Coal mine micro-seismic electrical method coupling monitoring data joint inversion method

Abstract

The invention provides a coal mine micro-seismic electrical method coupling monitoring data joint inversion method, which comprises the following steps of adopting micro-seismic event triggering type electrical method data acquisition, dynamically adjusting an electrical method sampling frequency according to micro-seismic energy, and carrying out normalization processing on data; constructing a combined inversion model of micro-seismic and electrical methods, and establishing quantitative correlation between micro-seismic energy and resistivity gradient; a mixed optimization solver adopts a hierarchical iteration strategy, and GPU parallel design is carried out; and a dynamic feedback and early warning and real-time updating mechanism is adopted, and when a trigger condition is detected based on an inversion result, early warning is triggered. The method is based on a multi-physics field cooperative constraint and dynamic feedback mechanism, and has the characteristics of high precision and real-time performance.

Description

Technical Field

[0001] The present invention mainly relates to the technical field related to coal mine disaster early warning, and specifically is a method for joint inversion of coal mine microseismic electrical coupling monitoring data. Background Art

[0002] When mining underground, early warning of water disasters in the coal mine roof is very important. Conventional early warning methods mainly include microseismic inversion method and electrical inversion method. The microseismic method mainly monitors the development of surrounding rock cracks caused by mining through sensors, and the electrical method, also known as the DC resistivity method, mainly monitors the water content or water conductivity of the surrounding rock.

[0003] In current technology, microseismic inversion methods and electrical inversion methods are generally independent of each other and are deployed separately. Simple microseismic inversion and electrical inversion methods are time-consuming and have a high false alarm rate for water inrush warnings, so it is necessary to improve traditional warning methods. Summary of the invention

[0004] In order to solve the shortcomings of the current technology, the present invention combines the existing technology and, starting from practical application, provides a method for joint inversion of coal mine microseismic electrical coupling monitoring data, which is based on multi-physical field collaborative constraints and dynamic feedback mechanism and has the characteristics of high precision and real-time performance.

[0005] The technical solution of the present invention is as follows:

[0006] A method for joint inversion of coal mine microseismic electrical coupling monitoring data is as follows:

[0007] (1) Microseismic event-triggered electrical data acquisition is adopted, the electrical sampling frequency is dynamically adjusted according to the microseismic energy, and the data is normalized;

[0008] (2) Construct a joint inversion model of microseismic and electrical methods to establish a quantitative correlation between microseismic energy and resistivity gradient;

[0009] (3) Hybrid optimization solver, using a hierarchical iteration strategy and GPU parallelization design;

[0010] (4) Dynamic feedback and early warning are adopted, and the real-time update mechanism is updated. When the triggering conditions are detected based on the inversion results, the early warning is triggered.

[0011] Furthermore, in step (1), the corresponding expression for microseismic event-triggered electrical data acquisition is as follows:

[0012] T 电法 =T 微震 +Δt·log(E / E0)

[0013] Where, T 电法 is the time parameter obtained by electrical measurement, T微震 is the time parameter obtained from microseismic measurement, Δt is the time difference, which indicates the time difference between the electrical method and the microseismic measurement, E0 is the reference electric field strength, which is used for normalization or comparison, and E is the current electric field strength, which represents the energy state during electrical measurement.

[0014] Furthermore, in step (1), the data is normalized as follows:

[0015]

[0016] Where A is the microseismic amplitude, ρ is the resistivity, μ is used to describe the coupling relationship between microseismic and electrical response, and is dimensionless. current is the current measured resistivity, reflecting the electrical characteristics of the current underground medium, ρ base is the background resistivity without water inrush risk.

[0017] Furthermore, in step (2), the objective function of the joint inversion model is as follows:

[0018]

[0019] Where α, β, and γ are weight coefficients used to balance the contribution of microseismic data, electrical data, and coupling constraints in the objective function, dimensionless, N is the number of observation points or events of microseismic data, dimensionless, and L(X 微震 ) is the forward operator of microseismic wave propagation, A i is the microseismic amplitude, M is the number of observation points of electrical data, dimensionless, F(X 电法 ) is the finite element forward model of resistivity field, ρ i is the resistivity of the electrical observation point, R(X 耦合 ) is a cross-coupling constraint term, which is used to describe the relationship between microseismic data and electrical data.

[0020] Furthermore, the cross-coupling constraint term R(X 耦合 ) is expressed as follows:

[0021]

[0022] Where N is the number of data points or model parameters, dimensionless, ω is the weight coefficient, dimensionless, f is the first function describing the relationship between microseismicity and electrical method, g is the second function describing the relationship between microseismicity and electrical method, and v is i is the microseismic velocity model parameter, ρ i is the electrical resistivity model parameter.

[0023] Furthermore, the forward operator L(X 微震 ) is obtained based on ray tracing or finite difference method.

[0024] Furthermore, in step (3), in the global search phase, an improved genetic algorithm is used to determine the initial values ​​of the inversion parameters. For the fitness function, the objective function value plus the model smoothness penalty term is used. For the crossover operator, the differential mutation probability is designed for the resistivity and microseismic parameters.

[0025] In the local convergence stage, the optimal solution is quickly approached based on the conjugate gradient method, and the Jacobian matrix calculation is accelerated using the adjoint state method;

[0026] For GPU parallel design, the finite element mesh is divided into subdomains, and each CUDA thread block processes the forward calculation of one subdomain.

[0027] Furthermore, in step (4), the inversion results are updated every several minutes, and an early warning is triggered when the following conditions are detected:

[0028]

[0029] Where, L crit is the critical crack length, v crit Critical velocity, H is the thickness of the coal seam.

[0030] Beneficial effects of the present invention:

[0031] 1. The present invention constructs a spatiotemporal collaborative constraint model through the dynamic coupling relationship between the energy of microseismic events and resistivity changes, breaking through the limitations of traditional single-field inversion. Based on multi-physical field collaborative constraints and dynamic feedback mechanisms, the early warning method has the characteristics of high precision and real-time performance, and can provide accurate and rapid early warning for coal mine roof water disasters.

[0032] 2. The present invention dynamically adjusts the electrical sampling frequency according to the microseismic energy, can balance the data density and the computational load, and enhances the sensitivity of water-bearing fracture identification by establishing a quantitative correlation between microseismic energy and resistivity gradient.

[0033] 3. Compared with traditional microseismic inversion and electrical inversion, the method provided by the present invention has greatly improved performance in terms of positioning error, resistivity resolution, inversion time and false alarm rate of water inrush warning. DETAILED DESCRIPTION

[0035] The embodiment of the present invention provides a method for joint inversion of microseismic-electrical coupled monitoring data in coal mines. The core principle is to construct a spatiotemporal collaborative constraint model through the dynamic coupling relationship between the energy of microseismic events and resistivity changes, thereby breaking through the limitations of traditional single-field inversion. The specific process is as follows.

[0036] 1. Data synchronization and preprocessing.

[0037] Hardware synchronization mechanism:

[0038] The microseismic event-triggered electrical data acquisition is adopted (non-uniform sampling, microseismic sensors and electrical electrodes are arranged in the boreholes of the underground working face, and the coal-rock fracture signal is collected when the coal-rock fractures, and the resistivity signal at the same time is collected). The expression is as follows:

[0039] T 电法 =T 微震 +Δt·log(E / E0)

[0040] Where, T 电法 T is the time parameter obtained by electrical measurement, in seconds. 微震 is the time parameter obtained from microseismic measurement, with the unit of s. Δt is the time difference, which represents the time difference between the electrical method and the microseismic measurement. E0 is the reference electric field strength, which is used for normalization or comparison. E is the current electric field strength, which represents the energy state during electrical measurement, with the unit of volt per meter (V / m).

[0041] In the data acquisition method provided in this embodiment, the electrical sampling frequency is dynamically adjusted according to the microseismic energy, so as to balance the data density and the computational load.

[0042] Data normalization:

[0043] The dimensions of microseismic signals (amplitude (A)) and resistivity (ρ) are unified:

[0044] The expression is as follows:

[0045]

[0046] Where A is the microseismic amplitude, in meters (m) or micrometers (μm). ρ is the resistivity, in ohm-meters (Ω·m). μ describes the coupling relationship between microseismicity and electrical response, and is dimensionless. ρ current The resistivity currently measured reflects the electrical characteristics of the current underground medium. The unit is ohm-meter (Ω m). base It is the background resistivity when there is no risk of water inrush, usually a benchmark value determined by historical data or theoretical models, in ohm-meter (Ω·m).

[0047] 2. Construction of joint inversion model.

[0048] This step is a joint inversion model based on dynamic weight cross constraints. The objective function expression of the joint inversion model is as follows:

[0049]

[0050] Where α, β, and γ are weight coefficients used to balance the contribution of microseismic data, electrical data, and coupling constraints in the objective function, and are dimensionless; N is the number of observation points or events of microseismic data, and is dimensionless; L(X 微震 ) is the forward operator of microseismic wave propagation, with the unit of m / N; A i is the microseismic amplitude of the ith event, in meters (m) or micrometers (μm); M is the number of observation points of electrical data, dimensionless; F(X 电法 ) is the finite element forward model of resistivity field, the unit is V / (Ω·m), ρ i R(X 耦合 ) is a cross-coupling constraint term, which is used to describe the relationship between microseismic data and electrical data and ensure that the joint inversion results are physically consistent. It is dimensionless.

[0051] As the cross-constraint term of design emphasis, R(X 耦合 ) is expressed as follows:

[0052]

[0053] Where N is the number of data points or model parameters, dimensionless, ω is the weight coefficient, dimensionless, f is the first function describing the relationship between microseismicity and electrical method, g is the second function describing the relationship between microseismicity and electrical method, and v is i is the microseismic velocity model parameter, unit: m / s, ρ i It is the electrical resistivity model parameter, unit is Ω·m.

[0054] Through the construction of the joint inversion model in this step, a quantitative correlation between microseismic energy and resistivity gradient is established, enhancing the sensitivity of water-bearing fracture identification.

[0055] 3. Hybrid optimization solver.

[0056] Layered Iteration Strategy:

[0057] 1) In the global search stage: an improved genetic algorithm (GA) is used to determine the initial values ​​of the inversion parameters. For the fitness function: the objective function value + model smoothness penalty term is used. For the crossover operator: the differentiated mutation probability is designed for the resistivity and microseismic parameters. The improved genetic algorithm (GA) is used to determine the initial values ​​of the inversion parameters. The genetic algorithm simulates natural selection and genetic mechanisms, and searches for the optimal solution through the continuous evolution of the population. In this method, the traditional genetic algorithm is improved to meet the needs of the joint inversion of the microseismic electrical method in coal mines. The objective function value plus the model smoothness penalty term is used as the fitness function. The objective function value reflects the degree of match between the inversion result and the actual observation data, and the model smoothness penalty term is used to avoid excessive fluctuations or unreasonable mutations in the inversion results.

[0058] 2) In the actual coal mine geological environment, the coal seam structure and geological structure have a certain degree of continuity and stability. If the microseismic velocity model or resistivity model in the inversion result changes too dramatically, it is likely to be inconsistent with the actual situation. After adding the model smoothness penalty term, it can guide the algorithm to tend to choose a smooth model that is more in line with geological laws in the process of finding the optimal solution, thereby improving the reliability and rationality of the inversion results. Differentiated mutation probabilities are designed for resistivity and microseismic parameters.

[0059] 3) In coal mine monitoring data, resistivity and microseismic parameters have different physical meanings and change characteristics. Resistivity mainly reflects the conductivity of underground media and is closely related to the water content of rocks; microseismic parameters are related to mechanical processes such as rock fracture and deformation. Therefore, in order to better mine the information of these two types of parameters and improve the search efficiency of the algorithm, the crossover operator is improved so that the resistivity and microseismic parameters have different mutation probabilities during the genetic operation process. For the resistivity parameter, since it is more sensitive to water-containing fractures, the mutation probability is appropriately reduced to maintain the search stability near the better solution; for the microseismic parameter, according to its change characteristics under different geological conditions, the mutation probability is reasonably adjusted to ensure that the algorithm can fully explore the microseismic parameter space and find a better solution.

[0060] 2) In the local convergence stage: Based on the conjugate gradient method (CG), the optimal solution is quickly approached, and the Jacobian matrix calculation is accelerated by the adjoint state method. The conjugate gradient method is an effective iterative optimization algorithm. It uses the gradient information of the current point to determine the search direction. It has the advantages of fast convergence speed and high computational efficiency when solving large-scale linear equations or optimization problems. In the joint inversion of coal mine microseismic and electrical methods, when the improved genetic algorithm finds a relatively good initial solution in the global search stage, the conjugate gradient method can be further optimized on this basis to quickly approach the global optimal solution or the local optimal solution. In the iterative process of the conjugate gradient method, the calculation of the Jacobian matrix is ​​a key step, and its calculation amount is large, which will affect the operation efficiency of the algorithm. In order to speed up the calculation process, the adjoint state method is used to calculate the Jacobian matrix. The adjoint state method transforms the sensitivity analysis of the original problem into the solution of an adjoint problem by introducing adjoint variables, which greatly reduces the calculation amount and storage amount of the Jacobian matrix. In the joint inversion of coal mine microseismic and electrical methods, the structure and parameters of the underground medium are complex, and the traditional method of calculating the Jacobian matrix may consume a lot of time and memory resources. The adjoint state method can significantly improve the calculation speed of the Jacobian matrix while ensuring the calculation accuracy, thereby accelerating the convergence speed of the conjugate gradient method and making the entire inversion process more efficient.

[0061] GPU parallel design:

[0062] The finite element grid is divided into subdomains, and each CUDA thread block processes the forward calculation of a subdomain, and the iteration time is reduced to less than 10% of the traditional CPU. GPU has powerful parallel computing capabilities and can process a large amount of data and computing tasks at the same time compared with traditional CPUs. In the joint inversion of microseismic and electrical methods in coal mines, finite element forward modeling is a huge computational link, involving the simulation of microseismic wave propagation and resistivity field distribution in underground media. By dividing the finite element grid into multiple subdomains and using CUDA (Compute Unified Device Architecture) technology, each thread block can process the forward calculation of a subdomain in parallel, which can give full play to the parallel computing advantages of GPU. In practical applications, a large finite element model is divided into hundreds or even thousands of subdomains, and each subdomain is independently calculated by a CUDA thread block, which can reduce the iteration time to less than 10% of the traditional CPU calculation. This not only greatly shortens the inversion calculation time and meets the needs of real-time monitoring and early warning, but also improves the overall performance and processing power of the system, so that the joint inversion analysis of microseismic and electrical methods can be carried out quickly and accurately under complex geological conditions.

[0063] 4. Dynamic feedback and early warning.

[0064] This embodiment adopts a real-time update mechanism:

[0065] The inversion results are updated every 5 minutes, and an alert is triggered when the following conditions are detected:

[0066]

[0067] Where, L crit is the critical crack length, in meters. Physical meaning: The critical crack length is the threshold of crack development. When the crack length exceeds this value, it may cause water inrush or other geological disasters. crit is the critical velocity, in m / s. The critical velocity is the threshold velocity of fluid flowing in the fracture. When the flow velocity exceeds this value, it may cause water inrush or further expansion of the fracture. H is the thickness of the coal seam, in m.

[0068] The applicant compared the inversion method of the present application with the traditional microseismic inversion and the traditional electrical inversion, as shown in Table 1 below. It can be seen from the content of Table 1 that the inversion method provided by the present application has greatly improved performance in terms of positioning error, resistivity resolution, single inversion time and false alarm rate of water inrush warning.

[0069] Table 1 Performance comparison of various inversion methods

[0070]

[0071] The following is a simulation implementation case using the inversion method provided in this application.

[0072] 1) Scene setting:

[0073] Coal seam thickness H = 5m, roof sandstone aquifer;

[0074] Microseismic event energy E = 1 × 10 4 J, resistivity background value ρ_base = 100Ω·m;

[0075] 2) Inversion results:

[0076] The length of the water-bearing fissure detected was L = 1.8 m (≥ L_crit = 1.5 m);

[0077] The crack extends toward the goaf at v = 0.15 m / min;

[0078] The system triggered a red alert 2.3 hours in advance, with an on-site verification accuracy of 92%.

Claims

1. A method for joint inversion of coal mine microseismic electrical coupling monitoring data, characterized in that: Here’s how: (1) Microseismic event-triggered electrical data acquisition is adopted, the electrical sampling frequency is dynamically adjusted according to the microseismic energy, and the data is normalized; (2) Construct a joint inversion model of microseismic and electrical methods to establish a quantitative correlation between microseismic energy and resistivity gradient; (3) Hybrid optimization solver, using a hierarchical iteration strategy and GPU parallelization design; (4) Dynamic feedback and early warning are adopted, and the real-time update mechanism is updated. When the triggering conditions are detected based on the inversion results, the early warning is triggered.

2. The method for joint inversion of coal mine microseismic electrical coupling monitoring data according to claim 1 is characterized in that: In step (1), the corresponding expression for microseismic event-triggered electrical data acquisition is as follows: T 电法 =T 微震 +Δt·log(E / E0) Where, T 电法 is the time parameter obtained by electrical measurement, T 微震 is the time parameter obtained from microseismic measurement, Δt is the time difference, which indicates the time difference between the electrical method and the microseismic measurement, E0 is the reference electric field strength, which is used for normalization or comparison, and E is the current electric field strength, which represents the energy state during electrical measurement.

3. The method for joint inversion of coal mine microseismic electrical coupling monitoring data according to claim 1, characterized in that: In step (1), the data is normalized as follows: Where A is the microseismic amplitude, ρ is the resistivity, μ is used to describe the coupling relationship between microseismic and electrical response, and is dimensionless. current is the current measured resistivity, reflecting the electrical characteristics of the current underground medium, ρ base is the background resistivity without water inrush risk.

4. The method for joint inversion of coal mine microseismic electrical coupling monitoring data according to claim 1, characterized in that: In step (2), the objective function of the joint inversion model is as follows: Where α, β, and γ are weight coefficients used to balance the contribution of microseismic data, electrical data, and coupling constraints in the objective function, dimensionless, N is the number of observation points or events of microseismic data, dimensionless, and L(X 微震 ) is the forward operator of microseismic wave propagation, A i is the microseismic amplitude, M is the number of observation points of electrical data, dimensionless, F(X 电法 ) is the finite element forward model of resistivity field, ρ i is the resistivity of the electrical observation point, R(X 耦合 ) is a cross-coupling constraint term, which is used to describe the relationship between microseismic data and electrical data.

5. The method for joint inversion of coal mine microseismic electrical coupling monitoring data according to claim 4, characterized in that: The cross-coupling constraint term R(X 耦合 ) is expressed as follows: Where N is the number of data points or model parameters, dimensionless, ω is the weight coefficient, dimensionless, f is the first function describing the relationship between microseismicity and electrical method, g is the second function describing the relationship between microseismicity and electrical method, and v is i is the microseismic velocity model parameter, ρ i is the electrical resistivity model parameter.

6. The method for joint inversion of coal mine microseismic electrical coupling monitoring data according to claim 5, characterized in that: Microseismic wave propagation forward operator L(X 微震 ) is obtained based on ray tracing or finite difference method.

7. The method for joint inversion of coal mine microseismic electrical coupling monitoring data according to claim 1, characterized in that: In step (3), in the global search phase, an improved genetic algorithm is used to determine the initial values ​​of the inversion parameters, the objective function value plus the model smoothness penalty term is used for the fitness function, and the crossover operator is used to design the differential mutation probability for the resistivity and microseismic parameters; In the local convergence stage, the optimal solution is quickly approached based on the conjugate gradient method, and the Jacobian matrix calculation is accelerated by the adjoint state method; For GPU parallel design, the finite element mesh is divided into subdomains, and each CUDA thread block processes the forward calculation of one subdomain.

8. The method for joint inversion of coal mine microseismic electrical coupling monitoring data according to claim 1, characterized in that: In step (4), the inversion results are updated every several minutes, and an alert is triggered when the following conditions are detected: Where, L crit is the critical crack length, v crit Critical velocity, H is the thickness of the coal seam.

Images