A dynamic prediction method for surface subsidence in coal mining based on dual-parameter time function

Abstract

The present invention discloses a method for dynamic prediction of surface settlement in coal mining based on a dual-parameter time function, and belongs to the field of dynamic and accurate prediction of coal mining settlement. First, the mining working face information and parameter data are determined through the coal mining engineering plan and mining data, and the working face is divided into n units; the settlement influence values ​​of the n units on the surface settlement prediction point are calculated respectively by using the probability integral method formula; a dual-parameter time function model and an objective function model of the dual parameters to be solved are constructed, and the time of mining of the n units relative to the start of mining is used to calculate the time influence value of the n units on the surface settlement prediction point; finally, the settlement value of the surface settlement prediction point at time t is calculated by summing the product of the settlement influence value of the n units and the time influence value. The function model of this method is refined, and the hybrid optimization algorithm can be used to obtain the dual parameters of the time function with high accuracy, and dynamic and accurate prediction of mining settlement can be realized.

Description

Technical Field

[0001] The invention relates to a method for dynamic prediction of surface subsidence in coal mining based on a dual-parameter time function, and belongs to the field of dynamic and accurate prediction of subsidence in coal mining. Technical Background

[0002] The deformation of rock strata and ground surface caused by underground coal mining is a complex spatiotemporal problem. If the design is based only on the deformation value after the surface settlement of coal mining, it will often cause damage to ground buildings (structures). Therefore, when designing "three-down" coal mining, it is necessary to conduct a dynamic prediction of the surface settlement of coal mining to ensure the safety of the protected objects. Especially with the protection requirements of large-scale infrastructure and the need to implement mining and reconstruction in settlement areas, how to accurately predict the dynamic settlement of coal mining surface is particularly important.

[0003] At present, the parameters of the time function are often obtained by inversion based on measured data. This requires the inversion algorithm to be able to fully fit the function model curve with the measured curve while ensuring that the fitting error function value is minimized. However, there is little research on how to improve the accuracy of the parameter inversion algorithm, which leads to low prediction accuracy and underfitting problems when using the time function to perform dynamic prediction of coal mining settlement. At the same time, when the algorithm is fitted with the root mean square error as the objective function to solve the parameters, the settlement curve corresponding to the single-parameter time function model has a single shape, and the curve is often unable to be fully fitted; the three-parameter and multi-parameter time function models contain more variables, which makes the model formula deduction and parameter inversion more difficult and the model complexity increases. The time function model based on dual parameters is more refined, with moderate model complexity and good curve shape.

[0004] Currently, the commonly used algorithms for inverting time function parameters based on measured settlement data include: genetic algorithm, particle swarm algorithm, fruit fly algorithm, simulated annealing algorithm, spatial surface fitting parameterization and software fitting, etc. In experimental research, it is found that single optimization algorithm often has problems such as local optimum, sensitivity to algorithm initial value and parameter selection, slow convergence speed, etc., while in the case of obvious nonlinearity of objective function in dynamic prediction of surface settlement, hybrid optimization algorithm has the ability to overcome local extreme value and obtain global extreme value, and can solve the problems of traditional and relatively mature algorithms. Combining the relatively mature particle swarm algorithm that arbitrarily falls into local extreme value with the artificial fish swarm algorithm that has not been widely used in dynamic prediction of surface settlement can achieve the optimal value solution of the parameters of dual-parameter time function.

[0005] The working face is divided into sections by using the effective size segmentation method, and the complex Simpson's formula is introduced to perform numerical integration on the definite integral of the prediction model. Then, a dynamic prediction model for surface subsidence of the working face in a horizontal or gently inclined coal seam is constructed based on the superposition principle in combination with the time function to achieve dynamic and accurate prediction of mining subsidence. Summary of the invention

[0006] In view of the shortcomings of the above-mentioned technologies, a method for dynamic prediction of surface subsidence in coal mining based on dual-parameter time function is provided. The function model of this method is refined, and a hybrid optimization algorithm is used to obtain the dual parameters of the time function with high precision, which can realize accurate prediction of mining subsidence dynamics.

[0007] In order to achieve the above technical objectives, a method for dynamic prediction of surface subsidence in coal mining based on a dual-parameter time function is provided in the present invention, and the specific steps are as follows:

[0008] Step 1: First, determine the plane coordinates of the four turning points of the rectangular working face and the mining depth, coal seam mining thickness, coal seam inclination, and coal seam strike azimuth through the coal mining engineering plan and mining data, determine the mining advancement speed v of the working face strike at any dynamic prediction time t and time t, in meters per day, and simulate the inversion of the probability integral method parameters of the region according to the actual mining subsidence and extreme value of the movement deformation value of the coal mine, including the subsidence coefficient q, the horizontal movement coefficient b, the main influence angle tangent tanβ, the mining influence propagation angle θ, the downhill turning point offset S1, the uphill turning point offset S2, the strike left turning point offset S3, the strike right turning point offset S4, and the coal seam working face strike and dip mining dimensions D3 and D1;

[0009] Step 2: Determine the direction dimension of the calculated working surface as l = D3-S3-S4, and the inclination dimension is The working face is mined forward along the strike direction. The strike length of the working face at time t is v*t, and the dip width remains unchanged. The calculated working face rectangle is evenly divided into n rectangular subdivision units along the strike direction of the calculated working face with a width of 1 meter and a dip dimension L as the length.

[0010] Step 3: Assume that the surface point whose settlement state needs to be predicted is the surface settlement prediction point, and the coordinates of the surface settlement prediction point are (x, y). Use the probability integral method formula to calculate the settlement influence values ​​of n rectangular subdivision units on the surface settlement prediction point respectively;

[0011] Step 4: Construct a dual-parameter time function model and an objective function model of the dual parameters to be solved;

[0012] Step 5: The field of view and step size of the artificial fish swarm algorithm are changed to an adaptive field of view and step size to obtain an adaptive artificial fish swarm algorithm, and the adaptive artificial fish swarm algorithm is combined with the particle swarm algorithm to obtain an adaptive hybrid optimization artificial fish swarm algorithm, and the adaptive hybrid optimization artificial fish swarm algorithm is used to solve the time function dual parameters of the dual-parameter time function model;

[0013] Step 6: Calculate the time of mining of n rectangular subdivision units relative to the start of mining, and bring it into the dual-parameter time function model with solved dual parameters to calculate the time influence value of n rectangular subdivision units on the surface settlement prediction point;

[0014] Step 7: Multiply the settlement influence values ​​and time influence values ​​of the n rectangular subdivision units respectively and then sum them up to obtain the settlement value of the surface settlement prediction point at time t.

[0015] Furthermore, the steps of solving the dual parameters of the time function using the adaptive hybrid optimization artificial fish swarm algorithm are as follows:

[0016] Step 1: Define the time function double parameter as an artificial fish, initialize the parameters of the artificial fish swarm algorithm and the particle swarm algorithm, including the time function double parameter quantity N, the maximum number of iterations Maxgen, the maximum number of trials try_number, the crowding factor delta, the visual step coefficient, the initial step length, the initial field of view; the number of iterations, the population size, the learning factors c1 and c2, the maximum flight speed, the maximum and minimum inertia weights, and the upper and lower limits of the time function double parameter; initialize the time function double parameter value: generate N groups of time function double parameter values ​​through random numbers and the upper and lower limits of the time function double parameter;

[0017] Step 2: Select a surface settlement prediction point, calculate the objective function value corresponding to each set of time function double parameters according to the measured settlement value of the point, and record it on the bulletin board, calculate the distance between the current time function double parameter and the optimal time function double parameter in the bulletin board as the field of view of the current time function double parameter, and calculate the step length through the visual step coefficient;

[0018] Step 3: Each group of time function double parameters performs foraging, flocking, chasing and random behaviors respectively, and selects the time function double parameter with the best objective function value as the current optimal time function double parameter;

[0019] Step 4: Within the maximum number of iterations, after each time function double parameter executes the above behavior, compare the objective function value with the billboard. If the billboard update condition is met, update the billboard and the current optimal time function double parameter;

[0020] Step 5: Determine whether the maximum number of iterations has been reached. When the iteration is completed, the current optimal time function double parameter is used as the initial particle value of the particle swarm algorithm;

[0021] Step 6: Execute the particle swarm algorithm to further optimize the particle optimal solution under the dual parameter state of the current optimal time function;

[0022] Step 7: Compare the optimal dual parameter values ​​of the time function of the adaptive artificial fish swarm and the optimal solution of the particles, and take the better value of the two as the global optimal result and output it as the dual parameter of the time function of the dual parameter time function model.

[0023] Furthermore, the probability integral method is used to calculate the settlement influence values ​​of n rectangular subdivision units on the surface settlement prediction point at time t. The settlement value of the surface settlement prediction point (x, y) is calculated using the following formula:

[0024]

[0025] In the formula, η, ξ are the integral variables of the double integral, η corresponds to the differential value of x, and ξ corresponds to the differential value of y; D is the calculation boundary of the coal seam; r is the main influence radius, which is obtained by the formula: r = H / tanβ; H is the mining depth, tanβ is the main influence angle tangent; W cm The maximum surface settlement under full mining conditions is calculated by the formula: W cm =mq·cosα; q is the sedimentation coefficient; m is the normal mining thickness of the coal seam; α is the inclination angle of the coal seam;

[0026] For the i-th (i∈[1,n]) rectangular subdivision unit in the n rectangular working surfaces, equation (1) is further calculated to obtain:

[0027]

[0028] In the formula, The minimum and maximum values ​​of the coordinates of the rectangular subdivision unit; The minimum and maximum values ​​of the inclination coordinates of the rectangular subdivision unit;

[0029] The complex Simpson formula (3) is used to numerically integrate the definite integral in formula (2):

[0030]

[0031] The settlement influence value expression (4) of the rectangular subdivision unit i on any surface settlement prediction point is obtained:

[0032]

[0033] The superposition principle is used to calculate the settlement impact value of mining in each rectangular subdivision unit on any surface settlement prediction point, that is, the static settlement value of the surface settlement prediction point after mining in the prediction area is completed is formula (5):

[0034]

[0035] In the formula: In the formula: is the formula (2) Using the numerical integration value of formula (3), is the formula (2) The numerical integration value of formula (3) is adopted.

[0036] Furthermore, the method for calculating the settlement value of the surface settlement prediction point at time t by using the settlement influence value and time influence value is as follows:

[0037] Using v i(i=1,2,…,n) represents the advancement speed of the working face when mining to the i-th subdivision unit along the mining advancement direction, in meters per day;

[0038] Using t i (i=1,2,…,n) is the number of days that the working face takes to mine from the i-1th subdivision unit to the i-th subdivision unit along the mining advancement direction, in days;

[0039] Then the mining dimension of each mining subdivision unit is v i t i , in meters;

[0040] The last rectangular unit is mined with a strike dimension of v n t n =lv n-1 t n-1 , l is the mining dimension of the working face calculated at time t;

[0041] Remember i is the static settlement impact value of the surface settlement prediction point (x, y) when the mining of the i-th mining subdivision unit is completed. When the working face is mined to the i-th subdivision unit, that is, t = t1 + t2 + ... + t i The time when the working surface of the 1st to ith subdivision unit produces settlement influence on the surface settlement prediction point is t-t1, t-t1-t2, …, t-t1-t2-…-t i-1 , the settlement influence value of each subdivision unit working surface on the surface settlement prediction point is assigned a time function coefficient and superimposed to obtain the surface settlement value of the surface settlement prediction point at time t. The surface settlement dynamic prediction model of the surface settlement prediction point is formula (6):

[0042]

[0043] In the formula, w i (x, y) is calculated according to formula (5), (x, y) is the coordinate of the surface settlement prediction point, is t-t1-t2-…-t i-1 The time function value at .

[0044] Beneficial effects: This method introduces an adaptive step size and field of view model and combines the global optimization of artificial fish swarms with the local optimization characteristics of particle swarm algorithms. The optimal value of artificial fish is added with random disturbances as the initial solution of particles for hybrid optimization, which can quickly and accurately fit the measured settlement data to obtain the dual parameters of the time function. This method uses rectangular subdivision and introduces the complex Simpson formula to numerically integrate the definite integral of the prediction model, and then combines the time function to construct a dynamic prediction model for surface settlement of horizontal or gently inclined coal seam working faces based on the superposition principle. The dynamic prediction model for surface settlement can realize the refined prediction and simulation of surface settlement. Accurate dynamic prediction of mining settlement can ensure the safety of ground buildings (structures) during mining and the need to implement mining and reclamation. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 The present invention is a flow chart of the method for dynamic prediction of surface subsidence in coal mining based on a dual-parameter time function.

[0046] Figure 2 It is a schematic diagram of coal seam mining and subdivision in an embodiment of the present invention.

[0047] Figure 3 This is a flow chart of the adaptive hybrid optimization artificial fish swarm algorithm in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The present invention will be further described in detail below with reference to the figures and specific implementation process:

[0049] like Figure 1 and Figure 2 As shown, the method for dynamic prediction of surface subsidence in coal mining based on dual-parameter time function of the present invention determines the plane coordinates of the four turning points of the rectangular coal seam working face and the mining depth, the coal seam mining thickness m, the coal seam inclination α, and the coal seam strike azimuth through the coal mining engineering plan and mining data, determines the dynamic prediction time t (day) and the mining advancement speed v (meters per day) of the working face strike at time t, and simulates and inverts the parameters of the regional probability integral method according to the actual mining subsidence and extreme value of the movement deformation value of the coal mine, including the subsidence coefficient q, the horizontal movement coefficient b, the main influence angle tangent tanβ, the mining influence propagation angle θ, the downhill turning point offset S1, the uphill turning point offset S2, the strike left turning point offset S3, the strike right turning point offset S4, the coal seam working face strike and the inclined mining size D3, D1;

[0050] Then determine the direction dimension of the calculated working face as l=D3-S3-S4, and the inclination dimension as The working face is mined forward along the strike direction. The strike length of the working face at time t is v*t, and the dip width remains unchanged. The calculated working face rectangle is evenly divided into n rectangular subdivision units along the strike direction of the calculated working face with a width of 1 meter and a dip dimension L as the length.

[0051] Assume that the surface point whose settlement state needs to be predicted is the surface settlement prediction point, and the coordinates of the surface settlement prediction point are (x, y). Use equations (1)-(5) to calculate the settlement influence values ​​of n rectangular subdivision units on the surface settlement prediction point.

[0052] The probability integral method is used to calculate the settlement influence values ​​of n rectangular subdivision units on the surface settlement prediction point at time t. The settlement value of the surface settlement prediction point (x, y) is calculated using the following formula:

[0053]

[0054] In the formula, η, ξ are the integral variables of the double integral, η corresponds to the differential value of x, and ξ corresponds to the differential value of y; D is the calculation boundary of the coal seam; r is the main influence radius, which is obtained by the formula: r = H / tanβ; H is the mining depth, tanβ is the main influence angle tangent; W cm The maximum surface settlement under full mining conditions is calculated by the formula: W cm =mq·cosα; q is the sedimentation coefficient; m is the normal mining thickness of the coal seam; α is the inclination angle of the coal seam;

[0055] For the i-th (i∈[1,n]) rectangular subdivision unit in the n rectangular working surfaces, equation (1) is further calculated to obtain:

[0056]

[0057] In the formula, The minimum and maximum values ​​of the coordinates of the rectangular subdivision unit; The minimum and maximum values ​​of the inclination coordinates of the rectangular subdivision unit;

[0058] The complex Simpson formula (3) is used to numerically integrate the definite integral in formula (2):

[0059]

[0060] The settlement influence value expression (4) of the rectangular subdivision unit i on any surface settlement prediction point is obtained:

[0061]

[0062] The superposition principle is used to calculate the settlement impact value of mining in each rectangular subdivision unit on any surface settlement prediction point, that is, the static settlement value of the surface settlement prediction point after mining in the prediction area is completed is formula (5):

[0063]

[0064] In the formula: In the formula: is the formula (2) Using the numerical integration value of formula (3), is the formula (2) The numerical integration value of formula (3) is adopted.

[0065] Constructing a dual-parameter time function model and an objective function model of the dual parameters to be solved;

[0066] like Figure 3 As shown, the time function double parameter is defined as an artificial fish, and the parameters of the artificial fish swarm algorithm and the particle swarm algorithm are initialized, including the time function double parameter quantity N, the maximum number of iterations Maxgen, the maximum number of trials try_number, the crowding factor delta, the visual step coefficient, the initial step length, the initial field of view; the number of iterations, the population size, the learning factors c1 and c2, the maximum flight speed, the maximum and minimum inertia weights, and the upper and lower limits of the time function double parameter; initialize the time function double parameter value: generate N groups of time function double parameter values ​​through random numbers and the upper and lower limits of the time function double parameter;

[0067] Write an adaptive hybrid optimization artificial fish swarm algorithm, and solve the dual parameters of the time function model according to the initialization parameters, the time function model and the measured settlement data of the surface settlement prediction point (x, y);

[0068] Calculate the time of mining of n rectangular subdivision units relative to the start of mining respectively, and bring the calculated time into the dual-parameter time function model to calculate the time influence value of n rectangular subdivision units on the surface settlement prediction point;

[0069] Using v i (i=1,2,…,n) represents the advancement speed of the working face when mining to the i-th subdivision unit along the mining advancement direction, in meters per day;

[0070] Using t i (i=1,2,…,n) is the number of days that the working face takes to mine from the i-1th subdivision unit to the i-th subdivision unit along the mining advancement direction, in days;

[0071] Then the mining dimension of each mining subdivision unit is v i t i , in meters;

[0072] The last rectangular unit is mined with a strike dimension of v n t n =lv n-1 t n-1 , l is the mining dimension of the working face calculated at time t;

[0073] Remember iis the static settlement impact value of the surface settlement prediction point (x, y) when the mining of the i-th mining subdivision unit is completed. When the working face is mined to the i-th subdivision unit, that is, t = t1 + t2 + ... + t i The time when the working surface of the 1st to ith subdivision unit produces settlement influence on the surface settlement prediction point is t-t1, t-t1-t2, …, t-t1-t2-…-t i-1 The surface settlement value of the surface settlement prediction point at time t is calculated using formula (6).

[0074]

[0075] In the formula, w i (x, y) is calculated according to formula (5), (x, y) is the coordinate of the surface settlement prediction point, is t-t1-t2-…-t i-1 The time function value at .

[0076] Implementation example:

[0077] Step 1: For the mining working face, collect and organize the working face information into an Excel table. The relevant information of the working face is as follows: the geodetic coordinates of each corner point and the mining depth (X, Y, H) are as shown in the following table, the mining thickness m = 6320 mm, the coal seam strike azimuth 86.5°, the coal seam inclination α = 3°, the subsidence coefficient q = 0.49, the main influence angle tangent tanβ = 1.2, the mining influence propagation angle θ = 61.5° and the inflection point offset S1 = S2 = S3 = S4 = 0.

[0078] Table 1 Coordinates of turning points and mining depth of working face

[0079]

[0080] Step 2: Determine the dynamic prediction time, measured settlement value and corresponding mining advancement speed, as shown in Table 2.

[0081] Table 2 Dynamic prediction time of working face, measured settlement value and mining advancement speed

[0082]

[0083] Step 3: Determine the dual-parameter time function model. Here we take the Weibull time function model as an example, that is, Where W m is the maximum settlement value of the surface settlement prediction point, mm; c and k are parameters related to the properties of the overlying rock strata in the goaf; t is the time of the surface settlement prediction point, day.

[0084] Step 4: Initialize the parameters of the artificial fish swarm algorithm: time function dual parameter N = 100, maximum number of iterations Maxgen = 100, maximum number of attempts try_number = 100, crowding factor delt 0.418, visual step coefficient 0.1, initial step length 0.1, initial field of view 1; number of iterations 30, population size 60, learning factors c1 = 2 and c2 = 2, maximum flight speed 10, maximum and minimum inertia weights 0.9 and 0.4, time function dual parameter upper and lower limits c∈[0,1], k∈[0,1].

[0085] Step 5: Write an adaptive hybrid optimization artificial fish swarm algorithm and use the measured settlement data for the first 10 periods to obtain the dual parameters of the Weibull time function model as c = 0.055 and k = 0.508.

[0086] Step 6: Use the dynamic prediction model formulas (1)-(6) and the dual-parameter time function model to calculate the settlement values ​​of the surface settlement prediction points (x, y) for periods 1-16 as shown in Table 3.

[0087] Issue 1 2 3 4 5 6 7 8 Predicted settlement(mm) -4.86 -10.25 -18.12 -29.72 -48.04 -96.46 -152.23 -195.63 Issue 9 10 11 12 13 14 15 16 Predicted settlement(mm) -230.67 -269.15 -289.09 -299.22 -309.33 -316.91 -328.04 -359.45

[0088] Step 7: The calculated relative error between the predicted settlement value and the measured settlement value is 10.5%, which proves that the method for dynamic prediction of surface settlement in coal mining based on dual-parameter time function of the present invention can achieve more accurate dynamic prediction of mining settlement.

Claims

1. A method for dynamic prediction of surface subsidence in coal mining based on dual-parameter time function, characterized in that: The specific steps are as follows: Step 1: First, determine the plane coordinates of the four turning points of the rectangular working face and the mining depth, coal seam mining thickness, coal seam inclination, and coal seam strike azimuth through the coal mining engineering plan and mining data, determine the mining advancement speed v of the working face strike at any dynamic prediction time t and time t, in meters per day, and simulate the probability integral method parameters of the rectangular working face according to the actual mining sinking and movement deformation value extreme value of the coal mine, including the sinking coefficient q, the horizontal movement coefficient b, the main influence angle tangent tanβ, the mining influence propagation angle θ, the downhill turning point offset S1, the uphill turning point offset S2, the strike left turning point offset S3, the strike right turning point offset S4, and the coal seam working face strike and dip mining dimensions D3 and D1; Step 2: Determine the direction dimension of the calculated working surface as l = D3-S3-S4, and the inclination dimension is The working face is mined forward along the strike direction. The strike length of the working face at time t is v*t, and the dip width remains unchanged. The calculated working face rectangle is evenly divided into n rectangular subdivision units along the strike direction of the calculated working face with a width of 1 meter and a dip dimension L as the length. Step 3: Assume that the surface point whose settlement state needs to be predicted is the surface settlement prediction point, and the coordinates of the surface settlement prediction point are (x, y). Use the probability integral method formula to calculate the settlement influence values ​​of n rectangular subdivision units on the surface settlement prediction point respectively; Step 4: Construct a dual-parameter time function model and an objective function model of the dual parameters to be solved; Step 5: The field of view and step size of the artificial fish swarm algorithm are changed to an adaptive field of view and step size to obtain an adaptive artificial fish swarm algorithm, and the adaptive artificial fish swarm algorithm is combined with the particle swarm algorithm to obtain an adaptive hybrid optimization artificial fish swarm algorithm, and the adaptive hybrid optimization artificial fish swarm algorithm is used to solve the time function dual parameters of the dual-parameter time function model; Step 6: Calculate the time of mining of n rectangular subdivision units relative to the start of mining, and bring it into the dual-parameter time function model with solved dual parameters to calculate the time influence value of n rectangular subdivision units on the surface settlement prediction point; Step 7: Multiply the settlement influence values ​​and time influence values ​​of the n rectangular subdivision units respectively and then sum them up to obtain the settlement value of the surface settlement prediction point at time t.

2. The method for dynamic prediction of surface subsidence in coal mining based on dual-parameter time function according to claim 1 is characterized in that The steps of solving the dual parameters of the time function using the adaptive hybrid optimization artificial fish swarm algorithm are as follows: Step 1: Define the time function double parameter as an artificial fish, initialize the parameters of the artificial fish swarm algorithm and the particle swarm algorithm, including the time function double parameter quantity N, the maximum number of iterations Maxgen, the maximum number of trials try_number, the crowding factor delta, the visual step coefficient, the initial step length, the initial field of view; the number of iterations, the population size, the learning factors c1 and c2, the maximum flight speed, the maximum and minimum inertia weights, and the upper and lower limits of the time function double parameter; initialize the time function double parameter value: generate N groups of time function double parameter values ​​through random numbers and the upper and lower limits of the time function double parameter; Step 2: Select a surface settlement prediction point, calculate the objective function value corresponding to each set of time function double parameters according to the measured settlement value of the point, and record it on the bulletin board, calculate the distance between the current time function double parameter and the optimal time function double parameter in the bulletin board as the field of view of the current time function double parameter, and calculate the step length through the visual step coefficient; Step 3: Each group of time function double parameters performs foraging, flocking, chasing and random behaviors respectively, and selects the time function double parameter with the best objective function value as the current optimal time function double parameter; Step 4: Within the maximum number of iterations, after each time function double parameter executes the above behavior, compare the objective function value with the billboard. If the billboard update condition is met, update the billboard and the current optimal time function double parameter; Step 5: Determine whether the maximum number of iterations has been reached; when the iteration is completed, use the current optimal time function double parameter as the initial particle value of the particle swarm algorithm; Step 6: Execute the particle swarm algorithm to further optimize the particle optimal solution under the dual parameter state of the current optimal time function; Step 7: Compare the optimal dual parameter values ​​of the time function of the adaptive artificial fish swarm and the optimal solution of the particles, and take the better value of the two as the global optimal result and output it as the dual parameter of the time function of the dual parameter time function model.

3. The method for dynamic prediction of surface subsidence in coal mining based on dual-parameter time function according to claim 1 is characterized in that The probability integral method is used to calculate the settlement influence values ​​of n rectangular subdivision units on the surface settlement prediction point at time t. The settlement value of the surface settlement prediction point (x, y) is calculated using the following formula: In the formula, η, ξ are the integral variables of the double integral, η corresponds to the differential value of x, and ξ corresponds to the differential value of y; D is the calculation boundary of the coal seam; r is the main influence radius, which is obtained by the formula: r = H / tanβ; H is the mining depth, tanβ is the main influence angle tangent; W cm The maximum surface settlement under full mining conditions is calculated by the formula: W cm =mq·cosα; q is the sedimentation coefficient; m is the normal mining thickness of the coal seam; α is the inclination angle of the coal seam; For the i-th (i∈[1,n]) rectangular subdivision unit in the n rectangular working surfaces, equation (1) is further calculated to obtain: In the formula, The minimum and maximum values ​​of the coordinates of the rectangular subdivision unit; The minimum and maximum values ​​of the inclination coordinates of the rectangular subdivision unit; The complex Simpson formula (3) is used to numerically integrate the definite integral in formula (2): The settlement influence value expression (4) of the rectangular subdivision unit i on any surface settlement prediction point is obtained: The superposition principle is used to calculate the settlement impact value of mining in each rectangular subdivision unit on any surface settlement prediction point, that is, the static settlement value of the surface settlement prediction point after mining in the prediction area is completed is formula (5): Where: is the formula (2) Using the numerical integration value of formula (3), is the formula (2) The numerical integration value of formula (3) is adopted.

4. The method for dynamic prediction of surface subsidence in coal mining based on dual-parameter time function according to claim 1 is characterized in that: The method for calculating the settlement value of the surface settlement prediction point at time t by using the settlement influence value and time influence value is: Using v i (i=1,2,…,n) represents the advancement speed of the working face when mining to the i-th subdivision unit along the mining advancement direction, in meters per day; Using t i (i=1,2,…,n) is the number of days that the working face takes to mine from the i-1th subdivision unit to the i-th subdivision unit along the mining advancement direction, in days; Then the mining dimension of each mining subdivision unit is v i t i , in meters; The last rectangular unit is mined with a strike dimension of v n t n =lv n-1 t n-1 , l is the mining dimension of the working face calculated at time t; Remember i is the static settlement impact value of the surface settlement prediction point (x, y) when the mining of the i-th mining subdivision unit is completed. When the working face is mined to the i-th subdivision unit, that is, t = t1 + t2 + ... + t i The time when the working surface of the 1st to ith subdivision unit produces settlement influence on the surface settlement prediction point is t-t1, t-t1-t2, …, t-t1-t2-…-t i-1 , the settlement influence value of each subdivision unit working surface on the surface settlement prediction point is assigned a time function coefficient and superimposed to obtain the surface settlement value of the surface settlement prediction point at time t. The surface settlement dynamic prediction model of the surface settlement prediction point is formula (6): In the formula, w i (x, y) is calculated according to formula (5), (x, y) is the coordinate of the surface settlement prediction point, is t-t1-t2-…-t i-1 The time function value at .

Images