A high slope blasting cumulative damage monitoring method

Abstract

The present application relates to the technical field of high slope blasting, in particular to a high slope blasting cumulative damage monitoring method, in the high slope to be monitored area, according to the occurrence of rock mass structure plane and the propagation direction of blasting vibration, adaptively arrange the multidimensional perception node including distributed optical fiber sensing network, microseismic monitoring array and hyperspectral imaging unit, and collect the initial reference data of each perception node;Before and after each blasting operation, respectively, pre-explosion detection and post-explosion detection are carried out on each perception node, and vibration response data, strain field data, microseismic event data, hyperspectral image data and slope three-dimensional shape data are synchronously acquired;The beneficial effect is that: by adaptively arranging multidimensional perception nodes, the synchronous collection and environmental decoupling of blasting vibration, rock mass strain, microseismic event and slope spectral characteristics are realized, the interference of temperature, rainfall and other environmental factors on damage quantification is effectively eliminated, and the accuracy of single damage index is fundamentally improved.

Description

Technical Field

[0001] This invention relates to the field of high slope blasting technology, specifically a method for monitoring cumulative damage during high slope blasting. Background Technology

[0002] In the construction of large-scale infrastructure projects such as highways, water conservancy, mining, and railways, blasting excavation of rock slopes is an indispensable construction method. However, when blasting loads are repeatedly applied to the rock mass, stress concentration and microcrack propagation occur near the structural planes, leading to the gradual accumulation of damage within the rock mass. When the accumulated damage exceeds a critical threshold, the slope may experience local spalling, block slippage, or even overall instability, resulting in serious safety accidents and economic losses.

[0003] Currently, slope blasting damage monitoring mainly relies on single or a few monitoring methods. For example, traditional methods using blasting vibration velocity as a control index only reflect the peak response of a single blast and cannot characterize the cumulative evolution of damage; methods mainly based on microseismic monitoring can capture internal rock fracture events, but are severely affected by environmental noise and have difficulty distinguishing between forced and natural microseismic events; methods mainly based on distributed fiber optic sensing can obtain continuous strain fields, but lack the accuracy to locate damage on internal rock structural surfaces; hyperspectral imaging can reflect surface damage signs such as changes in slope moisture content, but lacks the ability to perceive deep damage. All of the above single methods suffer from common defects such as insufficient information dimensions, difficulty in eliminating environmental interference, and lack of quantitative models for cumulative damage.

[0004] Furthermore, existing technologies for multi-source data fusion often employ simple weighted averaging or empirical threshold judgments, failing to fully consider the physical correlations and spatiotemporal weight differences between different monitoring dimensions, resulting in insufficient reliability of damage quantification results. In cumulative damage calculation, most methods use equal-weighted linear superposition, neglecting the nonlinear influence of rock mass stress relaxation and time effects on the damage accumulation rate. Regarding early warning and decision-making, existing systems are mostly passive alarm systems, lacking feedforward optimization capabilities based on trend prediction, and unable to proactively adjust blasting parameters before damage reaches a critical threshold, making true closed-loop control difficult.

[0005] Therefore, there is an urgent need for a slope blasting cumulative damage monitoring method that can integrate multi-dimensional sensing data, eliminate environmental interference, quantify non-equal weighted cumulative damage, and have the ability to predict trends and adaptively optimize schemes. Summary of the Invention

[0006] The purpose of this invention is to provide a method for monitoring cumulative damage from blasting on high slopes, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for monitoring cumulative damage from blasting on high slopes, comprising the following steps: Step S1: In the high slope monitoring area, according to the rock mass structure surface orientation and the direction of blasting vibration propagation, adaptively deploy multi-dimensional sensing nodes including distributed optical fiber sensing network, microseismic monitoring array and hyperspectral imaging unit, and collect the initial reference data of each sensing node. Step S2: Before and after each blasting operation, pre-blast detection and post-blast detection are carried out on each sensing node. Vibration response data, strain field data, microseismic event data, hyperspectral image data and three-dimensional slope morphology data are acquired simultaneously. Vibration reference nodes deployed on stable bedrock outside the blasting influence range are used to obtain pure blasting vibration source signal characteristics. Step S3: Perform environmental decoupling preprocessing on the multi-source data obtained in step S2, remove data drift caused by non-blasting factors such as temperature changes and rainfall infiltration, and perform cross-correlation analysis on the vibration source signal characteristics and slope response signal to separate the signal distortion component caused by rock mass damage. Step S4: Input the multi-dimensional damage feature data after environmental decoupling into the pre-trained lightweight damage recognition network, and combine it with hyperspectral differential fingerprint features to output the single-explosion damage quantification index of each sensing node. Step S5: Introduce a time decay-spatial coupling correction mechanism and an adaptive encryption strategy triggered by damage threshold. The single damage quantification index of each blast is accumulated and superimposed with non-equal weights, and independent damage verification is performed by combining microseismic spatiotemporal clustering analysis to obtain the corrected cumulative damage value of each sensing node. Step S6: Using digital twin-driven co-kriging interpolation, combined with the corrected cumulative damage value and the blasting vibration field distribution, construct a three-dimensional cumulative damage evolution model for the entire high slope, and use this model to determine the damage level, provide early warning of instability, and dynamically optimize the blasting scheme.

[0008] Preferably, in step S1, the adaptive deployment method is as follows: First, a three-dimensional scan of the internal structural surfaces of the high slope is performed using ground-penetrating radar to identify the orientation, dip, and inclination angle of the main structural surfaces; then, taking the main propagation direction of the blasting vibration as the axis, the initial spacing of the sensing nodes is determined along the normal direction of the structural surfaces according to the vibration attenuation law, and the nodes are automatically densified at the intersection of structural surfaces and at the corners of the slope; the distributed optical fiber sensing network is deployed in a spiral winding manner to fit the slope surface, so as to simultaneously acquire the vibration velocity and strain distribution along the path; the hyperspectral imaging unit is fixed on the UAV platform and flies along a fixed route before and after each blast to acquire a full-coverage hyperspectral image of the slope surface.

[0009] Preferably, in step S3, the specific method of environmental decoupling preprocessing is as follows: using reference nodes deployed outside the blasting influence range, temperature field change data and rainfall infiltration data are collected simultaneously; by constructing a temperature-strain coupling model and a rainfall-pore water pressure coupling model, the contribution of environmental factors to the strain data of each sensing node is calculated and eliminated; using the source mechanism solution analysis of microseismic events, forced microseismic events caused by blasting and natural seismic events caused by environmental factors are distinguished.

[0010] Preferably, in step S4, the lightweight damage recognition network adopts a multi-branch parallel architecture, including a vibration signal branch, a strain signal branch, a microseismic event branch, and a hyperspectral difference branch. After each branch extracts the deep features of the corresponding data, feature-level fusion is performed in the fusion layer through an attention weighting mechanism. The attention weights are adaptively adjusted according to the sensitivity of each parameter to damage under different lithological conditions. The training data of the network comes from indoor rock blasting simulation tests, calibration data of known damage cases in the field, and virtual damage scenarios generated by numerical simulation. During the training process, an adversarial example enhancement strategy is introduced to improve the model's ability to generalize and recognize unknown damage patterns.

[0011] Preferably, in step S5, the specific implementation of the time decay-spatial coupling correction mechanism is as follows: based on the time interval between two adjacent blasts and combined with the rock mass stress relaxation characteristic curve, the time decay correction factor is dynamically calculated. The shorter the time interval, the larger the decay factor, reflecting the accelerated accumulation effect of damage. Based on the spatial orientation, distance, and occlusion effect of the sensing node relative to the blast source and the intermediate structural surface, the spatial coupling correction coefficient is calculated so that nodes far from the blast source and subject to structural surface occlusion receive lower weights in the cumulative superposition. The non-equal weighted cumulative superposition is also constrained by the damage threshold: when the corrected cumulative damage value of a certain sensing node reaches the primary damage threshold for the first time, the monitoring density enhancement strategy of the area to which the node belongs is automatically triggered, and a temporary sensing node is activated in the area.

[0012] Preferably, in step S5, the independent damage verification method for the spatiotemporal clustering analysis of microseismic events is as follows: during the blasting interval, microseismic event data is continuously collected, and spatial scanning statistical methods are used to identify spatiotemporal clusters of microseismic events; for each cluster, the event density growth rate and energy release acceleration ratio are calculated; when the event density growth rate or energy release acceleration ratio of any cluster exceeds a critical value, it is determined that there is a natural damage evolution trend unrelated to blasting in that area; this natural damage trend is quantified as a natural damage increment, which is subtracted from the corrected cumulative damage value to obtain the pure blasting cumulative damage value.

[0013] Preferably, in step S6, the specific method of co-kriging interpolation driven by the digital twin is as follows: First, a geomechanical digital twin of the high slope is established, and the rock mass structure plane, lithological zoning, and initial stress field are embedded as prior information into the interpolation model; then, the corrected cumulative damage value of each sensing node is used as the observation sample, and the peak value of the blasting vibration velocity and the main frequency are used as covariates to perform co-kriging interpolation, generating a three-dimensional cumulative damage cloud map that simultaneously reflects the monitored and measured data and the physical propagation law of vibration; after each blast, the digital twin is updated using measured vibration data and microseismic data to achieve continuous synchronization between the twin model and the actual slope state.

[0014] Preferably, step S6 further includes cross-slope damage correlation and transmission analysis, specifically: the high slope is cut into several sub-blocks according to the rock mass structural plane, and the cumulative damage transmission between each sub-block through the structural plane is calculated; when the modified cumulative damage value of any sub-block exceeds the damage threshold corresponding to the shear strength of its structural plane, it is determined that the structural plane has been damaged and the modified cumulative damage values ​​of the sub-block and the adjacent sub-blocks are correlated and merged; by tracking the transmission path and connection sequence of damage between sub-blocks, the potential sliding mode and instability time window of the high slope are predicted.

[0015] Preferably, in step S6, the specific method for dynamically optimizing the blasting scheme is as follows: the modified cumulative damage value of each high-priority sensing node is extrapolated as the number of blasts changes, and the damage evolution trajectory after several future blasts is predicted using a long short-term memory network; when the predicted trajectory shows that the critical damage threshold will be reached within a preset number of blasts, a multi-dimensional blasting parameter adjustment scheme is automatically generated, including reducing the maximum charge per segment, optimizing the micro-delay initiation sequence, adjusting the borehole layout direction, and adding pre-splitting vibration damping holes; the adjustment scheme is input into a digital twin for scheme pre-simulation verification, and the optimal scheme that passes verification is pushed to the blasting designer for confirmation and execution, forming a closed-loop control of "monitoring-prediction-optimization-verification-execution".

[0016] Preferably, step S7 is also included: constructing a real-time visualization and graded early warning mechanism for the cumulative damage evolution trend, specifically: visualizing the calculation results of the three-dimensional cumulative damage evolution model in the form of dynamic evolution animation, and dividing the high slope into five levels according to the magnitude of the corrected cumulative damage value: intact area, lightly damaged area, moderately damaged area, severely damaged area, and critically unstable area; when any area is determined to be a severely damaged area or above, a graded early warning signal is automatically triggered, and the damage evolution trend of the area, the predicted instability time window, and the recommended blasting parameter adjustment scheme are simultaneously pushed to the site management personnel and blasting designers.

[0017] Compared with the prior art, the beneficial effects of the present invention are: The proposed method for monitoring cumulative damage from blasting on high slopes utilizes adaptive deployment of multi-dimensional sensing nodes to simultaneously acquire and decouple blasting vibration, rock mass strain, microseismic events, and slope spectral characteristics from the environment. This effectively eliminates the interference of environmental factors such as temperature and rainfall on damage quantification, fundamentally improving the accuracy of single-instance damage indicators. Furthermore, a lightweight damage identification network is used to perform attention-weighted fusion of the multi-dimensional decoupled data. Incorporating the stress relaxation characteristics of the rock mass, a non-equal-weighted time decay correction factor and an independent verification mechanism for the spatiotemporal clustering of microseismic events are introduced, achieving high-precision quantification of cumulative blasting damage and avoiding the limitations of traditional equal-weighted methods. The superposition method has the drawbacks of overestimating or underestimating damage. Furthermore, by generating a three-dimensional cumulative damage cloud map through co-kriging interpolation driven by digital twins, and combining it with sub-block division and structural surface transmission analysis, it is possible to identify the damage penetration path and potential sliding mode across sub-blocks, overcoming the limitation that single-point monitoring cannot reflect the spatial transmission law of damage. At the same time, the closed-loop optimization mechanism based on LSTM trend prediction and digital twin pre-simulation verification can automatically generate blasting parameter adjustment schemes and complete pre-simulation verification before the damage reaches the critical threshold, realizing the transformation from passive early warning to active intervention, and significantly improving the safety and economy of slope blasting construction. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the present invention clear and complete, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only some, not all, embodiments of the present invention, and are merely illustrative of the embodiments of the present invention. They are not intended to limit the embodiments of the present invention. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1: Cumulative Damage Monitoring of Blasting on a High Rock Slope of a Highway At kilometer marker K127+350 of a highway, there is a high rock slope with a maximum excavation height of 68 meters. The slope is mainly composed of alternating layers of moderately weathered sandstone and mudstone, and features three sets of dominant structural planes (J1: dip 215° / dip angle 42°; J2: dip 310° / dip angle 55°; J3: dip 140° / dip angle 38°). This slope requires 12 controlled blasting excavations, with a maximum charge of 480 kg per blast, using micro-delay initiation technology and an inter-stage delay of 25 ms.

[0021] Multi-dimensional adaptive deployment of sensing nodes: First, a three-dimensional scan of the internal structural surfaces of the slope is performed using ground-penetrating radar. This identifies a convergence zone between structural surfaces J1 and J2 at a slope height of 42m, and a free face formed by structural surface J3 at a slope height of 55m. Using the direction of blasting vibration from the blast source towards the slope normal as the axis, the spacing of sensing nodes is determined according to the vibration attenuation law along the structural surface normal direction: within 50m of the blast source, the fiber optic sensing node spacing is set to 5m; within 50m to 150m, the spacing is set to 10m; and above 150m, the spacing is set to 20m.

[0022] Three distributed fiber optic sensor networks, totaling approximately 1800m in length, are spirally wound to conform to the slope surface, covering the J1-J2 intersection zone and slope corners. The initial wave velocity is 4200m / s, and the initial strain is 12με. The microseismic monitoring array comprises 12 sensors, spaced 8m apart at the J1-J2 intersection zone and 15m apart in other areas. The background microseismic event rate is 3 events / day, with an energy range of 10. 2 ~10 3 J. One hyperspectral imaging unit, fixed to the UAV platform, flies along a fixed flight path before and after each blast to acquire full-coverage hyperspectral images of the slope. The initial NDVI index is 0.35, and the water content inversion value is 8.2%. Two vibration reference nodes are deployed on stable bedrock outside the blast influence range, with an initial vibration velocity of 0.1 cm / s.

[0023] Data acquisition and environmental decoupling: Taking the first blast as an example.

[0024] The pre-blast detection data are as follows: fiber optic sensing detected a slope vibration velocity of 0.3 cm / s and an average strain field of 15 με; microseismic monitoring detected 5 background events, all with energies below 10. 3 J; The NDVI index measured by hyperspectral analysis was 0.34, and the water content inversion value was 8.0%; 1.2 million point clouds were obtained for the three-dimensional morphology of the slope.

[0025] The post-explosion monitoring data are as follows: the peak vibration velocity detected by fiber optic sensing was 3.8 cm / s, and the peak strain field was 87 με; 186 microseismic events were detected, with a maximum energy of 5.2 × 10⁻⁶. 4 J; Hyperspectral analysis showed an NDVI index of 0.31 and a moisture content inversion value of 9.1%; localized spalling area of ​​approximately 12m² was detected in the three-dimensional morphology of the slope. 2 .

[0026] The environmental decoupling process is as follows: The reference node collected data showing a daily temperature change of +3.5°C and rainfall of 0 mm. Using the temperature-strain coupling model, the strain increment caused by the 3.5°C temperature increase was calculated to be 4.2 με. Subtracting this from the post-blast strain of 87 με, the net strain increment caused by the blast was obtained as 82.8 με. Through microseismic source mechanism analysis, 142 of the 186 microseismic events were forced microseisms, and 44 were natural microseisms, with a uniform distribution and energy below 10. 3 J, only 142 forced micro-seismic events were included in the subsequent calculations. Using cross-correlation analysis between the source signal and the slope response signal, the signal distortion component was separated, and the wave velocity reduction caused by the blast was found to be 180 m / s. This distortion variable was used as an independent input for damage quantification.

[0027] Single-instance damage quantification and cumulative calculation: Multidimensional data after environmental decoupling is input into a lightweight damage identification network, outputting single-instance damage quantification indices for each monitoring node during the first blast. The vibration damage index at the J1-J2 intersection zone at the slope crest is 0.12, the strain damage index is 0.18, the microseismic damage index is 0.15, and the hyperspectral damage index is 0.08. The fusion damage index after attention weighting by the fusion layer is 0.14. The fusion damage index at the middle of the slope is 0.05, and the fusion damage index at the slope toe is 0.02.

[0028] The non-equal weighted cumulative superposition process is as follows: The second blast was 72 hours after the first blast. According to the rock mass stress relaxation characteristic curve, the time decay correction factor is 1.15. The short interval reflects the accelerated damage accumulation effect. The single-blast fusion damage index for the second blast is 0.11. The cumulative damage value at the top of the slope after the second blast is calculated as: 0.14×1.0+0.11×1.15=0.14+0.127=0.267.

[0029] The independent verification process of microseismic spatiotemporal clustering is as follows: After the second blast, a microseismic cluster was identified in the J1-J2 intersection zone. The event density growth rate was 0.35 / day and the energy release acceleration ratio was 1.4, neither of which exceeded the critical value, which was 0.5 / day and 2.0, respectively. It was determined that there was no natural damage evolution trend unrelated to the blast in this area, the natural damage increment was 0, and the corrected cumulative damage value was still 0.267.

[0030] Since 0.267 does not reach the primary damage threshold of 0.30, the encryption strategy will not be triggered for the time being.

[0031] Digital twin modeling and early warning: Using digital twin-driven co-kriging interpolation, with the corrected cumulative damage values ​​of 12 microseismic monitoring nodes and 3 fiber optic lines as observation samples, and with the peak blast vibration velocity of 3.8 cm / s and the dominant frequency of 45 Hz as covariates, a three-dimensional cumulative damage cloud map that simultaneously reflects the monitored measurements and the physical propagation law of vibration is generated.

[0032] The damage level assessment results are as follows: The corrected cumulative damage value at the intersection of slope top J1-J2 is 0.267, classifying it as a slightly damaged area, with a yellow alert. The corrected cumulative damage value at the middle of the slope is 0.085, classifying it as an intact area, with a green alert. The corrected cumulative damage value at the toe of the slope is 0.042, classifying it as an intact area, with a green alert.

[0033] Example 2: Monitoring and Optimization of Cumulative Damage from Blasting on the Bank Slope of a Water Conservancy Project Dam The right bank slope of a certain water conservancy project dam has a maximum excavation height of 92m. The lithology is thick-layered limestone with two sets of dominant structural planes (J1: dip 195° / dip angle 35°; J2: dip 280° / dip angle 50°). It requires 20 blasting excavations in stages, with a maximum charge of 620kg per blast and a delay of 30ms between stages.

[0034] Sensing node deployment: After scanning with ground-penetrating radar, a microseismic monitoring array was densified to a spacing of 5 meters at the intersection of J1 and J2 (approximately 58m above the slope), with a total of 20 microseismic sensors deployed. The distributed fiber optic sensor network, approximately 2500m long, was spirally wound across the entire slope. The hyperspectral UAV flight path covered an area of ​​approximately 45,000 square meters. 2 Three vibration reference nodes were set up on stable bedrock 500m outside the blasting influence area.

[0035] The initial baseline data are as follows: initial wave velocity of the optical fiber is 5100 m / s, initial strain is 8 με; microseismic background event rate is 2 events / day, energy range is 10. 2 ~10 3 J; The hyperspectral initial moisture content inversion value is 5.6%.

[0036] Cumulative damage and scheme optimization after the 8th blast: After the first 7 blasts, the corrected cumulative damage value of the intersection zone J1-J2 at the top of the slope has reached 0.412, the primary damage threshold is 0.30, and the encryption strategy has been triggered. Four temporary sensing nodes have been activated in this area, and the spatial resolution has been improved to 3m.

[0037] The data from the 8th blast are as follows: the peak vibration velocity of the optical fiber after the blast was 4.5 cm / s, and the peak strain was 112 με; the net strain increment after environmental decoupling was 98 με, minus the 5.3 με caused by temperature; a total of 287 forced microseismic events occurred, with a maximum energy of 8.7 × 10⁻⁶. 4 J; The hyperspectral moisture content inversion value rose to 7.8%.

[0038] The lightweight damage identification network outputs the 8th single-sample fusion damage index: 0.165 for the slope crest intersection zone. Regarding time decay correction, the interval between the 7th and 8th samples is 48 hours, with a correction factor of 1.22. The cumulative damage value after the 8th sample is calculated as: 0.412 × 1.0 + 0.165 × 1.22 = 0.412 + 0.201 = 0.613.

[0039] Microseismic clustering analysis results: The density growth rate of clustered events in the J1-J2 intersection zone is 0.62 / day, and the energy release acceleration ratio is 2.3. It is determined that there is a natural damage evolution trend in this area, with a natural damage increment of 0.038. The corrected cumulative damage value is 0.613-0.038=0.575.

[0040] The LSTM trend prediction and scheme optimization process is as follows: The cumulative damage value of high-priority nodes is extrapolated using a Long Short-Term Memory (LSTM) network. The prediction results show that the measured cumulative damage value in the 8th test is 0.575, and the predicted cumulative damage value in the 12th test is 0.782, with 4 tests remaining before reaching the critical threshold (0.85). The predicted cumulative damage value in the 13th test is 0.891, exceeding the critical threshold. The prediction indicates that the critical damage threshold will be reached after the 13th blast. The system automatically generates a blasting parameter adjustment scheme: the maximum charge per segment is reduced from 620kg to 480kg, the inter-segment delay is increased from 30ms to 50ms, the borehole arrangement direction is adjusted from vertical to the slope to a 15° offset towards the J1 structural surface, and a row of pre-splitting vibration-damping holes is added.

[0041] The adjusted plan was input into the digital twin for pre-simulation verification. The verification results showed that the predicted cumulative damage value after the 13th blast was reduced to 0.721, and the cumulative damage value after the 20th blast was 0.818, which did not exceed the critical threshold of 0.85, and the verification was successful. The optimal plan was pushed to the blasting designer for confirmation and execution.

[0042] Example 3: Monitoring of Cumulative Damage from Blasting on an Open-Pit Slope in a Mine and Analysis of Sub-Block Transmission The slope of an open-pit mine has a maximum height of 115m. The lithology is interbedded granite and gneiss with 4 sets of structural planes. It requires 35 blasting excavations, with a maximum charge of 850kg per blast and an inter-section delay of 20ms.

[0043] Sensing node deployment and initial benchmark: The slope was divided into 8 sub-blocks (B1~B8) according to the attitude of the structural planes, among which the structural plane J3 (dipping 260° / dipping angle 48°) between B3 and B4 is a potential sliding surface. A total of 4 distributed optical fibers, with a total length of 3200m and an initial wave velocity of 5800m / s, were deployed; 24 microseismic arrays were used, with a background event rate of 4 times / day; one hyperspectral UAV was deployed, with an initial water content of 4.3%; and 3 vibration reference nodes were deployed on stable bedrock 800m away.

[0044] Damage propagation analysis of sub-blocks after the 20th blast: After the first 20 blasts, the three-dimensional cumulative damage cloud map generated by digital twin co-kriging interpolation shows the status of each sub-block as follows: B1 has a corrected cumulative damage value of 0.312, and the damage threshold corresponding to the shear strength of the structural surface is 0.50, which is judged as slight damage; B2 has a value of 0.278, threshold 0.48, slight damage; B3 has a value of 0.534, threshold 0.55, close to the critical point; B4 has a value of 0.489, threshold 0.52, moderate damage; B5 has a value of 0.198, threshold 0.45, intact; B6 has a value of 0.215, threshold 0.46, intact; B7 has a value of 0.356, threshold 0.49, slight damage; B8 has a value of 0.287, threshold 0.47, slight damage.

[0045] The damage transmission analysis process across sub-blocks is as follows: The cumulative damage transmission between B3 and B4 via the J3 structural plane is calculated. The damage value on the B3 side is 0.534, the damage value on the B4 side is 0.489, and the damage threshold corresponding to the shear strength of the J3 structural plane is 0.52. It is determined that the damage value of 0.534 on the B3 side exceeds the damage threshold of 0.52 for the J3 structural plane, and damage penetration is determined to have occurred at the J3 structural plane. The corrected cumulative damage values ​​of B3 and B4 are correlated and merged for calculation. The merged value = (0.534 + 0.489) × 0.85 = 0.870, where the weighting coefficient 0.85 reflects the damage coupling effect after the structural plane penetration.

[0046] The damage propagation path tracing results are as follows: The propagation quantity from B3 to J3 to B4 is 0.341, indicating a complete penetration; instability is predicted after the 25th to 28th blasts. The propagation quantity from B1 to J1 to B5 is 0.112, indicating a non-penetrating penetration; instability is predicted after the 32nd blast. The propagation quantity from B7 to J4 to B8 is 0.089, indicating a non-penetrating penetration; instability is predicted after the 35th blast. The system determines that the B3-B4 sub-block group is a high-risk sliding body, with a potential sliding mode of planar sliding along the J3 structural surface, and is predicted to enter a critical instability state after the 25th to 28th blasts.

[0047] Quantification of single-shot damage in the 20th blast: In the 20th blast, the vibration damage index of B3 (high-risk area) was 0.198, the strain damage index was 0.245, the microseismic damage index was 0.212, the hyperspectral damage index was 0.134, and the fusion damage index was 0.205. The fusion damage index of B4 was 0.153, and the fusion damage index of B1 was 0.075.

[0048] Regarding the time decay correction, the interval between the 19th and 20th times was 96 hours, and the correction factor was 0.88 (exceeding the stress relaxation characteristic time by 60 hours, reflecting the damage self-healing effect). The cumulative damage value of B3 after the 20th time was calculated as: 0.489×1.0+0.205×0.88=0.489+0.180=0.669.

[0049] Regarding independent microseismic verification: the cluster event density growth rate in region B3 is 0.71 / day (supercritical value 0.5 / day), the energy acceleration ratio is 2.6 (supercritical value 2.0), the natural damage increment is 0.045, and the corrected cumulative damage value is 0.669 - 0.045 = 0.624. This value deviates from the result of 0.534 in the digital twin model. The system automatically triggers a digital twin Bayesian update to incorporate the deviation into the model correction.

[0050] Example 4: Monitoring and Classified Early Warning of Cumulative Damage from Blasting on a Gravel Rock Slope Along a Railway Line A gravelly rock slope along a railway line, with a maximum height of 45m, consists of strongly weathered gravelly rock with highly developed joints. Eight blasting operations are required, with a maximum charge of 180kg per blast and a delay of 15ms between blasts. Due to its proximity to the existing railway line (only 18m from the slope toe), safety requirements are extremely high, and the critical damage threshold is set at 0.60 (lower than that of typical rock slopes).

[0051] Sensing node deployment and initial baseline: Two distributed optical fibers, totaling 900m in length, are spirally wound, primarily covering the railway side at the toe of the slope, with an initial wave velocity of 2800m / s and an initial strain of 25με. Eight microseismic arrays are deployed, with a spacing of 3m along the railway side at the toe of the slope, and a background event rate of 8 events / day. One hyperspectral UAV is deployed, flying once before and after each blast, with an initial water content of 14.2%. Two vibration reference nodes are deployed on stable bedrock 200m away, with an initial vibration velocity of 0.08cm / s.

[0052] Graded early warning and closed-loop optimization after the 5th blast: The data of the 5th blast are as follows: Before the blast, the peak optical fiber vibration velocity was 0.4 cm / s, the average strain was 28 με, there were 6 microseismic events, and the hyperspectral water content was 14.2%; After the blast, the peak optical fiber vibration velocity was 2.9 cm / s, the peak strain was 156 με, there were 234 microseismic events, and the hyperspectral water content was 17.6%.

[0053] The environmental decoupling process is as follows: The daily temperature rose by 5.2°C, and the rainfall was 2 mm. The temperature-strain coupling model calculated the strain increment caused by temperature to be 7.8 με, and the rainfall-pore water pressure coupling model calculated the strain increment caused by rainfall to be 3.2 με, deducting a total of 11 με. The net strain increment = 156 - 28 - 11 = 117 με. Microseismic focal mechanism analysis: Of the 234 earthquakes, 198 were forced microseisms and 36 were natural microseisms; only the 198 earthquakes were considered.

[0054] The quantitative results of single-instance damage are as follows: the fusion damage index at the toe of the slope on the railway side is 0.218, at the middle of the slope it is 0.095, and at the top of the slope it is 0.052.

[0055] The cumulative damage calculation process is as follows: After the 4th earthquake, the cumulative damage value on the railway side at the toe of the slope is 0.382. The interval between the 4th and 5th earthquakes is 36 hours, and the correction factor is 1.18. The cumulative value after the 5th earthquake = 0.382 × 1.0 + 0.218 × 1.18 = 0.382 + 0.257 = 0.639. Microseismic clustering analysis: The density growth rate of clustered events in the toe area is 0.85 / day (supercritical value 0.5 / day), the energy acceleration ratio is 3.1 (supercritical value 2.0), the natural damage increment is 0.052, and the corrected cumulative damage value = 0.639 - 0.052 = 0.587.

[0056] The five-level zoning and early warning push results are as follows: The corrected cumulative damage value on the railway side at the toe of the slope is 0.587, classifying it as a severely damaged area, triggering a red alert. The damage evolution trend, instability time window, and blasting adjustment plan are simultaneously pushed out. The corrected cumulative damage value in the middle of the slope is 0.246, classifying it as a lightly damaged area, triggering a yellow alert. A trend monitoring report is pushed out. The corrected cumulative damage value at the top of the slope is 0.138, classifying it as an intact area, triggering a green alert. A routine report is pushed out.

[0057] The closed-loop optimization process is as follows: LSTM prediction results show that, based on the current trend, the cumulative damage value on the railway side at the toe of the slope will reach 0.712 after the 7th blast (supercritical value 0.60), leaving only 2 safe blasting attempts. The system automatically generates an adjustment plan: the maximum charge per section is reduced from 180kg to 120kg (a reduction of 33.3%), the inter-section delay is extended from 15ms to 30ms (an extension of 100%), the borehole layout direction is adjusted from perpendicular to the slope to a 20° inward deviation, and 2 rows of pre-splitting vibration damping holes are added.

[0058] The digital twin simulation results are as follows: After adjustment, the predicted cumulative damage value for the 7th blast was 0.548, and for the 8th blast it was 0.611, still slightly exceeding the limit. The system further suggested splitting the 8th blast into two micro-blasts (each with a charge of 90kg). The final predicted cumulative damage value for the 8th blast was 0.556, below the critical threshold of 0.60, and the verification was successful. The optimal solution was simultaneously pushed to the on-site management personnel and blasting designers, requiring the solution adjustment to be completed before the 6th blast.

[0059] Example 1, for a 68m rock interbedded slope with 12 blasts, demonstrates the complete methodology. The initial cumulative damage increased from 0.14 to 0.267 through non-weighted cumulative growth. Example 2, for a 92m limestone slope with 20 blasts, demonstrates closed-loop optimization and scheme adjustment. After LSTM prediction exceeded the threshold at the 13th blast, reducing the charge by 22.6% controlled the cumulative damage at the 20th blast at 0.818. Example 3, for a 115m granite slope with 35 blasts, demonstrates sub-block conduction and penetration determination. After the B3-J3-B4 damage penetrated, the combined value reached 0.870, predicting instability after the 25th to 28th blasts. Example 4, for a 45m scree slope with 8 blasts, demonstrates graded early warning and emergency optimization. With only 2 safe blasts remaining, the charge was reduced by 33.3%, and a secondary split blasting scheme reduced the final cumulative damage to 0.556. The four embodiments validated the completeness, operability, and engineering applicability of the method from different dimensions.

[0060] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring cumulative damage from blasting on high slopes, characterized in that: Includes the following steps: Step S1: In the high slope monitoring area, according to the rock mass structure surface orientation and the direction of blasting vibration propagation, adaptively deploy multi-dimensional sensing nodes including distributed optical fiber sensing network, microseismic monitoring array and hyperspectral imaging unit, and collect the initial reference data of each sensing node. Step S2: Before and after each blasting operation, pre-blast detection and post-blast detection are carried out on each sensing node. Vibration response data, strain field data, microseismic event data, hyperspectral image data and three-dimensional slope morphology data are acquired simultaneously. Vibration reference nodes deployed on stable bedrock outside the blasting influence range are used to obtain pure blasting vibration source signal characteristics. Step S3: Perform environmental decoupling preprocessing on the multi-source data obtained in step S2, remove data drift caused by temperature changes and non-blasting factors such as rainfall infiltration, and perform cross-correlation analysis on the vibration source signal characteristics and slope response signal to separate the signal distortion component caused by rock mass damage. Step S4: Input the multi-dimensional damage feature data after environmental decoupling into the pre-trained lightweight damage recognition network, and combine it with hyperspectral differential fingerprint features to output the single-explosion damage quantification index of each sensing node. Step S5: Introduce a time decay-spatial coupling correction mechanism and an adaptive encryption strategy triggered by damage threshold. The single damage quantification index of each blast is accumulated and superimposed with non-equal weights, and independent damage verification is performed by combining microseismic spatiotemporal clustering analysis to obtain the corrected cumulative damage value of each sensing node. Step S6: Using digital twin-driven co-kriging interpolation, combined with the corrected cumulative damage value and the blasting vibration field distribution, construct a three-dimensional cumulative damage evolution model for the entire high slope, and use this model to determine the damage level, provide early warning of instability, and dynamically optimize the blasting scheme.

2. The method for monitoring cumulative damage from blasting on high slopes according to claim 1, characterized in that: In step S1, the adaptive deployment method is as follows: First, a three-dimensional scan of the internal structural surfaces of the high slope is performed using ground-penetrating radar to identify the orientation, dip, and inclination angle of the main structural surfaces; then, taking the main propagation direction of the blasting vibration as the axis, the initial spacing of the sensing nodes is determined along the normal direction of the structural surfaces according to the vibration attenuation law, and the nodes are automatically densified at the intersection of structural surfaces and at the corners of the slope; the distributed optical fiber sensing network is deployed in a spiral winding manner to fit the slope surface, so as to simultaneously acquire the vibration velocity and strain distribution along the path; the hyperspectral imaging unit is fixed on the UAV platform and flies along a fixed route before and after each blast to acquire a full-coverage hyperspectral image of the slope surface.

3. The method for monitoring cumulative damage from blasting on high slopes according to claim 1, characterized in that: In step S3, the specific method of environmental decoupling preprocessing is as follows: using reference nodes deployed outside the blasting influence range, temperature field change data and rainfall infiltration data are collected simultaneously; by constructing a temperature-strain coupling model and a rainfall-pore water pressure coupling model, the contribution of environmental factors to the strain data of each sensing node is calculated and eliminated; using the source mechanism solution analysis of microseismic events, forced microseismic events caused by blasting and natural seismic events caused by environmental factors are distinguished.

4. The method for monitoring cumulative damage from blasting on high slopes according to claim 1, characterized in that: In step S4, the lightweight damage recognition network adopts a multi-branch parallel architecture, including a vibration signal branch, a strain signal branch, a microseismic event branch, and a hyperspectral difference branch. After each branch extracts the deep features of the corresponding data, feature-level fusion is performed in the fusion layer through an attention weighting mechanism. The attention weights are adaptively adjusted according to the sensitivity of each parameter to damage under different lithological conditions. The training data of the network comes from indoor rock blasting simulation tests, calibration data of known damage cases in the field, and virtual damage scenarios generated by numerical simulation. During the training process, an adversarial example enhancement strategy is introduced to improve the model's ability to generalize and recognize unknown damage patterns.

5. The method for monitoring cumulative damage from blasting on high slopes according to claim 1, characterized in that: In step S5, the specific implementation of the time decay-spatial coupling correction mechanism is as follows: based on the time interval between two adjacent blasts and combined with the rock mass stress relaxation characteristic curve, the time decay correction factor is dynamically calculated. The shorter the time interval, the larger the decay factor, reflecting the accelerated accumulation effect of damage. Based on the spatial orientation, distance, and occlusion effect of the sensing node relative to the blast source and the intermediate structural surface, the spatial coupling correction coefficient is calculated so that nodes far from the blast source and subject to structural surface occlusion receive lower weights in the cumulative superposition. The non-equal weighted cumulative superposition is also constrained by the damage threshold: when the corrected cumulative damage value of a certain sensing node reaches the primary damage threshold for the first time, the monitoring density enhancement strategy of the area to which the node belongs is automatically triggered, and a temporary sensing node is activated in the area.

6. The method for monitoring cumulative damage from blasting on high slopes according to claim 1, characterized in that: In step S5, the independent damage verification method for the spatiotemporal clustering analysis of microseismic events is as follows: during the blasting interval, microseismic event data is continuously collected, and spatial scanning statistical methods are used to identify spatiotemporal clusters of microseismic events; for each cluster, the event density growth rate and energy release acceleration ratio are calculated; when the event density growth rate or energy release acceleration ratio of any cluster exceeds the critical value, it is determined that there is a natural damage evolution trend unrelated to blasting in the area; this natural damage trend is quantified as a natural damage increment, which is subtracted from the corrected cumulative damage value to obtain the pure blasting cumulative damage value.

7. The method for monitoring cumulative damage from blasting on high slopes according to claim 1, characterized in that: In step S6, the specific method of co-kriging interpolation driven by the digital twin is as follows: First, a geomechanical digital twin of the high slope is established, and the rock mass structure plane, lithological zoning, and initial stress field are embedded as prior information into the interpolation model; then, the corrected cumulative damage value of each sensing node is used as the observation sample, and the peak value of the blasting vibration velocity and the main frequency are used as covariates to perform co-kriging interpolation, generating a three-dimensional cumulative damage cloud map that simultaneously reflects the actual monitoring and the physical propagation law of vibration; after each blast, the digital twin is updated using measured vibration data and microseismic data to achieve continuous synchronization between the twin model and the actual slope state.

8. The method for monitoring cumulative damage from blasting on high slopes according to claim 1, characterized in that: Step S6 further includes cross-slope damage correlation and transmission analysis, specifically: the high slope is cut into several sub-blocks according to the rock mass structural plane, and the cumulative damage transmission between each sub-block through the structural plane is calculated; when the modified cumulative damage value of any sub-block exceeds the damage threshold corresponding to the shear strength of its structural plane, it is determined that the structural plane has been damaged through, and the modified cumulative damage values ​​of the sub-block and the adjacent sub-blocks are correlated and merged for calculation; by tracking the transmission path and connection sequence of damage between sub-blocks, the potential sliding mode and instability time window of the high slope are predicted.

9. The method for monitoring cumulative damage from blasting on high slopes according to claim 1, characterized in that: In step S6, the specific method for dynamically optimizing the blasting scheme is as follows: the modified cumulative damage value of each high-priority sensing node is extrapolated as the number of blasts changes, and the damage evolution trajectory after several future blasts is predicted using a long short-term memory network; when the predicted trajectory shows that the critical damage threshold will be reached within a preset number of blasts, a multi-dimensional blasting parameter adjustment scheme is automatically generated, including reducing the maximum charge per segment, optimizing the micro-delay initiation sequence, adjusting the borehole layout direction, and adding pre-splitting vibration damping holes; the adjustment scheme is input into a digital twin for scheme pre-simulation verification, and the optimal scheme that passes verification is pushed to the blasting designer for confirmation and execution, forming a closed-loop control of "monitoring-prediction-optimization-verification-execution".

10. The method for monitoring cumulative damage from blasting on high slopes according to claim 1, characterized in that: It also includes step S7: constructing a real-time visualization and graded early warning mechanism for the cumulative damage evolution trend. Specifically, the calculation results of the three-dimensional cumulative damage evolution model are visualized in the form of dynamic evolution animation. At the same time, the high slope is divided into five levels according to the magnitude of the corrected cumulative damage value: intact area, lightly damaged area, moderately damaged area, severely damaged area, and critical instability area. When any area is determined to be severely damaged or above, a graded early warning signal is automatically triggered, and the damage evolution trend, predicted instability time window, and recommended blasting parameter adjustment scheme of the area are simultaneously pushed to the site management personnel and blasting designers.

Images