A borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden fractures

Abstract

The present invention discloses a borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden fissures, comprising: constructing n geoelectric models and high-resistance and low-resistance geoelectric models under full-space conditions; performing simulation calculations on the n geoelectric models and the high-resistance and low-resistance geoelectric models to obtain the apparent resistivity data sets corresponding to the n geoelectric models and the apparent resistivity data sets corresponding to the high-resistance and low-resistance geoelectric models as training sets and test sets; using the training sets and test sets to train and verify the BP neural network until the error between the output of the BP neural network and the resistivity set by the high-resistance and low-resistance geoelectric models meets the target error; using the trained BP neural network to perform full-space inversion on the geoelectric response data collected at different mining stages at the mining site, obtain the real resistivity data at the mining site, and determine the development range of overburden mining fissures at different mining stages. The present invention can accurately measure the development range and height of the overburden mining fissure zone.

Description

Technical Field

[0001] The present invention relates to the field of mining engineering and geotechnical engineering monitoring technology, and more particularly to a borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden cracks. Background Art

[0002] With the mining of coal seams, a large number of mining cracks are generated in the coal seam roof under the constantly changing stress disturbance. The overburden mining cracks are the channels for water inrush and gas migration in the coal seam roof. Therefore, it is very important to study the evolution characteristics of overburden mining cracks for the prevention and control of coal mine disasters.

[0003] The borehole resistivity method has the advantage of real-time dynamic monitoring in the dynamic monitoring of the evolution of overburden mining cracks, and can meet the needs of real-time monitoring of the development range and height of overburden mining cracks. However, the electric field data collected by the borehole resistivity method for in-situ monitoring is essentially the comprehensive geoelectric response of the surrounding rock under the full-space conditions of the overburden, while the resistivity inversion method based on the half-space theory and conventional resistivity inversion software can only achieve half-space resistivity inversion and imaging, and cannot achieve the requirements of full-space resistivity inversion and imaging, which leads to a large error between the detection results and the actual situation.

[0004] Therefore, how to provide a full-space borehole resistivity method for monitoring the evolution of overburden mining-induced fractures that can achieve accurate measurement of the development height of overburden water-conducting fracture zones is an urgent problem that technicians in this field need to solve. Summary of the invention

[0005] In view of this, the present invention provides a borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden cracks, which can accurately locate the development height of overburden mining cracks.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A borehole resistivity full-space imaging method for monitoring coal seam overburden fracture evolution, comprising:

[0008] According to the geological data and rock resistivity data of the unmined stage, the mined stage and the suspended stage in the mining process of the coal face, the formation attributes and rock resistivity attributes are determined, and n geoelectric models of the unmined stage, the mined stage and the suspended stage are constructed under full space conditions;

[0009] Construct high-resistance and low-resistance geoelectric models under full-space conditions;

[0010] Performing simulation calculations on n geoelectric models to obtain apparent resistivity data sets corresponding to the n geoelectric models as training sets; performing simulation calculations on the high-resistance and low-resistance geoelectric models to obtain apparent resistivity data sets corresponding to the high-resistance and low-resistance geoelectric models as test sets;

[0011] The pre-constructed BP neural network is trained using the training set until the minimum error of the preset training target is met;

[0012] The performance of the trained BP neural network is tested using the test set, and the minimum root mean square error is used to determine whether the test accuracy meets the preset value. If so, the final BP neural network is obtained;

[0013] The final BP neural network is used to perform full-space inversion on the geoelectric response data collected at different mining stages at the mining site to obtain the real apparent resistivity data of the mining site;

[0014] According to the variation law of the real resistivity data at the mining site, the development range and maximum development height of the overburden mining fractures at different mining stages are determined.

[0015] Furthermore, the high-resistance and low-resistance geoelectrical model includes: a single-layer high-resistance geoelectrical model and a double-layer high-resistance and low-resistance geoelectrical model; the single-layer high-resistance and low-resistance geoelectrical model includes a layer of high-resistance strata; the double-layer high-resistance and low-resistance geoelectrical model includes two strata, one of which is a high-resistance stratum whose resistivity is higher than the resistivity of the background field in the double-layer high-low group geoelectric model, and the other is a low-resistance stratum.

[0016] Furthermore, the construction process of the single-layer high-resistance geoelectric model is as follows:

[0017] A 500*500*500 three-dimensional uniform full-space model was established, and the resistivity of the three-dimensional uniform full-space model was regarded as the resistivity of the surrounding rock background field, which was set to 200Ωm;

[0018] A high-resistance stratum is established in the middle of the three-dimensional uniform full-space model, and the resistivity of the high-resistance stratum is set to 500Ωm.

[0019] Furthermore, the construction process of the double-layer high-resistance and low-resistance geoelectric model is as follows:

[0020] A 500*500*500 three-dimensional uniform full-space model was established, and the resistivity of the three-dimensional uniform full-space model was regarded as the resistivity of the surrounding rock background field, which was set to 200Ωm;

[0021] Two layers of strata are established in the middle of the three-dimensional uniform full-space model, one is a high-resistance stratum with a resistivity set to 500Ωm, and the other is a low-resistance stratum with a resistivity set to 100Ωm.

[0022] Furthermore, simulation calculations are performed on the n geoelectric models and the high-resistance and low-resistance geoelectric models, including:

[0023] For the n geoelectric models and the high-resistance and low-resistance geoelectric models, during the advancement of the coal mining working face, the resistivity value of the geoelectric model is adjusted every 2.5 m, and the step length of the rock formation resistivity change is 10Ω·m;

[0024] According to the above resistivity variation law, the apparent resistivity data sets of n geoelectric models in the non-mining stage, mining stage and suspension stage during the dynamic development of fractures and the apparent resistivity data sets of the high-resistance and low-resistance geoelectric models are obtained.

[0025] Furthermore, according to the variation law of the real resistivity data at the mining site, the development range and maximum development height of the overburden mining cracks at different mining stages are determined, including:

[0026] The real resistivity data of the mining site output by the BP neural network is processed by imaging, and different colors are used to represent the resistivity in different value ranges;

[0027] Analyze the change trend of rock resistivity at the mining site during the unmined stage, mining stage and suspension stage based on the imaging results;

[0028] According to the changing trend of rock resistivity, the development range of overburden mining cracks and the maximum development height of the fracture zone at the mining site at different mining stages were delineated.

[0029] Furthermore, the BP neural network includes: an input layer, two hidden layers and an output layer; the number of neurons in the hidden layer is 12; the activation function of the hidden layer is a sigmoid function, the training function is a trainscg function, the learning function is a learngd function, the performance function is an mse function, and the minimum error of the training target is 10 -6 , and the learning rate is 0.01.

[0030] It can be seen from the above technical scheme that, compared with the prior art, the present invention discloses a borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden cracks, determines the crack development law of overburden damage caused by mining according to previous geological data and rock resistivity, and establishes multiple three-dimensional geoelectric models under different advancement stages of the working face; at the same time, under full-space conditions, high-resistance and low-resistance geoelectric models are established, and high-resistance and low-resistance strata can approximately refer to high-resistance and low-resistance anomalies that appear in engineering practice; the apparent resistivity data sets obtained after simulation and calculation of the above models are used as training sets and test sets to train and verify the performance of the BP neural network, so that the trained BP neural network can invert the borehole resistivity method and image the borehole section under full-space conditions, realize full-space inversion and imaging of the borehole profile, and accurately locate the development height of the overburden mining crack zone, greatly improving the accuracy of dynamic monitoring of the borehole resistivity method, and can clearly reflect the development range and resistivity value of the "two zones" (caving zone and crack zone), and realize the fine measurement of mining overburden cracks. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0032] Figure 1 A flow chart of a method for monitoring the evolution of overburden mining cracks based on full-space borehole resistivity provided by the present invention;

[0033] Figure 2 The present invention provides a full-space imaging cross-section diagram of borehole resistivity based on a BP neural network when the working face is 40m away from the stop-mining line.

[0034] Figure 3 The present invention provides a full-space imaging cross-section diagram of borehole resistivity based on a BP neural network when the working face is 31 m away from the stop-mining line.

[0035] Figure 4 The present invention provides a full-space imaging cross-section diagram of borehole resistivity based on a BP neural network when the working face is 20.5 m away from the stop-mining line.

[0036] Figure 5 The present invention provides a full-space imaging cross-section diagram of borehole resistivity based on a BP neural network when the working face is 20m away from the stop-mining line.

[0037] Figure 6The present invention provides a full-space imaging cross-section diagram of borehole resistivity based on a BP neural network when the working face is 9.5 m away from the stop-mining line.

[0038] Figure 7 The present invention provides a full-space imaging cross-section diagram of borehole resistivity based on a BP neural network when the working face is 3.5 m away from the stop-mining line.

[0039] Figure 8 The present invention provides a full-space imaging cross-section diagram of borehole resistivity based on a BP neural network when the working face is 3m away from the stop-mining line.

[0040] Fig. 9 The present invention provides a full-space imaging cross-section diagram of borehole resistivity based on a BP neural network when the working face is 2m away from the stop-mining line.

[0041] Fig.10 The present invention provides a full-space imaging cross-section diagram of borehole resistivity based on a BP neural network when the working face is 0.8 m away from the stop-mining line.

[0042] Fig.11 The present invention provides a full-space imaging cross-section diagram of borehole resistivity based on a BP neural network when the working face is 0 m away from the stop-mining line. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] like Figure 1 As shown, the embodiment of the present invention discloses a borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden fractures, comprising the following steps:

[0045] S1. Design of geoelectric model parameters for dynamic monitoring of overburden mining fracture zone height and modeling at different advancement stages:

[0046] Based on the geological data and rock resistivity data of the unmined stage, the mined stage and the suspended stage in the mining process of the coal mining face, the formation properties and the rock resistivity properties are determined, and n geoelectric models of the unmined stage, the mining stage and the suspended stage are constructed; in the embodiment of the present invention, 900 geoelectric models are specifically constructed, and the geoelectric model is a layered geoelectric model under full-space conditions.

[0047] S2. Construct high-resistance and low-resistance geoelectric models under full space conditions;

[0048] S3. Finite element calculation to obtain training set and test set:

[0049] Simulate and calculate n geoelectric models to obtain apparent resistivity data sets corresponding to the n geoelectric models as training sets; simulate and calculate high-resistance and low-resistance geoelectric models to obtain apparent resistivity data sets corresponding to the high-resistance and low-resistance geoelectric models as test sets;

[0050] In this step, the simulation of n geoelectric models and high-resistance and low-resistance geoelectric models is based on the basic electric field theory to simulate the apparent resistivity response characteristics of the geoelectric model in the geoelectric field when a specific stratum or anomaly exists. It is the transformation from model data to apparent resistivity data set (that is, the forward modeling process).

[0051] S4, using the training set to train the pre-built BP neural network until the minimum error of the preset training target is met;

[0052] In the training phase, the network inputs the apparent resistivity data corresponding to the responses of n geoelectric models, and outputs the real resistivity data. During the training process, if the error of the BP neural network does not meet the minimum error of the training objective, the BP neural network parameters are further optimized until the minimum error of the training objective is met, and the optimally trained BP neural network is obtained.

[0053] S5. Using the test set to test the performance of the trained BP neural network, judging whether the test accuracy meets the preset value through the minimum root mean square error, and if so, obtaining the final BP neural network.

[0054] When testing the BP neural network, the apparent resistivity data sets of the high-resistance and low-resistance geoelectric models under uniform full-space conditions are used as test sets and input into the BP neural network. The test accuracy is determined by the minimum root mean square error. If the accuracy is high, it is determined that the BP neural network has high accuracy and can meet the inversion requirements. Otherwise, it indicates that the BP neural network has defects and needs further optimization.

[0055] S6. Using the final BP neural network to perform full-space inversion on the geoelectric response data collected at different mining stages at the mining site, obtain the real apparent resistivity data at the coal seam mining site;

[0056] S7. According to the variation pattern of the real resistivity data at the coal mining site, determine the development range of overburden mining fractures at different mining stages.

[0057] In one embodiment, the high-resistance and low-resistance geoelectric model includes: a single-layer high-resistance geoelectric model and a double-layer high-resistance and low-resistance geoelectric model; the single-layer high-resistance geoelectric model includes a layer of high-resistance stratum; the double-layer high-resistance and low-resistance geoelectric model includes two layers of stratum, one of which is a high-resistance stratum and the other is a low-resistance stratum.

[0058] For example, for a single-layer high-resistance geoelectric model, the construction process is:

[0059] A 500*500*500 three-dimensional uniform full-space model was established. The resistivity of the model was regarded as the resistivity of the surrounding rock background field and was set to 200Ωm.

[0060] A high-resistance stratum is established in the middle of the three-dimensional uniform full-space model, and the resistivity of the high-resistance stratum is set to 500Ωm.

[0061] For the double-layer high-resistance and low-resistance geoelectric model, the construction process is:

[0062] A 500*500*500 three-dimensional uniform full-space model was established, and the resistivity of the three-dimensional uniform full-space model was regarded as the resistivity of the surrounding rock background field, which was set to 200Ωm;

[0063] Two layers of strata are established in the middle of the three-dimensional uniform full-space model, one is a high-resistance stratum with a resistivity of 500Ωm, and the other is a low-resistance stratum with a resistivity of 100Ωm.

[0064] In one embodiment, simulation calculations are performed on n geoelectric models and high-resistance and low-resistance geoelectric models, including:

[0065] For n geoelectric models and high-resistance and low-resistance geoelectric models, the apparent resistivity value of the geoelectric model is adjusted every 2.5 m during the advancement of the coal mining face, and the step length of the rock resistivity change is 10Ω·m;

[0066] According to the above resistivity variation law, the apparent resistivity data sets of n geoelectric models in the non-mining stage, mining stage and suspension stage during the dynamic development of fractures and the apparent resistivity data sets of the high-resistance and low-resistance geoelectric models are obtained.

[0067] For example, as the coal mining face advances, caving zones and fracture zones begin to develop. The resistivity of the caving zone expands to 4 to 6 times of the original, and the resistivity of the fracture zone expands to 2 to 3 times of the original. According to the law of resistivity change, the apparent resistivity data sets of n geoelectric models in the dynamic development process of fractures in the non-mining stage, mining stage and suspension stage, as well as the apparent resistivity data sets of high-resistance and low-resistance geoelectric models are obtained.

[0068] The following is a further explanation using the numerical calculation of the working face mining to 0m from the borehole as an example:

[0069] When the area of ​​the caving zone and the fracture zone develops to the upper left and lower part of the monitoring borehole profile during coal seam mining, the corresponding geoelectric model is established, the surrounding rock resistivity is 100Ω·m, the caving zone width is 180m*15m, the fracture zone width is 180m*25m, and the layered strata are planned to be divided into 5 layers. The model stratum resistivity from bottom to top is 450Ω·m, 350Ω·m, 200Ω·m, 150Ω·m, and 100Ω·m. Because the caving zone and the fracture zone develop near the borehole mouth, the resistivity of the first and second layers of the rock strata expands to 1800Ω·m and 900Ω·m, so the actual stratum resistivity is 1800Ω·m, 900Ω·m, 200Ω·m, 150Ω·m, and 100Ω·m. The caving zone and fracture zone develop to the upper left and lower part of the borehole section, which is equivalent to a high-resistance anomaly in the layered background field. The top of the caving zone develops to the 1st to 12th electrodes, and the top of the fracture zone intersects with the survey line and is located near the 12th to 30th electrodes. The resistivity change of the model working face during the advancement process is forward calculated using the point power source theory and calculation method, and the apparent resistivity data required for the neural network inversion is calculated.

[0070] In one embodiment, according to the variation law of the real resistivity data at the mining site, the development range and development height of the overburden mining fractures at different mining stages are determined, including:

[0071] The real resistivity data of the mining site output by the BP neural network is imaged and processed, with different colors representing resistivity in different numerical ranges. After resistivity imaging, continuous high-resistance or low-resistance areas can be displayed. For example, the resistivity will increase after the development of cracks, and the height of the crack development can be determined based on the change in resistivity.

[0072] Analyze the change trend of rock resistivity at the mining site during the unmined stage, mining stage and suspension stage based on the imaging results;

[0073] According to the changing trend of rock resistivity, the development range of overburden mining cracks and the maximum development height of the fracture zone at the mining site at different mining stages were delineated.

[0074] In a specific embodiment, the BP neural network includes: an input layer, two hidden layers and an output layer; the number of neurons in the hidden layer is 12; the activation function of the hidden layer is a sigmoid function, the training function is a trainscg function, the learning function is a learngd function, the performance function is an mse function, and the minimum error of the training target is 10 -6 , and the learning rate is 0.01.

[0075] The present invention is further described below with reference to specific examples.

[0076] Based on the actual mining geological data of a coal mine, the geoelectric model and high-resistance and low-resistance geoelectric models and their corresponding resistivity data sets were established. After the BP neural network model was trained, the output data of the neural network model was fully inverted and imaged to finally determine the height of the fracture development. The instrument used for dynamic monitoring is a mine-based intrinsically safe network parallel electrical method system, which conducted 158 effective dynamic monitorings. During the data collection process, 48 ​​electrodes were used for data collection. The collected data was first decoded and processed, and then the measurement points with incorrect voltage attenuation trends in the original data were removed. This detection was carried out in the borehole, and no terrain correction was required.

[0077] Combined with the application of borehole resistivity method dynamic monitoring engineering in a coal mine working face, the trained BP neural network is used to perform full-space nonlinear inversion of the measured data. The imaging results are as follows: Figure 2-Figure 11 As shown in the figure, they represent the BP neural network inversion resistivity cross-section diagrams when the working face is 40m, 31m, 20.5m, 20m, 9.5m, 3.5m, 3m, 2m, 0.8m and 0m away from the stop mining line. The imaging results show that the overburden mining fracture zone has multiple development characteristics. The first stage is 20.5m away from the stop mining line, the developed fractures are relatively large, and the development height is determined to be 41m. The second stage is 3m away from the stop mining line, and the final development height is determined to be 53.5m.

[0078] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0079] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden fractures, characterized in that: include: According to the geological data and rock resistivity data of the unmined stage, the mined stage and the suspended stage in the mining process of the coal face, the formation attributes and rock resistivity attributes are determined, and n geoelectric models of the unmined stage, the mined stage and the suspended stage are constructed under full space conditions; Construct high-resistance and low-resistance geoelectric models under full-space conditions; Performing simulation calculations on the n geoelectric models to obtain apparent resistivity data sets corresponding to the n geoelectric models as training sets; Performing simulation calculations on the high-resistance and low-resistance geoelectric models to obtain apparent resistivity data sets corresponding to the high-resistance and low-resistance geoelectric models as test sets; The pre-constructed BP neural network is trained using the training set until the minimum error of the preset training target is met; The performance of the trained BP neural network is tested using the test set, and the minimum root mean square error is used to determine whether the test accuracy meets the preset value. If so, the final BP neural network is obtained; The final BP neural network is used to perform full-space inversion on the geoelectric response data collected at different mining stages at the mining site to obtain the real resistivity data of the mining site; According to the variation law of the real resistivity data at the mining site, the development range and maximum development height of the overburden mining cracks at different mining stages are determined; Performing simulation calculations on the n geoelectric models and the high-resistance and low-resistance geoelectric models, including: For the n geoelectric models and the high-resistance and low-resistance geoelectric models, during the advancement of the coal mining working face, the resistivity value of the geoelectric model is adjusted every 2.5 m, and the step length of the rock formation resistivity change is 10Ω·m; According to the above resistivity variation law, the apparent resistivity data sets of n geoelectric models in the non-mining stage, mining stage and suspension stage during the dynamic development of fractures and the apparent resistivity data sets of the high-resistance and low-resistance geoelectric models are obtained.

2. A borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden fractures according to claim 1, characterized in that: The high-resistance and low-resistance geoelectrical models include: a single-layer high-resistance geoelectrical model and a double-layer high-resistance and low-resistance geoelectrical model; the single-layer high-resistance geoelectrical model contains a layer of high-resistance strata; the double-layer high-resistance and low-resistance geoelectrical model contains two layers of strata, one of which is a high-resistance stratum and the other is a low-resistance stratum.

3. A borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden fractures according to claim 2, characterized in that: The construction process of the single-layer high-resistance geoelectric model is as follows: A 500*500*500 three-dimensional uniform full-space model was established, and the resistivity of the three-dimensional uniform full-space model was regarded as the resistivity of the surrounding rock background field, which was set to 200Ωm; A high-resistance stratum is established in the middle of the three-dimensional uniform full-space model, and the resistivity of the high-resistance stratum is set to 500Ωm.

4. The borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden fractures according to claim 2 is characterized in that: The construction process of the double-layer high-resistance and low-resistance geoelectric model is as follows: A 500*500*500 three-dimensional uniform full-space model was established, and the resistivity of the three-dimensional uniform full-space model was regarded as the resistivity of the surrounding rock background field, which was set to 200Ωm; Two layers of strata are established in the middle of the three-dimensional uniform full-space model, one is a high-resistance stratum with a resistivity set to 500Ωm, and the other is a low-resistance stratum with a resistivity set to 100Ωm.

5. The borehole resistivity full-space imaging method for monitoring the evolution of coal seam overburden fractures according to claim 1, characterized in that: According to the variation law of the real resistivity data at the mining site, the development range and maximum development height of the overburden mining cracks at different mining stages are determined, including: The real resistivity data of the mining site output by the BP neural network is processed by imaging, and different colors are used to represent resistivity in different value ranges; Analyze the change trend of rock resistivity at the mining site during the unmined stage, mining stage and suspension stage based on the imaging results; According to the changing trend of rock resistivity, the development range of overburden mining cracks and the maximum development height of the fracture zone at the mining site under different mining stages are delineated.

Images