An underground mine exploitation blasting safety risk intelligent identification and dynamic early warning method

Abstract

The application discloses a kind of underground mine mining blasting safety risk intelligent identification and dynamic early warning method, it is related to underground mine safety mining technical field.The method includes: based on geological data to establish three-dimensional mechanical model and carry out blasting vibration simulation, and divide into high, middle, low risk area;Multi-dimensional monitoring network is laid out, and real-time collection microseism, stress, separation, anchor rod axial force and toxic and harmful gas data;Data are preprocessed and extract low-frequency energy proportion, main band and gas concentration increment feature;Features and blasting parameters are input into depth forest model to output comprehensive risk grade;Surrounding rock-supporting structure response coupling model is constructed to calculate safety factor;Finally, according to risk grade, safety factor and gas concentration, generate graded dynamic early warning instruction, for adjusting blasting parameters, ventilation scheme or triggering emergency measures.The application realizes the real-time, accurate identification and closed-loop dynamic regulation and control of blasting safety risk.

Description

Technical Field

[0001] This invention relates to the technical field of safe mining in underground mines, and more specifically, to a method for intelligent identification and dynamic early warning of safety risks associated with blasting in underground mines. Background Technology

[0002] Blasting is one of the main mining methods in underground mining. However, due to the enclosed underground environment and complex and variable geological conditions, blasting operations pose extremely high risks of safety accidents, including surrounding rock instability, roof falls, rock bursts, and the release of toxic and harmful gases. Especially in high-gas mines or mines with sulfide deposits, blasting operations may trigger a sudden and massive release of harmful gases (such as CO, NO2, SO2, CH4, etc.), causing personnel poisoning or gas explosions. Existing risk management methods for blasting safety mostly rely on threshold monitoring of a single physical quantity (such as monitoring only blasting vibration velocity) or static assessment methods based on post-event analysis, making it difficult to achieve real-time, accurate perception and early warning of risks (including gas risks) throughout the entire blasting process.

[0003] Existing technologies, such as some early warning methods for underwater or open-pit blasting, while incorporating multi-dimensional parameters and machine learning models, have fundamentally different application scenarios compared to underground mines. Underwater blasting primarily focuses on the propagation of shock waves in water and their impact on hydraulic structures; open-pit blasting emphasizes the impact of flyrock and vibration on the surrounding environment. The core risk of underground mine blasting lies in the cumulative damage of dynamic loads on the surrounding rock-support structure system, and the potential for secondary disasters (such as gas outbursts and roof collapses). Existing technologies lack refined monitoring and coupled evaluation models for the dynamic response of the unique "surrounding rock-support" structure in underground mines, and also fail to deeply integrate and dynamically feedback control blasting parameters, geological conditions, real-time monitoring data, and toxic and harmful gas monitoring. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide an intelligent identification and dynamic early warning method for safety risks in underground mining blasting, which solves the technical problems in existing underground mining blasting safety risk identification and early warning methods, such as single monitoring dimensions, delayed risk identification, fixed and rigid early warning thresholds, lack of dynamic modeling of surrounding rock-support structure coupling, and weak gas risk prevention and control.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for intelligent identification and dynamic early warning of safety risks in underground mining blasting, comprising the following steps: Step S1: Based on the geological exploration data and mining engineering map of the underground mine blasting operation area, establish a three-dimensional geomechanical model, use the three-dimensional geomechanical model to conduct numerical simulation of blasting vibration effect, and combine blasting parameters and preset safe vibration velocity threshold to divide the blasting impact area into a high-risk core area, a medium-risk transition area and a low-risk general area. Step S2: Deploy a multi-dimensional physical parameter monitoring network within the affected zone, and collect multi-source monitoring data in real time before, during, and after blasting. The multi-source monitoring data includes microseismic signals, surrounding rock stress, roof delamination amount, anchor bolt (cable) axial force, and concentration of toxic and harmful gases. Step S3: Preprocess the multi-source monitoring data, including using a combined filtering algorithm to filter out environmental and equipment noise, and using wavelet packet transform to decompose the preprocessed microseismic signal into multiple scales to extract the low-frequency energy ratio and main frequency band features that characterize the blasting energy release and surrounding rock dynamic response; at the same time, perform baseline correction and abrupt change detection on the toxic and harmful gas concentration data to extract the instantaneous increment features of gas concentration. Step S4: Input the extracted features, preprocessed multi-source monitoring data and current blasting parameters into the pre-built risk intelligent identification model, output the comprehensive risk level of the current blasting operation, and generate an initial warning signal; Step S5: Combining the surrounding rock stress, the amount of delamination of the top plate, and the axial force of the anchor bolts (cables), construct a structural response coupling risk assessment model and calculate the safety factor of the surrounding rock-support structure system; Step S6: Based on the comprehensive risk level, safety factor, and monitoring data of toxic and harmful gas concentrations, generate graded dynamic early warning instructions. These instructions are used to adjust subsequent blasting parameters, adjust ventilation plans, or trigger emergency control measures.

[0006] Furthermore, the method for dividing the blasting impact zone in step S1 includes: Based on a three-dimensional geomechanical model, the propagation coefficient and attenuation index of blasting vibration waves in the ore rock mass were calibrated using numerical simulation methods. Based on the preset blasting parameters, the expected vibration velocity at each measuring point is calculated using the Sadovsky formula. Different safe vibration velocity thresholds are set for different protected objects. The thresholds include a safe upper limit and a safe lower limit. The area where the expected vibration velocity peak value is greater than the corresponding safe upper limit value is designated as the high-risk core area, the area where the expected vibration velocity peak value is between the safe lower limit value and the safe upper limit value is designated as the medium-risk transition area, and the area where the expected vibration velocity peak value is less than the safe lower limit value is designated as the low-risk general area.

[0007] Furthermore, the monitoring of toxic and harmful gas concentrations in step S2 includes: at least three multi-parameter gas sensors are arranged in a gradient on the return air side of the blasting face to monitor the concentrations of CO, NO2, SO2 and CH4 in real time. The sensor sampling frequency is not less than 1 Hz and the response time is not more than 20 seconds.

[0008] Furthermore, step S3 extracts the low-frequency energy proportion characteristics and dominant frequency band characteristics that characterize the blasting energy release and surrounding rock dynamic response, specifically including: The db wavelet basis function is selected to perform N-level wavelet packet decomposition on the microseismic signal and reconstruct the signal of each frequency band. Calculate the percentage of energy in the 0-50Hz frequency band to the total energy, as a characteristic of the low-frequency energy proportion; Identify and extract the dominant frequency with the largest amplitude and its corresponding energy value in the microseismic signal as the dominant frequency band feature.

[0009] Furthermore, the extraction of instantaneous gas concentration increment features in step S3 specifically includes: The difference between the maximum gas concentration within a preset time window after the blast and the background concentration before the blast is calculated and used as the instantaneous concentration increment. First-order difference processing was performed on the concentration time series data to extract the rising rate characteristics of gas concentration. When the instantaneous increase in CO concentration exceeds 24 ppm or the CH4 concentration exceeds 0.5%, it is marked as a gas anomaly event.

[0010] Furthermore, the risk intelligent identification model is constructed based on the deep forest algorithm. Its input feature set includes: low-frequency energy ratio characteristics, main frequency band characteristics, surrounding rock stress change rate, roof delamination rate, instantaneous increase in toxic and harmful gas concentration, maximum charge per segment, detonation interval time, and detonation center distance; the output is the comprehensive risk level.

[0011] Furthermore, the process of constructing the structural response coupling risk assessment model and calculating the safety factor in step S5 includes: establishing a dynamic response equation that considers the interaction between the surrounding rock and the support structure, wherein the input load is the time history curve of the blasting impact load obtained by inverting the microseismic signal and the blasting parameters; Solving the dynamic response equation yields the dynamic stress distribution at key parts of the support structure; The calculated dynamic stress is corrected based on the low-frequency energy ratio characteristics. If the low-frequency energy ratio exceeds a set threshold, the dynamic stress is amplified by a preset ratio. The safety factor is calculated by comparing the corrected dynamic stress with the yield strength of the support structure material or the allowable stress of the surrounding rock.

[0012] The generation of hierarchical dynamic early warning instructions in step S6 specifically includes: Set a first security threshold and a second security threshold, wherein the first security threshold is greater than the second security threshold; Set gas warning thresholds, including green, yellow, and red gas warning thresholds; When the safety factor is greater than the first safety threshold, the overall risk level is low, and the gas concentration is at the green gas warning threshold, the original blasting plan shall be maintained. A yellow warning instruction is generated when any of the following conditions are met: the safety factor is between the first safety threshold and the second safety threshold, or the comprehensive risk level is medium risk, or the gas concentration reaches the yellow gas warning threshold but is lower than the red gas warning threshold; the yellow warning instruction is used to suggest increasing the detonation interval, reducing the maximum charge per stage, or extending the ventilation time after blasting. A red warning command is generated when any of the following conditions are met: the safety factor is less than the second safety threshold, the comprehensive risk level is high risk, or the gas concentration reaches the red gas warning threshold; the red warning command is used to forcibly stop the blasting operation and initiate ventilation and personnel evacuation procedures.

[0013] A system for intelligent identification and dynamic early warning of safety risks in underground mining blasting includes: The monitoring network module consists of microseismic sensors, stress gauges, displacement gauges, and gas sensors deployed within the affected zone, used to collect multi-source monitoring data; The data acquisition and transmission module is used to filter, amplify, and convert sensor signals from analog to digital, and then transmit them to the data processing center after adding a unified timestamp. The signal processing and feature extraction module is used to perform wavelet packet decomposition, feature parameter calculation, baseline correction and abrupt change detection of gas concentration data; The risk identification and assessment module has a built-in intelligent risk identification model and a structural response coupled risk assessment model, which are used to output the comprehensive risk level and safety factor; The early warning and control module is used to generate graded early warning instructions based on the comprehensive risk level, safety factor and gas concentration data, and transmit them to the blasting control system, ventilation control system or on-site alarm device via wireless network. The database and self-learning module are used to store historical blasting data, monitoring data, and control effects, and to periodically update the risk intelligent identification model.

[0014] When a computer program is executed by a processor, it implements the steps of the method for intelligent identification and dynamic early warning of safety risks in underground mining blasting as described in any one of claims 1-8.

[0015] The beneficial effects of this invention are as follows: By establishing a three-dimensional geomechanical model and dividing the influence zones, the risk identification work is moved from post-event analysis to the blasting design stage, realizing the early prediction of risk sources; by integrating multi-dimensional physical parameters such as microseismic, stress, displacement, and gas, the one-sidedness of single-index monitoring is overcome, and the response characteristics of the surrounding rock-support system under blasting dynamic load and the generation and migration laws of toxic and harmful gases are more comprehensively depicted; by using wavelet packet transform to extract deep features such as the proportion of low-frequency energy closely related to blasting energy and structural resonance effect, and performing baseline correction and abrupt change detection on gas concentration data, high-quality input is provided for the intelligent identification model; a dynamic response model coupling blasting load, surrounding rock geology, and support structure is constructed, and the proportion of low frequency is introduced for correction, making the calculation of the safety factor closer to the actual underground engineering. Based on this, the comprehensive risk level and safety factor output by the deep forest and structural dynamic response model, combined with gas concentration monitoring data, generate graded early warning instructions and directly intervene in blasting parameters and ventilation schemes, realizing dynamic and intelligent risk control. In particular, it effectively prevents poisoning from toxic and harmful gases and gas explosion accidents after blasting, and significantly improves the inherent safety level of underground mine blasting operations. Attached Figure Description

[0016] Figure 1 Flowcharts of steps S1 to S3 of the intelligent identification and dynamic early warning method for blasting safety risks in underground mining provided in this embodiment of the invention; Figure 2 Flowcharts of steps S4 to S5 of the intelligent identification and dynamic early warning method for blasting safety risks in underground mining provided in this embodiment of the invention; Figure 3 The flowchart shows step S6 of the method for intelligent identification and dynamic early warning of safety risks in underground mining blasting provided in the embodiments of the present invention. Detailed Implementation

[0017] To explain in detail the technical content, objectives, and effects of the present invention, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0018] Please refer to Figures 1 to 3 A method for intelligent identification and dynamic early warning of safety risks in underground mining blasting includes the following steps: S1: Impact Zone Division; Based on geological exploration data and mining engineering maps of the underground mine blasting operation area, a three-dimensional geomechanical model is established. This model is used for numerical simulation of blasting vibration effects. Combining blasting parameters and a preset safe vibration velocity threshold, the blasting impact area is divided into a high-risk core zone, a medium-risk transition zone, and a low-risk general zone. Specifically, the propagation coefficient K and attenuation index α of the blasting vibration wave in the ore mass are calibrated through numerical simulation. Based on the preset blasting parameters and the Sadovsky formula V... pred =K⋅(Q 1 / 3 / R) α Calculate the expected peak vibration velocity at each measuring point; set differentiated safe vibration velocity thresholds for different protected objects, including an upper safety limit and a lower safety limit; designate areas with expected peak vibration velocity greater than the corresponding upper safety limit as high-risk core areas, areas with expected peak vibration velocity between the lower and upper safety limits as medium-risk transition areas, and areas with expected peak vibration velocity less than the lower safety limit as low-risk general areas; S2: Deployment of a multi-dimensional physical parameter monitoring network; A multi-dimensional physical parameter monitoring network shall be deployed within the aforementioned affected zone to collect multi-source monitoring data in real time before, during, and after blasting. This multi-source monitoring data includes microseismic signals, surrounding rock stress, roof delamination, anchor bolt (cable) axial force, and concentration of toxic and harmful gases. Among these, at least three multi-parameter gas sensors shall be deployed in a gradient on the return air side of the blasting face to monitor the concentrations of CO, NO2, SO2, and CH4 in real time. The sensor sampling frequency shall not be less than 1 Hz, and the response time shall not exceed 20 seconds. S3: Data Preprocessing and Feature Extraction; Multi-source monitoring data is preprocessed, and a combined filtering algorithm is used to remove environmental and equipment noise. Wavelet packet transform is used to decompose the preprocessed microseismic signals at multiple scales to extract low-frequency energy proportion features and dominant frequency band features characterizing blasting energy release and surrounding rock dynamic response. Specifically, the db wavelet basis function is selected to perform N-level wavelet packet decomposition on the microseismic signals to reconstruct signals in each frequency band. The percentage of energy in the 0-50Hz frequency band is calculated as the low-frequency energy proportion feature. The dominant frequency with the largest amplitude and its corresponding energy value in the microseismic signal are identified and extracted as the dominant frequency band feature. Simultaneously, baseline correction and abrupt change detection are performed on the toxic and harmful gas concentration data to extract the instantaneous gas concentration increment features. Specifically, this includes calculating the difference between the maximum gas concentration within a preset time window after blasting and the background value before blasting as the instantaneous concentration increment. First-order difference processing is performed on the concentration time series data to extract the gas concentration rise rate feature. When the instantaneous increase in CO concentration exceeds 24ppm or the CH4 concentration exceeds 0.5%, it is marked as a gas anomaly event. S4: Intelligent Risk Identification; The extracted low-frequency energy proportion characteristics, main frequency band characteristics, preprocessed multi-source monitoring data, and current blasting parameters are input into a pre-constructed intelligent risk identification model, which outputs the comprehensive risk level of the current blasting operation and generates an initial warning signal. This intelligent risk identification model is built based on the deep forest algorithm, and its input feature set includes low-frequency energy proportion characteristics, main frequency band characteristics, surrounding rock stress change rate, roof delamination rate, instantaneous increase in toxic and harmful gas concentration, maximum charge per segment, detonation interval time, and blast center distance. S5: Structural response coupling risk assessment; establishing the dynamic response equation Mẍ+Cẋ+Px=F(t). Here, the load F(t) is not a simplified sine function, but rather a more realistic time-history curve of the blast impact load obtained through inversion of microseismic signals and blasting parameters. Solving this equation yields the dynamic stress distribution σ of key components in the support structure, such as anchor bolts (cables) and linings. max Subsequently, the calculated dynamic stress is corrected using the low-frequency energy proportion characteristic η extracted in step S3: when η > 40%, a significant resonance amplification effect is considered to exist in the structure, and the stress is multiplied by a correction factor β = 1 + 0.005. (η-40), meaning that for every 1% increase in the proportion of low frequencies, the stress amplification is 0.5%. Finally, the safety factor K is calculated. safe =σy / σmax, where σ y The yield strength of the structural material or the allowable stress of the surrounding rock; S6: Generation of graded dynamic early warning instructions; Based on the comprehensive risk level, safety factor, and monitoring data of toxic and harmful gas concentrations, a graded dynamic early warning instruction is generated. This instruction is used to adjust subsequent blasting parameters, adjust ventilation plans, or trigger emergency control measures. Specifically, a first safety threshold and a second safety threshold are set, with the first safety threshold being greater than the second safety threshold. Gas warning thresholds are set, including green, yellow, and red gas warning thresholds. When the safety factor is greater than the first safety threshold, the comprehensive risk level is low, and the gas concentration is at the green gas warning threshold, the original blasting plan is maintained. A yellow warning instruction is generated when any of the following conditions are met: the safety factor is between the first and second safety thresholds, the comprehensive risk level is medium, or the gas concentration reaches the yellow gas warning threshold but is lower than the red gas warning threshold. The yellow warning instruction is used to suggest increasing the detonation interval, reducing the maximum charge per segment, or extending the post-blast ventilation time. A red warning instruction is generated when any of the following conditions are met: the safety factor is less than the second safety threshold, the comprehensive risk level is high, or the gas concentration reaches the red gas warning threshold. The red warning instruction is used to forcibly stop blasting operations and initiate ventilation and personnel evacuation procedures.

[0019] Example 1.

[0020] Step S1: Influence Zone Division. Taking an underground metal mine as an example, geological exploration data (including lithology, degree of joint and fracture development, fault distribution) and mining engineering maps are collected, and a three-dimensional geomechanical model is established using FLAC3D software. The propagation coefficient K and attenuation index α of blasting vibration waves in the ore body are calibrated through numerical simulation. Based on preset blasting parameters (e.g., maximum charge per section Q = 50 kg), the Sadovsky formula V is used. pred =K⋅(Q 1 / 3 / R) α Calculate the expected peak vibration velocity at each point of interest. The expected vibration velocity V in the main transport roadway 50m from the blast source is also calculated. pred =2.3cm / s. For the main transport roadways of this mine, a safe upper limit for vibration velocity is set at 1.5cm / s, and a lower limit at 0.8cm / s. Since Vpred = 2.3cm / s > 1.5cm / s, this roadway area is designated as a high-risk core area. Areas with expected vibration velocities between 0.8 and 1.5cm / s are designated as medium-risk transitional areas; areas with expected vibration velocities less than 0.8cm / s are designated as low-risk general areas.

[0021] Step S2: Deployment of a multi-dimensional physical parameter monitoring network. Within the high-risk core area, deploy microseismic sensors with a sampling rate of 500Hz, borehole stress gauges, roof delamination meters, and anchor bolt force gauges. Simultaneously, at distances of 5m, 10m, and 15m from the working face on the return air side of the blasting face, deploy three CO and CH4 gas sensors (using electrochemical principles, CO range 0-1000ppm, resolution 1ppm, response time ≤20 seconds, equipped with explosion-proof housings) at varying distances. All sensors undergo GPS time synchronization calibration.

[0022] Step S3: Data Preprocessing and Feature Extraction. The acquired microseismic signals are filtered using a combination of mean filtering and wavelet threshold denoising. A 5-level wavelet packet decomposition is performed using the db4 wavelet basis function, resulting in 32 frequency bands. The signals of each frequency band are reconstructed, and the percentage of energy in the 0-50Hz band is calculated to obtain the low-frequency energy proportion feature η. If η = 45%, it indicates a high risk of low-frequency resonance. For CO concentration data, the average value in the 5 minutes before the blast is calculated to be 2 ppm, and the maximum value in the 2 minutes after the blast is 35 ppm. The instantaneous concentration increase is 33 ppm, exceeding the yellow warning threshold of 24 ppm, and is marked as a gas anomaly. Simultaneously, the first-order difference of the concentration is calculated to obtain the rate of increase feature.

[0023] Step S4: Intelligent Risk Identification Based on Deep Forest. The extracted parameters η, dominant frequency f_max, surrounding rock stress change rate, roof delamination rate, instantaneous gas concentration increment, ascent rate, maximum charge per segment, detonation interval, and detonation center distance are combined to form a multi-dimensional feature set, which is then input into a pre-trained deep forest (gcForest) model. The model outputs the current comprehensive risk level as "medium risk" (Level II) and generates an initial warning signal.

[0024] Step S5: Structural response coupling risk assessment. Establish the dynamic response equation Mẍ + Cẋ + Px = F(t). Obtain the more realistic blast impact load time history curve F(t) through inversion using microseismic signals and blasting parameters. Solve the equation to obtain the dynamic stress σ at key locations on the anchor bolt. max =280MPa. Given the anchor bolt yield strength σ y =400MPa. Since η extracted in step S3 is 52% > 40%, the correction coefficient β is calculated as 1 + 0.005 × (52 - 40) = 1.06. The corrected σ max' =280 × 1.06 = 296.8 MPa. Safety factor K safe =400 / 296.8≈1.35.

[0025] Step S6: Generate graded dynamic early warning commands and execute control measures: Set the first safety threshold to 2.0 and the second safety threshold to 1.2; set gas early warning thresholds, where a yellow warning includes a CO concentration increase ≥ 24 ppm, a CH4 concentration ≥ 0.5%, or a NO2 concentration ≥ 5 ppm, and a red warning includes a CO concentration increase ≥ 50 ppm, a CH4 concentration ≥ 1.0%, or a NO2 concentration ≥ 10 ppm; the current state meets the conditions of a medium risk level, a safety factor of 1.35 between 1.2 and 2.0, and a CO concentration increase of 24 ppm (reaching the yellow gas early warning threshold), triggering a yellow early warning condition; the system generates a yellow early warning command and sends it to the blasting control system and ventilation control system via the downhole 5G network; the blasting control system automatically... The interval between the next detonation cycles was increased from 25ms to 38ms, and the maximum charge per segment was reduced from 50kg to 35kg. The ventilation control system extended the local ventilation time after blasting from 15 minutes to 25 minutes. At the same time, the audible and visual alarms in the roadway were activated to remind workers to wear self-rescue devices. After the adjustment, the monitoring of the next round of blasting showed that the proportion of low-frequency energy decreased to 35%, the safety factor increased to 2.2, the CO concentration increase decreased to 8ppm, all indicators returned to the green range, the risk level was reduced to low risk, and the system resumed the original plan. If the concentration of any gas reaches the red warning threshold, a red warning command is generated directly independently of other indicators, forcibly stopping the blasting operation and initiating the emergency ventilation and personnel evacuation procedures for the entire mine, realizing a closed-loop feedback of "monitoring-identification-early warning-control".

[0026] Example 2.

[0027] A computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements all steps of a method for intelligent identification and dynamic early warning of safety risks in underground mine blasting. The computer-readable storage medium is a non-volatile memory, including but not limited to solid-state drives, flash memory chips, or embedded storage modules, used to stably store the computer program for a long period in an underground mine monitoring system.

[0028] When the computer program is invoked by the processor, it first executes step S1: Based on the geological exploration data and mining engineering drawings of the underground mine blasting operation area, a three-dimensional geomechanical model is established; the model is used to perform numerical simulation of blasting vibration effects, and the propagation coefficient and attenuation index of blasting vibration waves in the ore and rock mass are calibrated; according to the preset blasting parameters, the expected vibration velocity at each measuring point is calculated in combination with the Sadovsky formula; differentiated safe vibration velocity thresholds are set for different protected objects, including safe upper limit and safe lower limit; areas where the expected vibration velocity peak value is greater than the corresponding safe upper limit value are classified as high-risk core areas, areas where the expected vibration velocity peak value is between the safe lower limit value and the safe upper limit value are classified as medium-risk transition areas, and areas where the expected vibration velocity peak value is less than the safe lower limit value are classified as low-risk general areas.

[0029] Then, step S2 is executed: a multi-dimensional physical parameter monitoring network is deployed within the aforementioned affected zone to collect multi-source monitoring data in real time before, during, and after blasting; among them, microseismic signals, surrounding rock stress, roof delamination, and anchor bolt axial force are obtained through sensors installed on the tunnel arch, sidewalls, and key support locations; the concentration of toxic and harmful gases is monitored through at least three multi-parameter gas sensors arranged in a gradient on the return air side of the blasting face, with a sensor sampling frequency of not less than 1Hz and a response time of not more than 20 seconds, used to detect the concentrations of CO, NO2, SO2, and CH4 in real time.

[0030] Next, step S3 is executed: preprocessing of multi-source monitoring data, using a combined filtering algorithm to filter out environmental and equipment noise; using the db wavelet basis function to perform N-level wavelet packet decomposition on the microseismic signal and reconstruct the signals of each frequency band, calculating the percentage of energy in the 0-50Hz frequency band to the total energy as the low-frequency energy proportion feature, and identifying the dominant frequency with the largest amplitude and its corresponding energy value as the dominant frequency band feature; simultaneously, baseline correction and abrupt change detection are performed on the toxic and harmful gas concentration data, calculating the difference between the maximum gas concentration and the background value before the blast within a preset time window after the blast to obtain the instantaneous concentration increment, and performing first-order difference processing on the concentration time series data to extract the rise rate feature; when the instantaneous increase in CO concentration exceeds 24ppm or the CH4 concentration exceeds 0.5%, it is marked as a gas anomaly event.

[0031] Then, step S4 is executed: the input feature set consisting of the extracted low-frequency energy ratio feature, main frequency band feature, surrounding rock stress change rate, roof delamination rate, instantaneous increase in toxic and harmful gas concentration, maximum charge per segment, detonation interval time and blast center distance is input into the risk intelligent identification model built based on the deep forest algorithm, outputs the comprehensive risk level of the current blasting operation, and generates an initial warning signal.

[0032] Then, step S5 is executed: combining the surrounding rock stress, the amount of delamination of the top plate, and the axial force of the anchor bolts, a structural response coupling risk assessment model is constructed; this model uses the time history curve of the blasting impact load obtained by inverting the microseismic signal and blasting parameters as the input load, and establishes a dynamic response equation considering the interaction between the surrounding rock and the support structure; the dynamic stress distribution of key parts of the support structure is obtained by solving the equation; the dynamic stress is corrected based on the low-frequency energy ratio characteristics, and if the low-frequency energy ratio exceeds the set threshold, the dynamic stress is amplified by a preset ratio; the corrected dynamic stress is compared with the yield strength of the support structure material or the allowable stress of the surrounding rock, and the safety factor of the surrounding rock-support structure system is calculated.

[0033] Finally, in step S6: Based on the comprehensive risk level, safety factor, and monitoring data of toxic and harmful gas concentrations, generate a graded dynamic early warning instruction; set a first safety threshold and a second safety threshold, with the first safety threshold being greater than the second safety threshold; set three levels of gas warning thresholds: green, yellow, and red; when the safety factor is greater than the first safety threshold, the comprehensive risk level is low, and the gas concentration is at the green gas warning threshold, maintain the original blasting plan; when the safety factor is between the first and second safety thresholds, or the comprehensive risk level is medium, or the gas concentration reaches the yellow gas warning threshold but not the red gas warning threshold, generate a yellow warning instruction to suggest increasing the detonation interval, reducing the maximum charge per segment, or extending the post-blast ventilation time; when the safety factor is less than the second safety threshold, or the comprehensive risk level is high, or the gas concentration reaches the red gas warning threshold, generate a red warning instruction to forcibly stop the blasting operation and initiate ventilation and personnel evacuation procedures.

[0034] In a preferred embodiment, the yellow gas warning threshold is set to an increase in CO concentration ≥ 24 ppm, CH4 concentration ≥ 0.5%, or NO2 concentration ≥ 5 ppm, and the red gas warning threshold is set to an increase in CO concentration ≥ 50 ppm, CH4 concentration ≥ 1.0%, or NO2 concentration ≥ 10 ppm. When the concentration of any gas exceeds the red warning threshold, the system directly triggers a red warning command independently of vibration and stress indicators.

[0035] The program stored in this computer-readable storage medium integrates multi-source monitoring data, physical models, and intelligent algorithms to achieve accurate identification and dynamic closed-loop control of safety risks throughout the entire underground mine blasting process, effectively improving the safety and controllability of blasting operations.

[0036] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent modifications made based on the content of the present invention specification and drawings, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for intelligent identification and dynamic early warning of safety risks in underground mining blasting, characterized in that, Includes the following steps: Step S1: Based on the geological exploration data and mining engineering map of the underground mine blasting operation area, establish a three-dimensional geomechanical model, use the three-dimensional geomechanical model to conduct numerical simulation of blasting vibration effect, and combine blasting parameters and preset safe vibration velocity threshold to divide the blasting impact area into a high-risk core area, a medium-risk transition area and a low-risk general area. Step S2: Deploy a multi-dimensional physical parameter monitoring network within the affected zone, and collect multi-source monitoring data in real time before, during, and after blasting. The multi-source monitoring data includes microseismic signals, surrounding rock stress, roof delamination amount, anchor bolt (cable) axial force, and concentration of toxic and harmful gases. Step S3: Preprocess the multi-source monitoring data, including using a combined filtering algorithm to filter out environmental and equipment noise, and using wavelet packet transform to decompose the preprocessed microseismic signal into multiple scales to extract the low-frequency energy ratio and main frequency band features that characterize the blasting energy release and surrounding rock dynamic response; at the same time, perform baseline correction and abrupt change detection on the toxic and harmful gas concentration data to extract the instantaneous increment features of gas concentration. Step S4: Input the extracted features, preprocessed multi-source monitoring data and current blasting parameters into the pre-built risk intelligent identification model, output the comprehensive risk level of the current blasting operation, and generate an initial warning signal; Step S5: Combining the surrounding rock stress, the amount of delamination of the top plate, and the axial force of the anchor bolts (cables), construct a structural response coupling risk assessment model and calculate the safety factor of the surrounding rock-support structure system; Step S6: Based on the comprehensive risk level, safety factor, and monitoring data of toxic and harmful gas concentrations, generate graded dynamic early warning instructions. These instructions are used to adjust subsequent blasting parameters, adjust ventilation plans, or trigger emergency control measures.

2. The method for intelligent identification and dynamic early warning of safety risks in underground mining blasting according to claim 1, characterized in that, The method for dividing the blasting impact zone in step S1 includes: Based on a three-dimensional geomechanical model, the propagation coefficient and attenuation index of blasting vibration waves in the ore rock mass were calibrated using numerical simulation methods. Based on the preset blasting parameters, the expected vibration velocity at each measuring point is calculated using the Sadovsky formula. Different safe vibration velocity thresholds are set for different protected objects. The thresholds include a safe upper limit and a safe lower limit. The area where the expected vibration velocity peak value is greater than the corresponding safe upper limit value is designated as the high-risk core area, the area where the expected vibration velocity peak value is between the safe lower limit value and the safe upper limit value is designated as the medium-risk transition area, and the area where the expected vibration velocity peak value is less than the safe lower limit value is designated as the low-risk general area.

3. The method for intelligent identification and dynamic early warning of safety risks in underground mining blasting according to claim 1, characterized in that, The monitoring of toxic and harmful gas concentrations in step S2 includes: at least three multi-parameter gas sensors are arranged in a gradient on the return air side of the blasting face to monitor the concentrations of CO, NO2, SO2 and CH4 in real time. The sensor sampling frequency is not less than 1 Hz and the response time is not more than 20 seconds.

4. The method for intelligent identification and dynamic early warning of safety risks in underground mining blasting according to claim 1, characterized in that, Step S3 extracts the low-frequency energy proportion and dominant frequency band characteristics that characterize the blasting energy release and surrounding rock dynamic response, specifically including: The db wavelet basis function is selected to perform N-level wavelet packet decomposition on the microseismic signal and reconstruct the signal of each frequency band. Calculate the percentage of energy in the 0-50Hz frequency band to the total energy, as a characteristic of the low-frequency energy proportion; Identify and extract the dominant frequency with the largest amplitude and its corresponding energy value in the microseismic signal as the dominant frequency band feature.

5. The method for intelligent identification and dynamic early warning of safety risks in underground mining blasting according to claim 1, characterized in that, Step S3 specifically includes extracting the instantaneous increase features of gas concentration: The difference between the maximum gas concentration within a preset time window after the blast and the background concentration before the blast is calculated and used as the instantaneous concentration increment. First-order difference processing was performed on the concentration time series data to extract the rising rate characteristics of gas concentration. When the instantaneous increase in CO concentration exceeds 24 ppm or the CH4 concentration exceeds 0.5%, it is marked as a gas anomaly event.

6. The method for intelligent identification and dynamic early warning of safety risks in underground mining blasting according to claim 1, characterized in that, The risk intelligent identification model is built based on the deep forest algorithm. Its input feature set includes: low-frequency energy ratio characteristics, main frequency band characteristics, surrounding rock stress change rate, roof delamination rate, instantaneous increase in toxic and harmful gas concentration, maximum charge per segment, detonation interval time and detonation center distance; the output is the comprehensive risk level.

7. The method for intelligent identification and dynamic early warning of safety risks in underground mining blasting according to claim 1, characterized in that, The process of constructing the structural response coupling risk assessment model and calculating the safety factor in step S5 includes: establishing a dynamic response equation that considers the interaction between the surrounding rock and the support structure, wherein the input load is the time history curve of the blasting impact load obtained by inverting the microseismic signal and the blasting parameters; Solving the dynamic response equation yields the dynamic stress distribution at key parts of the support structure; The calculated dynamic stress is corrected based on the low-frequency energy ratio characteristics. If the low-frequency energy ratio exceeds a set threshold, the dynamic stress is amplified by a preset ratio. The safety factor is calculated by comparing the corrected dynamic stress with the yield strength of the support structure material or the allowable stress of the surrounding rock.

8. The method for intelligent identification and dynamic early warning of safety risks in underground mining blasting according to claim 1, characterized in that, The generation of hierarchical dynamic early warning instructions in step S6 specifically includes: Set a first security threshold and a second security threshold, wherein the first security threshold is greater than the second security threshold; Set gas warning thresholds, including green, yellow, and red gas warning thresholds; When the safety factor is greater than the first safety threshold, the overall risk level is low, and the gas concentration is at the green gas warning threshold, the original blasting plan shall be maintained. A yellow warning instruction is generated when any of the following conditions are met: the safety factor is between the first safety threshold and the second safety threshold, or the comprehensive risk level is medium risk, or the gas concentration reaches the yellow gas warning threshold but is lower than the red gas warning threshold; the yellow warning instruction is used to suggest increasing the detonation interval, reducing the maximum charge per stage, or extending the ventilation time after blasting. A red warning command is generated when any of the following conditions are met: the safety factor is less than the second safety threshold, the comprehensive risk level is high risk, or the gas concentration reaches the red gas warning threshold; the red warning command is used to forcibly stop the blasting operation and initiate ventilation and personnel evacuation procedures.

9. A system for implementing the intelligent identification and dynamic early warning method for safety risks in underground mining blasting as described in any one of claims 1-8, characterized in that, include: The monitoring network module consists of microseismic sensors, stress gauges, displacement gauges, and gas sensors deployed within the affected zone, used to collect multi-source monitoring data; The data acquisition and transmission module is used to filter, amplify, and convert sensor signals from analog to digital, and then transmit them to the data processing center after adding a unified timestamp. The signal processing and feature extraction module is used to perform wavelet packet decomposition, feature parameter calculation, baseline correction and abrupt change detection of gas concentration data; The risk identification and assessment module has a built-in intelligent risk identification model and a structural response coupled risk assessment model, which are used to output the comprehensive risk level and safety factor; The early warning and control module is used to generate graded early warning instructions based on the comprehensive risk level, safety factor and gas concentration data, and transmit them to the blasting control system, ventilation control system or on-site alarm device via wireless network. The database and self-learning module are used to store historical blasting data, monitoring data, and control effects, and to periodically update the risk intelligent identification model.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When a computer program is executed by a processor, it implements the steps of the method for intelligent identification and dynamic early warning of safety risks in underground mining blasting as described in any one of claims 1-8.

Images