areas, the ratio of the overlapping area formed by the intersection of the directional and inclined directions to the reference area is calculated to obtain the area overlap ratio. When the area overlap ratio exceeds the preset overlap ratio threshold, the adjusted event separation threshold is obtained by multiplying the preset event separation threshold by the adjustment coefficient associated with the detachment event. Based on the adjusted event separation threshold, the overlapping boundary segment identification is re-executed for the time windows corresponding to adjacent detachment events in the central region where the area overlap ratio exceeds the threshold, resulting in an updated time boundary.

8. The method for real-time early warning of rock mass blasting disturbance risk according to claim 1, characterized in that, Based on the temporal and spatial boundaries of the independent events, and combined with the time interval between two adjacent detachments, high-risk boundary segments within the collapse impact range are determined, and a real-time early warning signal for rock mass blasting disturbance risk is generated, including: Read the start and end times of each independent event, as well as the coordinates of the corresponding bounding box; For two independent events that are spatially adjacent or overlapping, read the time interval between them; When the time interval is less than the adjusted event separation threshold, the adjacent or overlapping boundary segments between the rectangular bounding boxes of the two independent events are marked as high-risk boundary segments. Based on the distribution of all high-risk boundary segments, a buffer distance is extended outward from each high-risk boundary segment as the center, and grids are divided within the extended area. Grid cells falling within the buffer zone of at least one high-risk boundary segment are marked as support candidate cells and merged into connected regions to delineate priority support areas. The percentage of grid cells in the priority support area whose strain values ​​exceed a preset strain risk threshold is used as a risk level indicator. When the risk level index exceeds the preset value, a real-time early warning signal for rock mass blasting disturbance risk is output for the corresponding priority support area.

Images