Optimized micro-seismic positioning method based on space search

Abstract

The invention discloses an optimized micro-seismic positioning method based on spatial search. The method comprises the following steps: S1, constructing a micro-seismic positioning model; s2, after spatial positioning solving is carried out on each microseismic event through the microseismic positioning model, optimization is carried out through a spatial search algorithm, and an optimal solution is obtained; and S3, introducing an inertia weight index and a learning factor to carry out multi-objective optimization. The invention provides an optimized micro-seismic positioning algorithm based on a space search algorithm, which combines the space search algorithm with a traditional positioning model, optimizes various key parameters in a micro-seismic positioning process, adopts a multi-target optimization strategy, comprehensively improves the positioning precision and the system efficiency, and overcomes the defects in a traditional method.

Description

Technical Field

[0001] The present invention relates to an optimized microseismic positioning method based on spatial search, belonging to the technical field of coal mine safety. Background Art

[0002] Currently, microseismic positioning technology is a technology for estimating the source location by analyzing the propagation process of seismic wave signals. Microseismic positioning is widely used in fields such as earthquake monitoring, mine safety, and oil and gas exploration, and can help detect and locate minor seismic events in a timely manner, warning of potential earthquake disasters or other safety hazards. With the development of microseismic monitoring technology, higher requirements are put forward for positioning accuracy and real-time performance. Especially in complex geological environments, how to overcome problems such as noise, non-linear propagation media, and uneven sensor layout has become the key to improving microseismic positioning performance.

[0003] Existing microseismic positioning technologies include methods based on the least squares method (LS), maximum likelihood estimation method (MLE), backpropagation neural network (BP), etc. These traditional methods usually have the following problems:

[0004] Insufficient positioning accuracy: In complex geological structures or multi-path propagation environments, the positioning accuracy is difficult to meet high-precision requirements.

[0005] Low computational efficiency: Since the positioning process involves a large amount of non-linear calculations, traditional methods often have a large amount of calculations and low efficiency, and it is difficult to achieve real-time positioning.

[0006] Sensitive to noise: Traditional algorithms have poor robustness to measurement errors and noise, and are easily affected by environmental changes or low data quality, resulting in inaccurate positioning results. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide an optimized microseismic positioning algorithm based on a spatial search algorithm. By combining the spatial search algorithm with a traditional positioning model, various key parameters in the microseismic positioning process are optimized, and a multi-objective optimization strategy is adopted to comprehensively improve the positioning accuracy and system efficiency, solving the deficiencies in traditional methods.

[0008] To solve the above technical problem, the technical solution of the present invention is:

[0009] An optimized microseismic positioning method based on spatial search, which includes the following steps:

[0010] Step S1, construct a microseismic positioning model;

[0011] Step S2, after performing spatial positioning solution on each microseismic event through the microseismic positioning model, use the spatial search algorithm for optimization to obtain the optimal solution;

[0012] Step S3: Introduce the inertia weight index and learning factor for multi-objective optimization.

[0013] Further, in the step S1, a microseismic positioning model is constructed, which specifically includes the following steps:

[0014] Let the source location coordinates be (X O , Y O , Z O ), the occurrence time be T O , and the location coordinates of each sensor be (X i , Y i , Z i ), and the time when the vibration waveform appears be T i . Assuming that the seismic wave propagation speed is uniform at V, the first equation is obtained, and the first equation is as follows:

[0015] (X i - X O ) 2 + (Y i - Y O ) 2 + (Z i - Z O ) 2 = V 2 (T i- - T O ) 2 ;

[0016] The first equation is reduced to a power. Substitute i = 1 and i = 2 into the first equation and combine to remove the power to obtain a four-variable linear equation, and the four-variable linear equation is as follows:

[0017] 2(X1 - X2)X O + 2(Y1 - Y2)Y O + 2(Z1 - Z2)Z O + 2V 2 (T2 - T1)T O = X1 2 - X2 2 + Y1 2 - Y2 2 + Z1 2 - Z2 2 + V 2 (T2 2 - T1 2 );

[0018] There are n monitoring points, (n ≥ 4); then m four-variable linear equations composed of the four-variable linear equations can be obtained, m = n(n - 1) / 2; the four-variable linear equations are as follows:

[0019] A1XO +B1Y O +C1Z O +D1T O =E1

[0020] A2X O +B2Y O +C2Z O +D2T O =E2

[0021] A3X O +B3Y O +C3Z O +D3T O =E3

[0022] A4X O +B4Y O +C4Z O +D4T O =E4

[0023] …

[0024] A m X O +B m Y O +C m Z O +D m T O =E m ;

[0025] According to the system of four linear equations, a linear matrix of m * 5 order is obtained. The linear matrix is as follows:

[0026]

[0027] Use the Gauss-Jordan elimination method to find the solution of the system of four linear equations. After obtaining several groups of solutions, each solution is verified to obtain the optimal solution, and the objective function method is used for verification. Substitute the solution to be verified into the first equation to obtain the verification formula. The verification formula is as follows:

[0028]

[0029] Among them, x0, y0, z0 are the source coordinates;

[0030] t0 is the source vibration time;

[0031] x i ,y i ,z i are the coordinates of the i-th measurement point,

[0032] n is the number of sensors;

[0033] t i is the arrival time of the wave at the i-th station;

[0034] v(x0, y0, z0) is the wave propagation velocity in the coal-rock medium;

[0035] p is the standard exponent used for description;

[0036] The objective function method is used to verify the objective function value of each solution one by one, and the approximation method is adopted to finally obtain the solution with the minimum objective function value as the optimal solution;

[0037] After obtaining the optimal solution, the objective function is compared for each effective monitoring point through the comparison formula, so that the calculation error of a single channel is within the expected accuracy range.

[0038] Furthermore, in step S2, after solving the spatial location of each microseismic event through the microseismic location model, it is optimized using the spatial search algorithm to obtain the optimal solution, which specifically includes the following steps;

[0039] Step S21, initialize the parameter group;

[0040] Step S22, randomly initialize the position and velocity of each parameter;

[0041] Step S23, determine whether the end condition of the spatial search algorithm is satisfied. If the end condition is satisfied, output the optimal solution; if the end condition is not satisfied, enter step S24;

[0042] Step S24, update the velocity and position of each parameter;

[0043] Step S25, calculate the fitness value of each parameter;

[0044] Step S26, update the individual historical optimal fitness value and position of each parameter;

[0045] Step S27, update the historical optimal fitness value and position of the parameter group;

[0046] Step S28, update the inertia, weight, and iteration times, and then enter step S23.

[0047] Furthermore, in step S24, the velocity update formula of the parameter is as follows:

[0048]

[0049] where N is the parameter group size;

[0050] i is the parameter serial number, i = 1, 2,..., N;

[0051] D is the parameter dimension;

[0052] d is the parameter dimension serial number, d = 1, 2,..., D;

[0053] k is the number of iterations;

[0054] w is the inertia weight;

[0055] c1 is the individual learning factor;

[0056] c2 is the swarm learning factor;

[0057] r1, r2 are random numbers within the interval [0, 1], which increases the randomness of the search;

[0058] is the velocity vector of parameter i in the d-th dimension at the k-th iteration;

[0059] is the position vector of parameter i in the d-th dimension at the k-th iteration;

[0060] is the historical optimal position of particle i in the d-th dimension at the k-th iteration, that is, after the k-th iteration, the optimal solution obtained by the i-th particle search;

[0061] is the historical optimal position of the swarm in the d-th dimension at the k-th iteration, that is, after the k-th iteration, the optimal solution in the entire particle swarm;

[0062] Obtain the moving direction of the parameter for the next iteration according to the velocity update formula, and the moving direction of the parameter for the next iteration = inertia direction + individual optimal direction + swarm optimal direction.

[0063] Furthermore, in step S24, the position update formula of the parameter is as follows:

[0064]

[0065] Furthermore, in step S3, the formula of the weight index is as follows:

[0066]

[0067] where w is the weight index;

[0068] w max is the maximum inertia weight;

[0069] w min is the minimum inertia weight;

[0070] iter is the current number of iterations;

[0071] iter maxis the maximum number of iterations.

[0072] Further, in the step S3, the learning factors include a first learning factor c1 and a second learning factor c2;

[0073] The first learning factor c1 represents the weight of the part where the next action of the particle comes from its own experience, and it is the acceleration weight that pushes the particle towards the individual optimal position ;

[0074] The second learning factor c2 represents the weight of the part where the next action of the particle comes from the experience of other particles, and it is the acceleration weight that pushes the particle towards the global optimal position ;

[0075] By adopting the above technical solution, the present invention combines the global search ability of the space search algorithm with the traditional microseismic location method, and proposes an efficient optimization solution for the multi-variable and high-complexity problems in microseismic location. By using the space search algorithm to optimize the key parameters in the microseismic location model, including the source location, epicenter depth, and distance between the source and the sensor, etc., the accuracy and robustness of microseismic location are improved. In addition, the present invention introduces a multi-objective optimization strategy, taking into account multiple aspects such as accuracy, calculation time, and resource consumption, and further optimizes the location process. The optimized microseismic location algorithm of the present invention can be successfully integrated into the microseismic monitoring system, and can provide accurate microseismic location results in a real-time monitoring environment, and can be widely applied to fields such as microseismic disaster warning, mine safety monitoring, oil and gas coal exploration, etc. It realizes the rapid location of microseismic events and has high practical value and promotion prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 is the flowchart of the optimized microseismic location method based on space search of the present invention;

[0077] Figure 2 is the schematic diagram of microseismic monitoring and location of the present invention;

[0078] Figure 3 is the flowchart of step S2 of the present invention;

[0079] Figure 4 is the schematic diagram of the moving direction of the parameters in the next iteration of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0080] In order to make the content of the present invention easier to be clearly understood, the present invention will be further described in detail below according to specific embodiments and in conjunction with the accompanying drawings.

[0081] As Figure 1 shown, this embodiment provides an optimized microseismic location method based on space search, which includes the following steps:

[0082] Step S1: Construct a microseismic location model, specifically as follows:

[0083] In this embodiment, the longitudinal wave (P-wave) first arrival time method is used as the calculation model. This model has the following advantages: 1. The first appearance of the P-wave is easy to confirm; 2. Compared with other waves, the determination error of the first arrival time of the P-wave is smaller; 3. The first arrival time can be automatically determined by a computer program.

[0084] As Figure 2 shown, let the source position coordinates be (X O , Y O , Z O ), the occurrence time be T O , the position coordinates of each sensor be (X i , Y i , Z i ), and the time when the vibration waveform appears be T i . Assuming that the seismic wave propagation speed is uniform at V, the first equation can be obtained as follows:

[0085] (X i - X O ) 2 + (Y i - Y O ) 2 + (Z i - Z O ) 2 = V 2 (T i- - T O ) 2 ;

[0086] Since what we want to solve are X O , Y O , Z O , T O , the first equation needs to be reduced to a lower power. Substitute i = 1 and i = 2 into the first equation and combine to eliminate the power to obtain a four-variable linear equation as follows:

[0087] 2(X1 - X2)X O + 2(Y1 - Y2)Y O + 2(Z1 - Z2)Z O + 2V 2 (T2 - T1)T O = X1 2 - X2 2 + Y1 2 - Y2 2 + Z1 2 - Z2 2 + V 2 (T22 -T1 2 );

[0088] In this quartic equation, except for X O , Y O , Z O , and T O , the rest are known parameters, which is a standard quartic equation. From this, it can be deduced that: Suppose there are n monitoring points, (n≥4); then m quartic equations can be obtained to form a system of quartic equations, where m = n(n - 1) / 2; the system of quartic equations is as follows:

[0089] A1X O +B1Y O +C1Z O +D1T O =E1

[0090] A2X O +B2Y O +C2Z O +D2T O =E2

[0091] A3X O +B3Y O +C3Z O +D3T O =E3

[0092] A4X O +B4Y O +C4Z O +D4T O =E4

[0093] …

[0094] A m X O +B m Y O +C m Z O +D m T O =E m ;

[0095] According to the system of quartic equations, a linear matrix of m * 5 order can be obtained. The linear matrix is as follows:

[0096]

[0097]

[0098] Since there are only 4 unknowns corresponding to this system of four linear equations and the values on the diagonal of its coefficient matrix are non-zero, when using a computer for matrix calculation and solution, the Gauss-Jordan elimination method is generally used to find the solution of the system of four linear equations. After obtaining several sets of solutions, each solution needs to be verified to obtain the optimal solution, and the objective function method is used for verification. Specifically, substitute the solution to be verified into the first equation to obtain the verification formula, and the verification formula is as follows:

[0099]

[0100] where x0, y0, z0 are the source coordinates;

[0101] t0 is the source vibration time;

[0102] x i , y i , z i are the coordinates of the i-th measurement point,

[0103] n is the number of sensors;

[0104] t i is the time when the wave arrives at the i-th station;

[0105] v(x0, y0, z0) is the wave propagation speed in the coal and rock medium;

[0106] p is the standard exponent used for description;

[0107] In particular: when p = 2, we use the L2 standard for positioning, that is, assume Gaussian error distribution according to the wave first arrival time; when p = 1, we use the L1 standard for positioning. Due to large timing errors in some monitoring points, in this case, the L1 standard is superior to the L2 standard, but this requires a much larger amount of data on the wave arrival time because this standard will produce many local extrema. Therefore, the L2 standard is usually used for positioning.

[0108] In this embodiment, the objective function method is used to verify the objective function value of each solution one by one, and the approximation method is adopted to finally obtain the solution with the smallest objective function value as the optimal solution;

[0109] After obtaining the optimal solution, it is also necessary to compare the objective function for each effective monitoring point through the comparison formula to make the calculation error of a single channel within the expected accuracy range (generally ≤ 20 ms). The comparison formula is as follows:

[0110]

[0111] Step S2. After spatially locating and solving each microseismic event using the microseismic location model, optimize it using the spatial search algorithm to obtain the optimal solution. The idea of the spatial search algorithm stems from the study of animal foraging behavior, enabling the group to find the optimal destination through collective information sharing. Applied to microseismic location, after the microseismic location model spatially locates and solves each microseismic event, each individual locates along the direction it determines and records the position with the minimum error during the location process. At the same time, all individuals share their each location result and location accuracy. By analyzing the location results of each individual each time, continuously adjust the location direction, and finally obtain the location result with the minimum error, that is, the optimal solution. As Figure 3 shown, specifically:

[0112] Step S21. Initialize the parameter group;

[0113] Step S22. Randomly initialize the position and velocity of each parameter;

[0114] Step S23. Determine whether the end condition of the spatial search algorithm is satisfied. If the end condition is satisfied, output the optimal solution; if the end condition is not satisfied, go to Step S24;

[0115] Step S24. Update the velocity and position of each parameter;

[0116] Step S25. Calculate the fitness value of each parameter;

[0117] Step S26. Update the individual historical optimal fitness value and position of each parameter;

[0118] Step S27. Update the historical optimal fitness value and position of the parameter group;

[0119] Step S28. Update the inertia, weight, and number of iterations, and then go to Step S23.

[0120] In Step S21, the parameter settings include the parameter group size, parameter dimension, number of iterations, inertia weight, learning factor, iteration step range; the initialized parameters include the individual historical optimal solution, group historical optimal solution, individual historical optimal fitness value, group historical optimal fitness value; the end condition is: reaching the maximum number of iterations or satisfying the solution condition.

[0121] In Step S24, the core elements of the algorithm are velocity and position. Velocity represents the direction and distance of the particle's movement in the next iteration, and position is a solution to the problem to be solved.

[0122] The velocity update formula of the parameter is as follows:

[0123]

[0124] where N is the parameter group size;

[0125] i is the parameter serial number, i = 1, 2, ..., N;

[0126] D is the parameter dimension;

[0127] d is the parameter dimension serial number, d = 1, 2, ..., D;

[0128] k is the number of iterations;

[0129] w is the inertia weight;

[0130] c1 is the individual learning factor;

[0131] c2 is the swarm learning factor;

[0132] r1, r2 are random numbers within the interval [0, 1], which increases the randomness of the search;

[0133] is the velocity vector of parameter i in the d-th dimension at the k-th iteration;

[0134] is the position vector of parameter i in the d-th dimension at the k-th iteration;

[0135] is the historical optimal position of particle i in the d-th dimension at the k-th iteration, that is, after the k-th iteration, the optimal solution obtained by the search of the i-th particle (individual);

[0136] is the historical optimal position of the swarm in the d-th dimension at the k-th iteration, that is, after the k-th iteration, the optimal solution in the entire particle swarm;

[0137] Such as Figure 4 As shown, according to the velocity update formula, the moving direction of the parameter for the next iteration can be obtained. The moving direction of the parameter for the next iteration = inertia direction + individual optimal direction + swarm optimal direction.

[0138] The position update formula of the parameter is as follows:

[0139]

[0140] Step S3, introduce the inertia weight index and learning factors for multi-objective optimization;

[0141] Aiming at the problems faced by microseismic event location, such as location accuracy, calculation speed, noise interference, etc., introduce the inertia weight index w and learning factors c1, c2 to optimize the objective solution, and achieve the optimal processing method that balances the high and low location accuracy and the fast and slow calculation speed.

[0142] When solving the actual optimization problem of this embodiment, it is often desirable to first use global search to quickly converge the search space to a certain region, and then use local fine search to obtain a high-precision solution. Therefore, an adaptive adjustment strategy is proposed, that is, as the iteration progresses, the value of the inertia weight ω is linearly decreased. The linear change strategy adopted in this embodiment is as follows: as the number of iterations increases, the inertia weight ω continuously decreases, so that the space search algorithm has strong global convergence ability in the initial stage and strong local convergence ability in the later stage.

[0143] The formula for the weight index is as follows:

[0144]

[0145] where w is the weight index;

[0146] w max is the maximum inertia weight;

[0147] w min is the minimum inertia weight;

[0148] iter is the current number of iterations;

[0149] iter max is the maximum number of iterations.

[0150] The learning factors include the first learning factor c1 and the second learning factor c2;

[0151] The first learning factor c1 represents the weight of the part where the next action of the particle comes from its own experience, and is the acceleration weight that pushes the particle to the individual optimal position ;

[0152] The second learning factor c2 represents the weight of the part where the next action of the particle comes from the experience of other particles, and is the acceleration weight that pushes the particle to the global optimal position ;

[0153] In this embodiment, the algorithm model is optimized through the learning samples of more than 1000 microseismic events in the coal mine underground. Finally, the value ranges of the weight index w and the learning factors c1 and c2 are determined. Compared with the original microseismic event location model that does not use the space search optimization algorithm, the location accuracy is improved, the calculation speed is optimized, and the events that could not be located due to other factors such as noise before are located, realizing the all-round optimization of the microseismic location event algorithm.

[0154] The specific embodiments described above further elaborate on the technical problems solved by the present invention, the technical solutions, and the beneficial effects. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. An optimized microseismic positioning method based on spatial search, characterized in that, It includes the following steps: Step S1, construct a microseismic location model; Step S2, after solving the spatial location of each microseismic event through the microseismic location model, optimize it using a spatial search algorithm to obtain the optimal solution; Step S3, introduce an inertia weight index and a learning factor for multi-objective optimization.

2. The optimized microseismic positioning method based on spatial search according to claim 1, wherein In the said step S1, constructing a microseismic location model specifically includes the following steps: Suppose the source location coordinates are (X O , Y O , Z O ), the occurrence time is T O , and the location coordinates of each sensor are (X i , Y i , Z i ), the time when the vibration waveform appears is T i . Assuming that the seismic wave propagation speed is uniform at V, the first equation is obtained, and the first equation is as follows: (X i -X O ) 2 +(Y i -Y O ) 2 +(Z i -Z O ) 2 =V 2 (T i -T O ) 2 ; Reduce the power of the first equation, substitute i = 1 and i = 2 into the first equation and combine to eliminate the power to obtain a four-variable linear equation, and the four-variable linear equation is as follows: 2(X1 - X2)X O +2(Y1 - Y2)Y O +2(Z1 - Z2)Z O +2V 2 (T2 - T1)T O =X1 2 - X2 2 +Y1 2 - Y2 2 +Z1 2 - Z2 2 +V 2 (T2 2 - T1 2 ); Suppose there are n monitoring points, (n≥4); then m four-variable linear equations composed of the four-variable linear equation can be obtained, m = n(n - 1) / 2; the four-variable linear equation system is as follows: A1X O +B1Y O +C1Z O +D1T O =E1 A2X O +B2Y O +C2Z O +D2T O =E2 A3X O +B3Y O +C3Z O +D3T O =E3 A4X O + B4Y O + C4Z O + D4T O = E4 … A m X O +B m Y O +C m Z O +D m T O =E m ; According to the four-variable linear equation system, obtain an m*5 order linear matrix, and the linear matrix is as follows: Use the Gauss-Jordan elimination method to solve the four-variable linear equation system. After obtaining several groups of solutions, verify each solution to obtain the optimal solution, and use the objective function method for verification. Substitute the solution to be verified into the first equation to obtain the verification formula, and the verification formula is as follows: Among them, x0, y0, z0 are the source coordinates; t0 is the source vibration time; x i , y i , z i are the coordinates of the i-th measurement point. n is the number of sensors; t i is the arrival time of the wave at the i-th station; v(x0, y0, z0) is the wave propagation speed in the coal and rock medium; p is the standard exponent used for description; Use the objective function method to verify the objective function value of each solution one by one, and adopt the approximation method to finally obtain the solution with the smallest objective function value as the optimal solution; After obtaining the optimal solution, compare the objective function corresponding to each effective monitoring point through the comparison formula to make the calculation error of a single channel within the expected accuracy range.

3. The optimized microseismic positioning method based on spatial search according to claim 1, characterized in that In the said step S2, after solving the spatial location of each microseismic event through the microseismic location model, optimize it using a spatial search algorithm to obtain the optimal solution, specifically including the following steps; Step S21, initialize the parameter group; Step S22, randomly initialize the position and velocity of each parameter; Step S23, judge whether the end condition of the spatial search algorithm is satisfied. If the end condition is satisfied, output the optimal solution; if the end condition is not satisfied, enter step S24; Step S24, update the velocity and position of each parameter; Step S25, calculate the fitness value of each parameter; Step S26, update the individual historical optimal fitness value and position of each parameter; Step S27, update the historical optimal fitness value and position of the parameter group; Step S28, update the inertia, weight and iteration times, and then enter step S23.

4. The optimized microseismic positioning method based on spatial search according to claim 3, characterized in that In the said step S24, the velocity update formula of the parameter is as follows: Among them, N is the parameter group size; i is the parameter serial number, i = 1, 2,..., N; D is the parameter dimension; d is the parameter dimension serial number d = 1, 2,..., D; k is the iteration number; w is the inertia weight; c1 is the individual learning factor; c2 is the group learning factor; r1, r2 are random numbers within the interval [0, 1] to increase the randomness of the search; is the velocity vector of parameter i in the d-th dimension at the k-th iteration; is the position vector of parameter i in the d-th dimension at the k-th iteration; It is the historical optimal position of particle i in the d-th dimension during the k-th iteration, that is, after the k-th iteration, the optimal solution obtained by the i-th particle search; It is the historical optimal position of the group in the d-th dimension at the k-th iteration, that is, the optimal solution in the entire particle swarm after the k-th iteration; Obtain the moving direction of the parameter for the next iteration according to the speed update formula, where the moving direction of the parameter for the next iteration = inertial direction + individual optimal direction + swarm optimal direction.

5. The optimized microseismic positioning method based on spatial search according to claim 3, wherein In the step S24, the position update formula of the parameter is as follows:

6. The optimized microseismic positioning method based on spatial search according to claim 1, characterized in that In the step S3, the formula of the weight index is as follows: where w is the weight index; w max is the maximum inertia weight; w min is the minimum inertia weight; iter is the current iteration number; iter max is the maximum number of iterations.

7. The optimized microseismic positioning method based on spatial search according to claim 1, characterized in that: In the step S3, the learning factors include a first learning factor c1 and a second learning factor c2; The first learning factor c1 represents the weight of the part where the next action of the particle comes from its own experience, and pushes the particle towards the individual optimal position Acceleration weight; The next action of the second learning factor c2 particle comes from the weight of the experience part of other particles, which pushes the particle towards the global optimal position Acceleration weight.

Images