Intelligent evaluation method and pressure relief method for local impact risk of mining working face

Abstract

The invention discloses an intelligent evaluation method and a pressure relief method for local impact risk of an excavation working face. The method comprises the following steps: acquiring while-drilling monitoring data and coal body strength parameters of a to-be-monitored area; inputting the while-drilling monitoring data of the to-be-measured area and the coal body strength parameters into a pre-trained random forest model to obtain a borehole stress value output by the random forest model; the drilling stress value is compared with a set early warning value, when the drilling stress value is not lower than the early warning value, it is determined that the impact danger exists, and when the drilling stress value is lower than the early warning value, it is determined that the impact danger does not exist. According to the method, on the basis of real-time intelligent evaluation of the coal and rock mass strength and impact dangerousness while drilling and the pressure relief effect, the local drilling stress monitoring intelligent prediction model is researched and developed, regional micro-seismic monitoring is combined, regional and local combined space-time early warning and dynamic optimization of the pressure relief scheme are achieved, and the method has great significance in precise regulation and control of an underground stress field and optimization of the pressure relief scheme.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine excavation assessment, and in particular to an intelligent assessment method and a pressure relief method for local impact hazards in an excavation face. Background Art

[0002] As the depth of coal mine exploitation in China continues to extend below one thousand meters, mining activities have gradually entered areas with complex geological structures, and rock bursts have become the primary dynamic disasters threatening mine safety production. Under the "three highs" (high ground stress, high gas, and high osmotic pressure) environment of deep coal rock masses, the stress concentration phenomenon of coal bodies becomes more and more significant, and accurately grasping the stress distribution state of coal rock masses has become the core technical challenge for rock burst early warning and prevention. The current common methods for measuring coal body stress in mines mainly include the stress on-line monitoring method and the drill cuttings method: the former obtains the relative change amount of stress by burying sensors, but it is difficult to reflect the true stress field distribution due to the constraints of sensor installation technology and surrounding rock structure disturbance; the latter calculates the stress value based on the amount of drill cuttings discharged, but there are technical bottlenecks such as low coal powder collection efficiency and large positioning error of drill hole depth. Both methods cannot achieve accurate and real-time assessment of coal body stress. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide an intelligent assessment method and a pressure relief method for local impact hazards in an excavation face in view of the above deficiencies in the prior art, so as to solve the problems raised in the above background art.

[0004] To solve the above technical problems, the technical solution adopted by the present invention is: an intelligent assessment method for local impact hazards in an excavation face, including:

[0005] Obtaining the measurement-while-drilling monitoring data and coal body strength parameters of the area to be measured;

[0006] Inputting the measurement-while-drilling monitoring data and coal body strength parameters of the area to be measured into a pre-trained random forest model to obtain the borehole stress value output by the random forest model;

[0007] Comparing the borehole stress value with a set warning value. When the borehole stress value is not lower than the warning value, it is determined that there is an impact hazard. When the borehole stress value is lower than the warning value, it is determined that there is no impact hazard;

[0008] The random forest model is trained by obtaining the measurement-while-drilling monitoring data and coal body strength parameters from a number of previous coal body samples, and obtaining the borehole stress value of each coal body sample through experiments, so as to obtain a pre-trained random forest model.

[0009] Further, the training of the random forest model includes:

[0010] Obtain several previous coal samples;

[0011] Extract the measurement-while-drilling monitoring data and coal strength parameters of each of the coal samples, and obtain the borehole stress value of each of the coal samples through experiments;

[0012] Take the measurement-while-drilling monitoring data, coal strength parameters and corresponding borehole stress value of each of the coal samples as a set of data to obtain multiple sets of data;

[0013] Divide the multiple sets of data into a training set, a test set and a validation set according to a set ratio respectively;

[0014] Establish a random forest model, and input the data sets of the training set into the random forest model respectively in sequence to obtain a trained random forest model;

[0015] Use the data sets of the test set and the validation set to test and validate the trained random forest model respectively to obtain a validated random forest model.

[0016] Furthermore, the random forest model is composed of multiple decision tree models.

[0017] Furthermore, the inputting the data sets of the training set into the random forest model respectively in sequence includes:

[0018] Process the measurement-while-drilling monitoring data and coal strength parameters in the data sets of the training set through Bayes hyperparameter optimization and ten-fold cross-validation in sequence;

[0019] Use the processed measurement-while-drilling monitoring data and coal strength parameters in the data sets, along with their corresponding borehole stress values, to jointly train the random forest model.

[0020] Based on the same inventive concept, the present invention also provides a method for relieving local impact hazards in an excavation and working face, including:

[0021] Use the above-mentioned intelligent assessment method for local impact hazards in an excavation and working face to obtain the borehole stress value of the area to be measured;

[0022] Generate a distribution cloud map of the roadway side abutment pressure from the measurement-while-drilling monitoring data and borehole stress value of the area to be measured through a cloud platform on the industrial Ethernet;

[0023] Obtain the seismic waves of the area to be measured, and calculate the microseismic monitoring data and microseismic positioning based on the seismic waves;

[0024] Based on the distribution cloud map of the roadway side abutment pressure, combined with the microseismic monitoring data, judge the pressure relief and danger elimination effect of the area to be measured, and obtain a danger degree cloud map;

[0025] Based on the stress change in the danger level cloud map, determine whether it exceeds the set threshold. If it does not exceed the threshold, jump to the intelligent assessment method for local impact danger in a mining and excavation face described in any one of claims 1-4. If it exceeds the threshold, determine the dangerous local area based on the danger level cloud map and the microseismic positioning;

[0026] Based on the dangerous local area, determine the danger mitigation plan for the dangerous local area.

[0027] Further, after determining the danger mitigation plan for the dangerous local area, it further includes:

[0028] Form a comprehensive report based on the danger mitigation plan for the dangerous local area.

[0029] Further, the danger mitigation plan includes any one of the following: increasing the construction density of local large-diameter boreholes, increasing the construction density of local blasting holes, and high-pressure water injection in the dangerous area.

[0030] The present invention has the following advantages compared with the prior art:

[0031] The present invention provides an intelligent assessment method and a pressure relief method for local impact danger in a mining and excavation face, realizes the full information recording during the drilling process, improves the utilization efficiency of the drilling parameters, establishes a coal body stress prediction model based on the drilling parameters, realizes the intelligent identification of the drilling parameters and the test inversion of the coal body stress, improves the accuracy and timeliness of early warning, and promotes the intelligent monitoring of the borehole construction; realizes the full automation processing of the borehole parameters, reduces the manual work, improves the data processing efficiency, can quickly obtain the magnitude of the coal body stress, and timely evaluates the pressure relief and danger mitigation effect; realizes the combined time-space early warning of the area + local and the dynamic optimization of the pressure relief plan, and the real-time characterization of the danger mitigation effect, adopts the combined monitoring and early warning plan of microseismic monitoring (to master the dynamic load) and coal body drilling monitoring (to master the coal body loading characteristics), and optimizes the pressure relief plan in real time according to the pressure relief and danger mitigation effect; at the same time, the present invention has a more sensitive and timely test and evaluation plan for the coal body stress, minimizes the delay of the original stress test and analysis to the greatest extent, effectively improves the test accuracy, and ensures the correctness and timeliness of the test results; establishes the relationship between multiple drilling parameters and the coal body stress, effectively utilizes a large amount of effective data from the drilling monitoring, forms a fully automated analysis method, and real-time characterizes and evaluates the pressure relief and danger mitigation effect; combines regional monitoring means such as microseismic to realize combined monitoring and early warning, accurately feedbacks the pressure relief effect, can quantitatively evaluate the dynamic phenomenon, and realizes the dynamic optimization of the pressure relief plan.

[0032] The technical solution of the present invention will be further described in detail below through the drawings and embodiments. Brief Description of the Drawings

[0033] Figure 1 Schematic block diagram of the overall process of an intelligent evaluation method and a pressure relief method for local impact hazard in an excavation working face provided by the present invention.

[0034] Figure 2 Schematic structural diagram of the measuring device in the present invention.

[0035] Figure 3 Schematic diagram of the training process and structure of the random forest model in the present invention. Specific implementation manners

[0036] As Figures 1-3 shown, an intelligent evaluation method for local impact hazard in an excavation working face provided by the present invention includes:

[0037] Step 1: Obtain the measurement-while-drilling monitoring data and coal body strength parameters of the area to be measured;

[0038] Step 2: Input the measurement-while-drilling monitoring data and coal body strength parameters of the area to be measured into a pre-trained random forest model to obtain the borehole stress value output by the random forest model;

[0039] Step 3: Compare the borehole stress value with a set warning value. When the borehole stress value is not lower than the warning value, it is determined that there is an impact hazard. When the borehole stress value is lower than the warning value, it is determined that there is no impact hazard;

[0040] The random forest model is trained by obtaining the measurement-while-drilling monitoring data and coal body strength parameters from a number of previous coal body samples and obtaining the borehole stress value of each coal body sample through experiments to obtain a pre-trained random forest model.

[0041] The present invention forms a combined monitoring and warning scheme of microseismic monitoring for the roof (to master dynamic loads) and measurement-while-drilling monitoring for the coal body (to master the coal body loading characteristics) based on the intelligent identification of measurement-while-drilling parameters and the test inversion of coal body stress, and optimizes the pressure relief scheme according to the pressure relief and danger relief effect.

[0042] In Step 1, based on the measurement-while-drilling to obtain the coal body stress, first, coal body mechanical parameter tests need to be carried out according to different mines to obtain coal body strength mechanical parameters such as cohesion and internal friction angle (coal body strength parameters). Then, using large-diameter borehole measurement equipment, as Figure 2 shown, perform multiple borehole operations on the local area to obtain measurement-while-drilling test parameters (measurement-while-drilling monitoring data), namely cutting angle, cutting thickness, drilling displacement, drilling torque, drilling time, and bit rotation speed. Use a borehole stress gauge to obtain the magnitude of the coal body stress, and upload the measurement results to the computer in real time through the mine industrial ring network to construct an initial data set of measurement-while-drilling monitoring parameters and coal body stress

[0043] In step 2, a random forest regression prediction model optimized based on the Bayesian algorithm (Bayes - RF) is constructed. For the working condition of predicting coal seam stress with drilling - while - drilling parameters, the random forest algorithm can automatically handle non - linear relationships and feature interactions without complex feature engineering. It ensures strong anti - overfitting ability through voting of multiple decision trees, provides the ranking of feature importance, which is convenient for explaining key influencing factors. It is suitable for medium - and small - scale data. The rationality of features is verified in combination with geomechanics theory (such as the correlation between drilling resistance and stress), and the rationality of the model is verified by combining linear fitting or formulas. The prediction framework is as Figure 3 shown.

[0044] Among them, the training of the random forest model includes:

[0045] Obtain a number of previous coal seam samples;

[0046] Extract the drilling - while - drilling monitoring data and coal seam strength parameters of each coal seam sample, and obtain the borehole stress value of each coal seam sample through experiments;

[0047] Take the drilling - while - drilling monitoring data, coal seam strength parameters and the corresponding borehole stress value of each coal seam sample as a set of data to obtain multiple sets of data;

[0048] Divide the multiple sets of data into a training set, a test set and a validation set according to a set ratio respectively;

[0049] Establish a random forest model, and input each set of data in the training set into the random forest model in turn to obtain a trained random forest model;

[0050] Use each set of data in the test set and each set of data in the validation set to test and validate the trained random forest model respectively to obtain a validated random forest model.

[0051] Furthermore, the random forest model is composed of multiple decision tree models.

[0052] The step of inputting each set of data in the training set into the random forest model in turn includes:

[0053] Process the drilling - while - drilling monitoring data and coal seam strength parameters in each set of data in the training set through Bayes hyperparameter optimization and ten - fold cross - validation in turn;

[0054] Use the processed drilling - while - drilling monitoring data and coal seam strength parameters in each set of data, along with their corresponding borehole stress values, to jointly train the random forest model.

[0055] Using the dataset generated in the first step, taking the cutting angle, cutting thickness, drilling displacement, drilling torque, drilling time, and bit rotation speed of each borehole as characteristic parameters, and taking the measured coal seam stress of each borehole as the target feature, input them into the Bayes-RF model, normalize and divide the dataset, optimize the hyperparameters of the random forest model using the Bayes algorithm, and adopt a ten-fold cross-validation method to avoid overfitting of the model.

[0056] In step 3, the magnitude of the borehole stress based on the measurement while drilling is finally predicted, and the degree of local impact hazard is judged according to the magnitude of the stress warning value. When the predicted stress exceeds the warning value set by the mine, it is considered that there is an impact hazard; if it is lower, it is considered that there is no impact hazard.

[0057] Based on the above-mentioned intelligent evaluation method for local impact hazard in an excavation and working face, the present invention also provides a pressure relief method for local impact hazard in an excavation and working face, including:

[0058] Step 1: Use the above-mentioned intelligent evaluation method for local impact hazard in an excavation and working face to obtain the borehole stress value of the area to be measured;

[0059] Step 2: Generate a distribution cloud map of the roadway side abutment pressure through the cloud platform on the industrial Ethernet network for the measurement while drilling data and borehole stress value of the area to be measured;

[0060] Step 3: Obtain the seismic waves of the area to be measured, and calculate the microseismic monitoring data and microseismic positioning based on the seismic waves;

[0061] Step 4: Based on the distribution cloud map of the roadway side abutment pressure, combined with the microseismic monitoring data, judge the pressure relief and danger relief effect of the area to be measured, and obtain a danger degree cloud map;

[0062] Step 5: According to the stress change of the danger degree cloud map, judge whether it exceeds the set threshold. If it does not exceed the threshold, jump to the above-mentioned intelligent evaluation method for local impact hazard in an excavation and working face according to any one of claims 1-4. If it exceeds the threshold, the dangerous local area is obtained according to the danger degree cloud map and the microseismic positioning;

[0063] Step 6: Determine the danger relief plan for the dangerous local area according to the dangerous local area.

[0064] In step 1, through the above-mentioned intelligent evaluation method for local impact hazard in an excavation and working face, according to the collected and calculated data, the borehole stress value (predicted stress data) of the area to be measured is obtained.

[0065] In Step 2, the measurement-while-drilling monitoring data parameters and the predicted stress data are uploaded to the cloud platform via the industrial ethernet. After parameter analysis and processing, the data is exported using software. Based on the borehole location and borehole number, the borehole pressure relief area is determined. According to the magnitude of the predicted borehole stress value, a distribution cloud map of the roadway side abutment pressure is generated to visually identify the impact hazard level in the borehole area.

[0066] In Step 3, the basic data of microseismicity is obtained, such as microseismic monitoring data and microseismic positioning, which is calculated based on seismic waves and can be directly adopted using a professional calculation system.

[0067] In Steps 4 to 6, based on the results of the distribution cloud map of the roadway side abutment pressure, combined with regional microseismic monitoring, the pressure relief and hazard elimination effect is comprehensively determined. According to the stress changes in the pressure distribution cloud map, the pressure relief and hazard elimination effect is classified into three levels: ineffective level, relatively effective level, and significantly effective level. Specifically: when the stress concentration coefficient drops below 1.2 (the original stress concentration coefficient may be between 1.5 and 2.5 or even higher), and the energy frequency in the microseismic event monitoring area decreases by more than 20% compared to the case without pressure relief (the safety mining microseismic early warning index value set by the mine), it is considered that the stress of the surrounding coal and rock mass has been significantly reduced after borehole pressure relief, and the pressure relief effect is at the significantly effective level at this time; when the stress concentration coefficient is between 1.3 and 1.8, and the energy frequency in the microseismic event monitoring area decreases by 10% - 20% compared to the case without pressure relief, and the stress has a relatively obvious reduction, the pressure relief effect is at the relatively effective level at this time; when the stress concentration coefficient is above 1.8, and the microseismic energy frequency has no change or shows an upward trend compared to the case without pressure relief, it is considered that the stress of the coal and rock mass has hardly changed significantly, and the pressure relief effect is at the ineffective level at this time. It is necessary to optimize and adjust the pressure relief plan by combining local borehole positioning and microseismic positioning, increase the construction density of large-diameter boreholes and blasting holes locally. When any index value is significantly higher than the early warning value, pressure relief measures such as high-pressure water injection can be implemented in the regional scope to effectively reduce the coal body stress. That is, the hazard elimination plan includes any one of the following: increasing the construction density of local large-diameter boreholes, increasing the construction density of local blasting holes, and high-pressure water injection in the dangerous area.

[0068] After Step 6, a comprehensive report is formed based on the measurement-while-drilling monitoring data and construction conditions, or according to the hazard elimination plan for the dangerous local area.

[0069] The present invention relates to a method for measuring in-situ stress of coal body based on large-diameter drilling, which can real-time monitor and extract multi-parameter stress data such as drilling speed, torque, rotational speed, drilling pressure, and torsion during the drilling process of a large amount of coal bodies underground, realize the method for inverse analysis of rock mass mechanical parameters, form an in-situ stress test inversion based on intelligent identification and analysis of drilling process parameters, establish a quantitative relationship between in-situ parameters and rock mass mechanical parameters, including real-time intelligent evaluation of the strength and impact hazard of coal and rock masses based on in-situ drilling, and the unloading effect, develop an intelligent prediction model for local borehole stress monitoring, combine with regional microseismic monitoring, realize local precise monitoring to find and make up for the blind area of regional monitoring, and finally realize the dynamic optimization of the space-time early warning and unloading plan of regional + local combination, and real-time characterize the effect of danger removal. The present invention is of great significance for accurately regulating the underground stress field, dividing the impact hazard area, intelligently evaluating the early warning effect, realizing the quantitative evaluation of dynamic phenomena, and optimizing the unloading plan.

[0070] The above are only the preferred embodiments of the present invention, and do not impose any limitations on the present invention. Any simple modification, change, and equivalent structural change made to the above embodiments according to the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent assessment method for local impact hazard in an excavation working face, characterized in that, Including: Obtaining the measurement-while-drilling monitoring data and coal mass strength parameters of the area to be measured; Inputting the measurement-while-drilling monitoring data and coal mass strength parameters of the area to be measured into a pre-trained random forest model to obtain the borehole stress value output by the random forest model; Comparing the borehole stress value with a set warning value. When the borehole stress value is not lower than the warning value, it is determined that there is an impact hazard. When the borehole stress value is lower than the warning value, it is determined that there is no impact hazard; The random forest model is trained by obtaining the measurement-while-drilling monitoring data and coal mass strength parameters from a number of previous coal mass samples, and obtaining the borehole stress value of each coal mass sample through experiments, so as to obtain a pre-trained random forest model.

2. The intelligent evaluation method for local impact hazard in an excavation working face according to claim 1, wherein, The training of the random forest model includes: Obtaining a number of previous coal mass samples; Extracting the measurement-while-drilling monitoring data and coal mass strength parameters of each coal mass sample, and obtaining the borehole stress value of each coal mass sample through experiments; Taking the measurement-while-drilling monitoring data, coal mass strength parameters and corresponding borehole stress value of each coal mass sample as a set of data to obtain multiple sets of data; Dividing the multiple sets of data into a training set, a test set and a validation set according to a set ratio respectively; Establishing a random forest model, and inputting the data sets of the training set into the random forest model in sequence to obtain a trained random forest model; Using the data sets of the test set and the validation set to test and validate the trained random forest model respectively to obtain a validated random forest model.

3. The intelligent evaluation method for local impact hazard in an excavation working face according to claim 2, characterized in that, The random forest model is composed of multiple decision tree models.

4. An intelligent assessment method for local impact hazard in an excavation working face according to claim 2, characterized in that, The step of inputting the data sets of the training set into the random forest model in sequence includes: Processing the measurement-while-drilling monitoring data and coal mass strength parameters in the data sets of the training set through Bayes hyperparameter optimization and ten-fold cross-validation in sequence; Using the processed measurement-while-drilling monitoring data, coal mass strength parameters in the data sets and their corresponding borehole stress values to jointly train the random forest model.

5. A local shock hazard pressure relief method for an excavation working face, characterized in that Including: Using the intelligent evaluation method for local impact hazard in an excavation face according to any one of claims 1-4 to obtain the borehole stress value of the area to be measured; Generating a distribution cloud map of the roadway side abutment pressure from the measurement-while-drilling monitoring data and the borehole stress value of the area to be measured through a cloud platform on the industrial network; Obtaining the seismic waves of the area to be measured, and calculating the microseismic monitoring data and microseismic positioning according to the seismic waves; Based on the distribution cloud map of the roadway side abutment pressure, combining with the microseismic monitoring data, judging the pressure relief and danger elimination effect of the area to be measured, and obtaining a danger degree cloud map; Judging whether it exceeds a set threshold according to the stress change of the danger degree cloud map. If it does not exceed the threshold, jump to the intelligent evaluation method for local impact hazard in an excavation face according to any one of claims 1-4. If it exceeds the threshold, obtain the dangerous local area according to the danger degree cloud map and the microseismic positioning; Determining the danger elimination plan for the dangerous local area according to the dangerous local area.

6. A local shock hazard pressure relief method for an excavation face according to claim 6, characterized in that, After determining the danger elimination plan for the dangerous local area, it further includes: Forming a comprehensive report according to the danger elimination plan for the dangerous local area.

7. A local impact hazard pressure relief method for an excavation working face according to claim 6, characterized in that, The danger elimination plan includes any one of the following: increasing the construction density of local large-diameter boreholes, increasing the construction density of local blasting holes, and high-pressure water injection in the dangerous area.

Images