A method and system for extracting induced polarization information from transient electromagnetic data

Abstract

The present invention discloses a method and system for extracting induced polarization information from transient electromagnetic data, which includes: forming an initial frog population, where each frog represents the resistivity, thickness, and polarization rate of each layer; converting the resistivity of each layer into complex resistivity and substituting it into the TEM forward modeling to calculate the vertical magnetic field; then calculating the fitness of each frog, and then performing an iterative search operation of the shuffled frog leaping algorithm with the minimization of fitness as the optimization goal. After the iteration terminates, the model parameters corresponding to the frog with the minimum fitness are output and used as the inversion optimal solution, and the polarization rates of each layer obtained are the induced polarization information extracted. In addition, the present invention uses the Tent chaotic distribution operator to improve the initial population distribution and uses an adaptive movement factor to replace the random operator in local search. By only adjusting the position of the worst individual during the evolution process, time-consuming iterative calculations are avoided, and the induced polarization information is effectively extracted, providing a new technology for actual transient electromagnetic data processing and interpretation.

Description

Technical Field

[0001] The present invention belongs to the technical field of geophysical exploration, and particularly relates to a method and system for extracting induced polarization information from transient electromagnetic data. Background Art

[0002] The transient electromagnetic method (TEM, Time Domain Electromagnetic Method) and the induced polarization method (IP, induced polarization method) are both widely used electrical prospecting techniques in geophysical exploration (mineral exploration, hydrogeology, engineering investigation, environmental investigation, etc.); the TEM mainly studies the difference in conductivity between rocks and ores, and the IP mainly studies the difference in induced polarization effect between underground rocks and ores. Existing research shows that extracting induced polarization information from electromagnetic method signals has relatively optimistic prospects and is theoretically feasible, but there are also some limitations: (1) The forward modeling research considering the induced polarization effect can analyze the influence of induced polarization parameters on the time-domain electromagnetic method exploration data, but it cannot extract the induced polarization parameters themselves; (2) Most of the research on extracting induced polarization information by inversion uses traditional linear inversion methods, and the inversion results depend on the selection of the initial model and are prone to falling into local extrema.

[0003] The fully nonlinear inversion method provides a new research idea for processing and interpreting electromagnetic method data and has received extensive attention from geophysical researchers. Although these nonlinear inversion methods have been widely used in the inversion of electromagnetic method data, there are the following deficiencies when directly used to interpret transient electromagnetic data containing induced polarization information: (1) When considering the induced polarization effect, the resistivity of the polarizable body underground is a complex number related to frequency, and the amount of calculation increases during the forward modeling iteration. When using Monte Carlo type search algorithms such as simulated annealing SA, particle swarm optimization PSO, and genetic algorithm GA, the operation efficiency is relatively low; (2) After considering the induced polarization effect, the inversion model parameters increase the induced polarization parameters on the basis of the original resistivity and thickness, making the nonlinear inversion process more difficult to converge.

[0004] The shuffled frog leaping algorithm (SFLA) is a heuristic swarm intelligence search algorithm that organically combines global information interaction and local search through grouping operators and memetic search. At the same time, only the position of the worst individual in the swarm is adjusted during the evolution process. Compared with swarm intelligence search algorithms such as GA and PSO that need to adjust all individuals, it has higher execution efficiency.

[0005] Therefore, in the context of considering the induced polarization effect, that is, in the application background of extracting induced polarization information from transient electromagnetic data, how to use the Shuffled Frog Leaping Algorithm (SFLA) to extract induced polarization information and solve the technical problems in the prior art that the deep learning method based on a large amount of data-driven requires repeated iterative gradient learning, has an excessive amount of iterative operations, and is too time-consuming is what this invention studies. Summary of the Invention

[0006] In order to solve the technical problems existing in the prior art of extracting induced polarization information from transient electromagnetic data, such as the need for repeated iterative gradient learning, excessive amount of iterative operations, and being too time-consuming, the present invention further provides a method and system for extracting induced polarization information from transient electromagnetic data. Specifically, it realizes the technology of extracting induced polarization information from transient electromagnetic data based on the Shuffled Frog Leaping Algorithm (SFLA), makes full use of the characteristics of the organic combination of global search and local search of the Shuffled Frog Leaping Algorithm, and only adjusts the position of the worst individual in the population during the evolution process to find the optimal parameters of the geoelectric model, avoiding time-consuming iterative training calculations.

[0007] On the one hand, the present invention provides a method for extracting induced polarization information from transient electromagnetic data, which includes the following steps:

[0008] Step 1: Obtain the parameter search range of the geoelectric model corresponding to the detection area, and form the initial frog population of the Shuffled Frog Leaping Algorithm based on the parameter search range;

[0009] Among them, the model parameters of the geoelectric model include: the resistivity, thickness, and polarization rate corresponding to each layer of the earth medium in the geoelectric model, and each frog represents a set of model parameters of the geoelectric model:

[0010] Step 2: Convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer;

[0011] Step 3: Substitute the complex resistivity of each layer corresponding to each frog into the transient electromagnetic TEM forward calculation to obtain the vertical magnetic field corresponding to each frog;

[0012] Step 4: Calculate the fitness of each frog in the frog population, and the fitness is constructed based on the error between the forward vertical magnetic field and the vertical magnetic field observation data of the detection area, where the vertical magnetic field observation data is obtained by TEM detection;

[0013] Step 5: Perform iterative search operations of the Shuffled Frog Leaping Algorithm with minimizing the fitness as the optimization goal until the iterative termination condition is met, then output the model parameters corresponding to the frog with the minimum fitness, and use the model parameters as the inversion optimal solution, and the polarization rates of each layer obtained are the extracted induced polarization information.

[0014] It should be noted that the Shuffled Frog Leaping Algorithm (SFLA) organically combines global information interaction and local search through grouping operators and memetic search. At the same time, during the evolution process, only the position of the worst individual in the population is adjusted. Compared with swarm intelligence search algorithms such as GA and PSO that need to adjust all individuals, it has higher execution efficiency, can better balance global and local search performance, is conducive to the effective search for the global optimal solution and the improvement of the overall execution efficiency. Therefore, the technical solution of the present invention introduces the Shuffled Frog Leaping Algorithm (SFLA) into the induced polarization information extraction technology of transient electromagnetic data. On the one hand, it provides a new technical means to realize the extraction of induced polarization information. On the other hand, by making full use of the advantages of the Shuffled Frog Leaping Algorithm, only the position of the worst individual in the population is adjusted during the evolution process, without the need for gradient learning through repeated iterations based on a large amount of data, avoiding time-consuming iterative training calculations, and finally obtaining an accurate geoelectric model of the detection area. Among them, the obtained polarization rates of each layer are the induced polarization information of the transient electromagnetic data of the detection area extracted.

[0015] Further preferably, in the Shuffled Frog Leaping Algorithm, the frog population is divided into k memetic groups through grouping operators, and then local search is performed on the frog with the worst fitness in each memetic group;

[0016] Among them, the local search is performed and updated according to the combination of formula 1 and formula 2 or the combination of formula 3 and formula 2 as follows;

[0017]

[0018] Formula 2: X w (t + 1) = X w (t) + Δ w (t + 1), S min ≤ Δ w (t) ≤ S max

[0019]

[0020] In the formula, Δ w (t + 1) is the moving step size, i represents the current global iteration number, t is the local search number of the current memetic group, iter g is the preset maximum global iteration number, X b 9t), X w 9t) respectively represent the frogs with the best fitness and the worst fitness in the current memetic group at the t-th local search; S min and S max are the minimum and maximum boundaries of the distance range allowed for the frog to move; X g (t) represents the frog with the best fitness in the frog population at the t-th local search of the current memetic group;

[0021] Among them, if the fitness of the frog updated using the combination of Formula 1 and Formula 2 or the combination of Formula 3 and Formula 2 is not better than that before the update, a new frog is randomly generated to replace the original frog and then local search is performed.

[0022] The standard Shuffled Frog Leaping Algorithm uses a single random operator rand, which limits the local search range within each meme group to X b -X w or X g -X w In the later stage of the algorithm, it is prone to fall into local extrema. The technical solution of the present invention optimizes it using an adaptive movement factor. Among them, in the technical solution proposed by the present invention, as the number of iterations increases, the movement step size will gradually increase; after using the adaptive movement factor to adjust the movement step size, in the early stage of local search evolution, the global search ability of the solution can be well restricted, and to a certain extent, the optimization range of frog individuals is expanded, so that most frog individuals can be better distributed in the feasible region. In the later stage of iteration, the movement step size is larger, which can ensure the global search ability of the solution as much as possible; in addition, the adaptive adjustment of the movement step size avoids unnecessary detailed optimization search, which is beneficial to accelerating the convergence speed of the algorithm, that is, it not only improves the solution accuracy of the algorithm, but also avoids the algorithm from falling into local optimum prematurely.

[0023] Further optionally, the initial frog population is constructed based on the Tent map, and the formula of the Tent map is as follows:

[0024]

[0025] In the formula, the distribution range x of the chaotic sequence value xn n ∈[0,1], and a is a control parameter;

[0026] Among them, according to the mapping relationship between the parameter search range and the distribution range of the chaotic sequence value, the model parameter value of the initial geoelectric model randomly generated within the parameter search range is mapped to the chaotic sequence value x n ;

[0027] Then, according to the Tent map, the chaotic sequence value x n is updated to the chaotic sequence value x n+1 ;

[0028] Then, according to the mapping relationship between the parameter search range and the distribution range of the chaotic sequence value, the chaotic sequence value x n+1 is converted into the model parameter of the geoelectric model, and the model parameter of the initial geoelectric model is updated, thereby forming the initial frog population.

[0029] It should be noted that there is a problem in the standard shuffled frog leaping algorithm that the initial population distribution is not uniform enough. To improve this performance, the technical solution of the present invention constructs the initial frog population based on the Tent mapping. Among them, the Tent mapping has the characteristic of piecewise linearity and can generate a relatively uniform initial population. Therefore, the present invention uses the Tent chaotic distribution to improve the construction of the initial population, which can effectively solve the problem of non-uniform initial population distribution in the standard shuffled frog leaping algorithm, and improve the global search ability of the algorithm by improving the initial population distribution.

[0030] Further optionally, the execution process of step 5 is as follows:

[0031] Execute the grouping operator to divide the frogs in the frog population into k meme groups;

[0032] Execute local search, and perform local search on the frog with the worst fitness in each meme group respectively. If the fitness of the frog updated by using the local search formula is not better than the fitness before updating, randomly generate a new frog to replace the original frog and then perform local search;

[0033] Execute global shuffling. After all meme groups have completed local search, shuffle all the frogs in the frog population globally, and then re-execute the grouping operator and local search;

[0034] Among them, repeat the operations of the grouping operator, local search, and global shuffling until the global search times reach the preset global maximum iteration times, and then output the model parameters corresponding to the frog with the minimum fitness as the inversion optimal solution.

[0035] Further optionally, when executing the grouping operator, first arrange the frogs in the frog population in descending order of fitness; then put the first frog into the first meme group, the second frog into the second meme group, until the kth frog is put into the kth meme group; then the k + 1th frog is put back into the first meme group, the k + 2th frog is put into the second meme group, and so on, divide the frogs in the frog population into k meme groups.

[0036] Further optionally, the formula for the fitness is:

[0037]

[0038] In the formula, Fitness is the fitness, N is the time sampling point, represents the vertical magnetic field observation data at the i-th time sampling point; represents the vertical magnetic field at the i-th time sampling point obtained by forward calculation based on the geoelectric model determined by the model parameters corresponding to the frog.

[0039] Further optionally, in step 2, the Dias induced polarization model or the Cole-Cole induced polarization model is used to convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer;

[0040] Among them, the formula for the complex resistivity of the Dias induced polarization model is:

[0041]

[0042] The formula for the complex resistivity of the Cole-Cole induced polarization model is:

[0043]

[0044] And there is:

[0045]

[0046] In the formula, ρ(ω) is the complex resistivity, ρ is the resistivity without considering the polarization effect, m is the polarizability; τ is the time constant; η is the electrochemical parameter; δ is the polarization resistivity coefficient, τ′, τ″, μ are all intermediate variables, i is the complex number symbol, ω is the angular frequency, and c is the frequency-dependent coefficient.

[0047] On the second aspect, the present invention provides a system based on the extraction method, which includes:

[0048] An initial frog population generation module, configured to obtain the parameter search range of the geoelectric model corresponding to the detection area, and form an initial frog population of the shuffled frog leaping algorithm based on the parameter search range;

[0049] Among them, the parameters of the geoelectric model include: the resistivity, thickness, and polarizability of each layer of the earth medium in the geoelectric model, and each frog represents a set of parameters of the geoelectric model:

[0050] A complex resistivity calculation module, configured to convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer;

[0051] A forward modeling module, configured to substitute the complex resistivity of each layer corresponding to each frog into the transient electromagnetic TEM forward modeling to obtain the vertical magnetic field corresponding to each frog;

[0052] A fitness calculation module, configured to calculate the fitness of each frog in the frog population, and the fitness is constructed based on the error between the forward modeled vertical magnetic field and the vertical magnetic field observation data of the detection area;

[0053] An extraction module is used to perform grouping operators, local search, and global shuffling of the shuffled frog leaping algorithm with the minimum fitness as the optimization goal until the iterative termination condition is met, and then output the model parameters corresponding to the frog with the minimum fitness, which are used as the inversion optimal solution, and the polarizabilities of each layer obtained are the extracted induced polarization information.

[0054] In a third aspect, the present invention provides a computer device, which includes:

[0055] One or more processors;

[0056] A memory storing one or more computer programs;

[0057] Wherein, the processor calls the computer program to implement:

[0058] The steps of the method for extracting induced polarization information from transient electromagnetic data.

[0059] In a fourth aspect, the present invention provides a computer-readable storage medium, which stores a computer program, and the computer program is called by a processor to implement:

[0060] The steps of the method for extracting induced polarization information from transient electromagnetic data.

[0061] Beneficial effects

[0062] 1. The technical solution of the present invention realizes the extraction technology of induced polarization information from transient electromagnetic data based on the shuffled frog leaping algorithm (SFLA). It makes full use of the characteristics of the organic combination of global search and local search of the shuffled frog leaping algorithm. During the evolution process, only the position of the worst individual in the population is adjusted, without the need for gradient learning through repeated iterations based on a large amount of data, avoiding time-consuming iterative training calculations. Compared with swarm intelligence search algorithms such as GA and PSO that need to adjust all individuals, it has higher execution efficiency, can better balance global and local search performance, is conducive to the effective search for the global optimal solution and the improvement of the overall execution efficiency.

[0063] 2. The technical solution of the present invention preferably optimizes the technology that uses a single random operator rand for local search in the standard shuffled frog leaping algorithm. Specifically, it is optimized by using an adaptive movement factor. Among them, after using the adaptive movement factor to adjust the movement step size, in the early stage of local search evolution, it can well limit the global search ability of the solution, and to a certain extent expand the optimization range of frog individuals, so that most frog individuals can be better distributed in the feasible domain. And when iterating to the later stage of the algorithm, the movement step size is larger, which can ensure the global search ability of the solution as much as possible; in addition, the adaptive adjustment of the movement step size avoids unnecessary detailed optimization searches, which is conducive to accelerating the convergence speed of the algorithm, that is, not only improves the solution accuracy of the algorithm, but also avoids the algorithm from falling into local optimality prematurely.

[0064] 3. The technical solution of the present invention preferably optimizes the initial population in the standard Shuffled Frog Leaping Algorithm (SFLA). Among them, the initial frog population is constructed based on the Tent mapping. By using the Tent chaotic distribution to improve the construction of the initial population, the problem of uneven distribution of the initial population in the standard SFLA can be effectively solved, and the global search ability of the algorithm can be improved by improving the initial population distribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 is a schematic diagram of layered earth provided by the present invention;

[0066] Figure 2 is a schematic flow chart of a method for extracting induced polarization information from transient electromagnetic data provided in Embodiment 1 of the present invention;

[0067] Figure 3 is a schematic diagram of the inversion result when the middle layer of a three-layer A-type geoelectric model is polarized, where the figures (a), (b), (c), and (d) respectively correspond to resistivity, polarizability, H Z fitting curve, fitness value curve;

[0068] Figure 4 is a schematic diagram of the inversion result when the middle layer of a four-layer KH-type geoelectric model is polarized, where the figures (a), (b), (c), and (d) respectively correspond to resistivity, polarizability, H Z fitting curve, fitness value curve;

[0069] Figure 5 is a schematic diagram of the inversion result when the middle layer of a five-layer HKH-type geoelectric model is polarized, where the figures (a), (b), (c), and (d) respectively correspond to resistivity, polarizability, H Z fitting curve, fitness value curve. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0070] A method for extracting induced polarization information from transient electromagnetic data provided by the present invention introduces the Shuffled Frog Leaping Algorithm (CSFLA) to extract induced polarization information, and further makes the following improvements on the basis of the non-linear inversion of the standard Shuffled Frog Leaping Algorithm: (1) The Tent chaotic distribution operator is used to improve the initial population distribution to increase the population diversity in the meme group evolution and enhance the global search ability of the algorithm; (2) An adaptive movement factor is designed to replace the random operator to balance the early search and late convergence capabilities of the algorithm.

[0071] For the detection area, the vertical magnetic field observation data of the detection area are obtained by TEM detection. The present invention will be further described below in conjunction with embodiments.

[0072] Embodiment 1:

[0073] A method for extracting induced polarization information from transient electromagnetic data provided by this embodiment includes the following steps:

[0074] Step 1: Obtain the parameter search range of the geoelectric model corresponding to the detection area, and form an initial frog population of the shuffled frog leaping algorithm based on the parameter search range.

[0075] Let the frog population size be P, and the initial frog population X(0) = {X1, X2, …, X P}. If the dimension of the feasible solution space is d, then the i-th frog in the solution space is represented as X i = [x i1 , x i2 , …, x id . During the TEM inversion process, the present invention is directed to a homogeneous horizontally layered underground medium, and its geoelectric model is as Figure 1 shown. Assume that the x and y axes of the Cartesian coordinate system are located on the ground surface, and the positive z axis is vertically downward. Then, a horizontal circular transmitting coil with a radius of a is located on the ground surface, and the transmitting current is I. And the earth medium has a total of Q layers, and the resistivity and thickness of each layer are ρ i and h i (i = 1, 2, …, Q); the model parameters to be obtained are the resistivity, thickness, and polarizability of each layer. Therefore, each set of model parameters includes: the resistivity, thickness, and polarizability of each layer, and is represented by the frog X i , and each parameter is represented by the frog x i . The total number of each set of parameters corresponds to the dimension d.

[0076] Generally, the standard shuffled frog leaping algorithm determines each frog in the frog population (i.e., using the formation resistivity, layer thickness, and polarizability model parameter values as the frog values in the frog population) in a random rand manner within a certain search parameter range, and then forms an initial population. In order to overcome the defects of the standard shuffled frog leaping algorithm, such as the non-uniform distribution of the initial population, the randomness of the moving step size, and the slow convergence speed, this embodiment constructs the initial frog population based on the Tent mapping. Among them, the chaotic distribution is a relatively uniform distribution function. Because of its characteristics of randomness, ergodicity, and regularity, it can well maintain the population diversity and has been widely used in swarm intelligence algorithms. Compared with other mappings, the Tent mapping has the characteristic of piecewise linearity and can generate a relatively uniform initial population. The formula of the Tent mapping is as follows:

[0077]

[0078] In the formula, the distribution range of the chaotic sequence value x n is x n ∈ [0, 1], a is a control parameter, and usually takes 2.

[0079] The specific construction process of using the Tent mapping to construct the initial frog population is as follows:

[0080] According to the mapping relationship between the parameter search range and the distribution range of the chaotic sequence values, the initial geoelectric model parameter values randomly generated within the parameter search range are mapped to the chaotic sequence value x n .

[0081] Then, according to the Tent mapping, the chaotic sequence value x n is updated to the chaotic sequence value x n+1 .

[0082] Then, according to the mapping relationship between the parameter search range and the distribution range of the chaotic sequence values, the chaotic sequence value x n+1 is converted into geoelectric model parameters, and the initial geoelectric model parameters are updated, thereby forming the initial frog population.

[0083] It should be understood that in this embodiment, the Tent chaotic distribution is used to improve the construction of the initial population, avoiding the influence of the uneven distribution of the initial population in the standard shuffled frog leaping algorithm. However, in other feasible embodiments, on the basis of not having such high requirements for model accuracy, using the standard shuffled frog leaping algorithm to generate the initial frog population also falls within the protection scope of the present invention.

[0084] Based on the above theoretical statements, this embodiment gives the global maximum number of iterations iter g and the local maximum number of iterations iter l , the frog population size P, the number of meme groups k, the control parameter a, and randomly generate the initial population according to the search ranges of the geoelectric model parameters ρ, h, and m.

[0085] Step 2: Convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer. Among them, the Dias induced polarization model or the Cole-Cole induced polarization model is used to convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer.

[0086] Dias model

[0087] Currently, there are many mathematical models describing the induced polarization characteristics of rocks and minerals. Among them, the Dias model proposed by Dias in 1968 is one of the widely used induced polarization response models. The complex resistivity calculation formula of the Dias model is:

[0088]

[0089] And there is:

[0090]

[0091] Where ρ(ω) is the complex resistivity, ρ is the resistivity without considering the polarization effect, m is the polarizability; τ is the time constant; η is the electrochemical parameter; δ is the polarization resistivity coefficient, τ′, τ″, and μ are all intermediate variables, i is the complex number symbol, ω is the angular frequency, and the Cole-Cole model

[0092] The Cole-Cole model is another induced polarization response model that is currently used more frequently. It was proposed by the Cole brothers in 1941. The calculation formula for the complex resistivity of the Cole-Cole model is:

[0093]

[0094] Where c is the frequency-dependent coefficient.

[0095] Step 3: Substitute the complex resistivity of each layer corresponding to each frog into the transient electromagnetic TEM forward calculation to obtain the vertical magnetic field corresponding to each frog.

[0096] According to the derivation formula of Kaufman, the vertical magnetic field generated by the transient electromagnetic central loop in the z-component of the frequency domain can be expressed as:

[0097]

[0098] Where H z is the vertical magnetic field, z is the vertical coordinate point of reception, J1 is the first-order Bessel function, λ is the Hankel transform integration variable, and R1 * is the input impedance of the first layer (calculated by the recurrence formula disclosed in Kaufman and substituting the complex resistivity for the resistivity). The above formula can be calculated by Hankel integration, that is, using 47-point Hankel filtering coefficients to calculate the electromagnetic response in the frequency domain, and then performing the Gaver-Stehfest transform and convolving with the transmitted current to calculate the time-domain electromagnetic response of any transmitted waveform (vertical magnetic field data H z ).

[0099] A brief description of the TEM forward process is as follows: (1) Construct a simplified layered geoelectric model based on the geological information of the detection area and add resistivity and formation thickness parameters to the geoelectric model; (2) By giving an induced polarization response model (Dias model or Cole-Cole model), add the polarizability information of the formation to the geoelectric model, and then calculate the complex resistivity ρ(w) related to frequency to replace the real resistivity ρ in the geoelectric model parameters; (3) Based on the complex resistivity of each layer, calculate the transformation impedance between adjacent layers, and then calculate the theoretical impedance on the ground by the recurrence formula, and further obtain the time-domain electromagnetic response with induced polarization response (vertical magnetic field data H z ).

[0100] Since the TEM forward modeling technique has been documented in existing literature, the present invention does not provide specific descriptions thereof. It should be understood that the present invention uses complex resistivity to replace resistivity to obtain the time-domain TEM response with induced polarization effect (vertical magnetic field data H z ).

[0101] Step 4: Calculate the fitness of each frog in the frog population. The fitness is constructed based on the error between the forward-calculated vertical magnetic field and the vertical magnetic field observation data in the detection area.

[0102] It should be understood that the purpose of inversion is to find a model that can best explain a set of observation data. Therefore, the present invention selects the relative error based on the theoretical value of the vertical magnetic field in the time domain as the fitness function to represent the error between the model value and the observation value. The definition expression of the fitness function is as follows:

[0103]

[0104] In the formula, Fitness is the fitness, N is the time sampling point, represents the vertical magnetic field observation data at the i-th time sampling point; is the geoelectric model determined based on the pattern parameters corresponding to the frog, and the vertical magnetic field at the i-th time sampling point obtained through forward calculation. During the inversion process, the optimization goal is to minimize the fitness function, that is, the relative error between the model value and the observation value. The smaller the result of Fitness, the closer the model is to the actual situation.

[0105] In other feasible embodiments, without departing from the idea of using fitness to characterize the error between the vertical magnetic field observation data and the vertical magnetic field data obtained through forward calculation, selecting other types of error functions also falls within the protection scope of the present invention.

[0106] In this embodiment, the data space is represented by N vertical magnetic field values, that is, H z =[H z1 ,H z2 ,…,H zN , and the model space is represented by the model parameters of l layers, that is, M = [ρ1, ρ2, …, ρ l ; h1, h2, …, h l1 ; m c1 , m c2 ,…, m cl , N is the time sampling point, a total of 91; ρ i , h i and m ci represent the resistivity, thickness, and polarization rate of the i-th layer respectively.

[0107] According to the above formula, each frog corresponds to a set of model parameters, and each set of model parameters determines a geoelectric model. Then, the vertical magnetic field data with induced polarization effect in all time domains is calculated for each geoelectric model, and finally, the fitness of each frog is calculated according to the formula of the fitness function.

[0108] Step 5: Take minimizing the fitness as the optimization goal, perform iterative search operations of the grouping operator, local search, and global shuffling of the shuffled frog leaping algorithm until the iterative termination condition is met, then output the model parameters corresponding to the frog with the minimum fitness, and use the model parameters as the inversion optimal solution. The polarization rates of each layer obtained are the induced polarization information extracted.

[0109] The process of using the shuffled frog leaping algorithm for search and update is as follows:

[0110] (1) Execute the grouping operator to divide the frogs in the frog population into k meme groups.

[0111] (2) Execute local search. Perform local search on the frog with the worst fitness in each meme group respectively. If the fitness of the frog updated using the local search formula is not better than the fitness before update, randomly generate a new frog to replace the original frog and then perform local search.

[0112] (3) Execute global shuffling. After all meme groups complete local search, perform global shuffling on all frogs in the frog population, and then re - execute the grouping operator and local search.

[0113] (4) Repeat the operations of the grouping operator, local search, and global shuffling until the global search times reach the preset global maximum iteration times, and then output the model parameters corresponding to the frog with the minimum fitness as the inversion optimal solution.

[0114] In this embodiment, regarding the grouping operator: After generating the initial frog group, arrange the frogs in the frog group in descending order of individual fitness, that is, sort the calculated objective function in descending order, and then divide the entire frog group into k meme groups. Each meme group contains n frogs, satisfying the relationship P = k×n. The specific grouping method is as follows: The first frog is put into the first meme group, the second frog is put into the second meme group, until the k - th frog is put into the k - th meme group; then the (k + 1) - th frog is put back into the first meme group, the (k + 2) - th frog is put into the second meme group, and so on until all frogs are grouped. The specific meme grouping formula is:

[0115] M i ={X i+m(l-1) ∈P|1≤l≤k},1≤i≤k

[0116] In the formula, M iis the i-th meme group. The frogs with the best and worst fitness in each meme group are denoted as X b and X w , and the frog with the best fitness in the entire frog population is denoted as X g .

[0117] Regarding local search:

[0118] First, regarding the local search of the standard Shuffled Frog Leaping Algorithm:

[0119] After the frog population is grouped, local search operations are performed on the worst frogs in each meme group (the number of local search times is iter l ), and its update formula is:

[0120]

[0121] Among them, Δ w (t + 1) is the moving step size, rand is an independent random number between [0, 1], t is the number of iterations, S min and S max are the distance ranges allowed for the frog to move. After the update, if the fitness of the obtained frog X w (t + 1) is better than that of X w (t), then it replaces X w (t) in the meme group; otherwise, update the moving step size according to the following formula and re-perform the local search:

[0122] Δ w (t + 1) = rand × (X g (t) - X w (t))

[0123] If the updated fitness is still not better than the fitness before the update, then randomly generate a new solution X w (t + 1) to replace X w (t). Among them, the local maximum number of iterations iter l is set, and the local search is performed in each meme group according to the above process until the local maximum number of iterations iter l is satisfied. After all meme groups have completed the local search, global shuffling is performed.

[0124] It should be understood that the above content is the local search of the standard Shuffled Frog Leaping Algorithm. On the basis that the required model accuracy meets the requirements, other feasible embodiments can adopt the above standard local search, which also falls within the protection scope of the present invention. In this embodiment, it is considered that the single random operator rand in the standard local search limits the local search range within each meme group to X b -X w or X g-X w , it is prone to falling into local extrema in the later stage of the algorithm. Therefore, it is improved by using an adaptive movement factor as follows:

[0125] After the frog group is grouped, local search operations are performed on the worst frogs in each meme group (the number of local search times is iter l ), and its update formula is:

[0126]

[0127] X w (t + 1) = X w (t) + Δ w (t + 1), S min ≤ Δ w (t) ≤ S max

[0128] Among them, i represents the current global iteration number.

[0129] After the update, if the fitness of the obtained frog X w (t + 1) is better than that of X w (t), then it replaces X w (t) in the meme group; otherwise, the movement step size is updated according to the following formula, and the local search is restarted:

[0130]

[0131] Similarly, if the fitness after the update is still not better than the fitness before the update, a new solution X w (t + 1) is randomly generated to replace X w (t). Among them, the local maximum iteration number iter l is set, and the local search is performed in each meme group according to the above process until the local maximum iteration number iter l is satisfied. After all meme groups have completed the local search, global shuffling is performed.

[0132] Regarding global shuffling: After the local search is completed, the frogs in all meme groups are globally shuffled, that is, re - mixed and sorted and local search is performed repeatedly until the defined global search number iter g is reached.

[0133] Result output: When the iteration termination condition is reached, the model parameters corresponding to the frog with the smallest fitness value are used as the optimal solution for inversion, and the model is output.

[0134] In summary, the present invention realizes the extraction technology of induced polarization information from transient electromagnetic data based on the Shuffled Frog Leaping Algorithm (SFLA). By making full use of the characteristics of the organic combination of global search and local search of the SFLA, only the position of the worst individual in the population is adjusted during the evolution process to find the optimal parameters of the geoelectric model, avoiding time-consuming iterative training calculations. More importantly, the present invention also optimizes the SFLA. On the one hand, the Tent chaotic distribution operator is used to improve the distribution of the initial population to enhance the population uniformity within the meme group during evolution and strengthen the global search ability of the algorithm. On the other hand, an adaptive movement factor is used to replace the random operator. In the early stage of the algorithm evolution, a smaller movement step size is preferably selected, which can appropriately limit the global search ability of the frog individuals, maintain the diversity of the population, and avoid premature convergence. In the middle and late stages of evolution, a larger movement step size is preferably selected to ensure the global search ability of the algorithm, avoid falling into local optimal solutions, and give full play to the role of local search of the frog leaping algorithm, enabling the algorithm to accelerate the convergence speed and improve the convergence accuracy during the optimization process.

[0135] Embodiment 2:

[0136] This embodiment provides a system based on the extraction method, which includes: an initial frog population generation module, a complex resistivity calculation module, a forward modeling module, a fitness calculation module, and an extraction module.

[0137] Among them, the initial frog population generation module is used to obtain the parameter search range of the geoelectric model corresponding to the detection area and form the initial frog population of the Shuffled Frog Leaping Algorithm based on the parameter search range.

[0138] The complex resistivity calculation module is used to convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer.

[0139] The forward modeling module is used to substitute the complex resistivity of each layer corresponding to each frog into the transient electromagnetic TEM forward modeling to obtain the vertical magnetic field corresponding to each frog.

[0140] The fitness calculation module is used to calculate the fitness of each frog in the frog population, and the fitness is constructed based on the error between the forward modeled vertical magnetic field and the vertical magnetic field observation data of the detection area.

[0141] The extraction module is used to perform the grouping operator, local search, and global shuffling of the Shuffled Frog Leaping Algorithm with the minimization of fitness as the optimization goal until the iterative termination condition is met, and then output the model parameters corresponding to the frog with the minimum fitness, which are used as the inversion optimal solution, and the polarization rates of each layer obtained are the extracted induced polarization information.

[0142] It should be understood that the implementation process of each module can refer to the content description of the foregoing method. The division of the above functional modules is only a division of logical functions. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. At the same time, the above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0143] Embodiment 3:

[0144] This embodiment provides a computer device, which includes: one or more processors, and a memory storing one or more computer programs; wherein, the processor calls the computer program to implement:

[0145] Steps of a method for extracting induced polarization information from transient electromagnetic data.

[0146] Specifically, the processor calls the computer program to implement:

[0147] Step 1: Obtain the parameter search range of the geoelectric model corresponding to the detection area, and form the initial frog population of the shuffled frog leaping algorithm based on the parameter search range;

[0148] Step 2: Convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer.

[0149] Step 3: Substitute the complex resistivity of each layer corresponding to each frog into the transient electromagnetic TEM forward calculation to obtain the vertical magnetic field corresponding to each frog.

[0150] Step 4: Calculate the fitness of each frog in the frog population, and the fitness is constructed based on the error between the forward vertical magnetic field and the vertical magnetic field observation data of the detection area.

[0151] Step 5: Perform iterative search operations of the grouping operator, local search, and global shuffling of the shuffled frog leaping algorithm with minimizing the fitness as the optimization goal until the iterative termination condition is met, then output the model parameters corresponding to the frog with the minimum fitness, and use the model parameters as the inversion optimal solution, and the polarizability of each layer obtained is the extracted induced polarization information.

[0152] For the specific implementation process of each step, please refer to the description of the foregoing method.

[0153] Among them, the memory may include high-speed RAM memory, and may also include non-volatile defibrillators, such as at least one disk memory.

[0154] If the memory and the processor are implemented independently, the memory, the processor, and the communication interface can be interconnected through a bus and communicate with each other. The bus can be an Industry Standard Architecture bus, a Peripheral Component Interconnect bus, or an Extended Industry Standard Architecture bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0155] Optionally, in a specific implementation, if the memory and the processor are integrated on a single chip, the memory and the processor can communicate with each other through an internal interface.

[0156] It should be understood that in the embodiments of the present invention, the so-called processor may be a Central Processing Unit (CPU), and this processor may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc. The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0157] Embodiment 4: The present invention provides a computer-readable storage medium that stores a computer program, and the computer program is called by a processor to implement: [[ID=))

[0158] Steps of a method for extracting induced polarization information from transient electromagnetic data.

[0159] Specifically, the computer program is called by a processor to implement:

[0160] Step 1: Obtain the parameter search range of the geoelectric model corresponding to the detection area, and form an initial frog population of the shuffled frog leaping algorithm based on the parameter search range;

[0161] Step 2: Convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer.

[0162] Step 3: Substitute the complex resistivity of each layer corresponding to each frog into the transient electromagnetic TEM forward calculation to obtain the vertical magnetic field corresponding to each frog.

[0163] Step 4: Calculate the fitness of each frog in the frog population, where the fitness is constructed based on the error between the forward vertical magnetic field and the vertical magnetic field observation data of the detection area.

[0164] Step 5: Perform iterative search operations of the grouping operator, local search, and global shuffling of the shuffled frog leaping algorithm with the goal of minimizing fitness until the iterative termination condition is met, then output the model parameters corresponding to the frog with the minimum fitness, and use the model parameters as the inversion optimal solution. The polarizability of each layer obtained is the extracted induced polarization information.

[0165] For the specific implementation process of each step, please refer to the description of the foregoing method.

[0166] The readable storage medium is a computer-readable storage medium, which can be the internal storage unit of the controller described in any of the foregoing embodiments, such as the hard disk or memory of the controller. The readable storage medium can also be an external storage device of the controller, such as a plug-in hard disk equipped on the controller, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the readable storage medium can also include both the internal storage unit and the external storage device of the controller. The readable storage medium is used to store the computer program and other programs and data required by the controller. The readable storage medium can also be used to temporarily store the data that has been output or will be output.

[0167] Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing readable storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs and other media that can store program codes.

[0168] Simulation experiment:

[0169] Experiment 1: (For comparing the inversion effects of two induced polarization models)

[0170] Taking a three-layer type-A geoelectric model as an example, the inversion performance of the method described in the present invention under different polarization models is studied. Among them, the geoelectric model parameters are set as ρ1 = 50 Ω·m, ρ2 = 100 Ω·m, ρ3 = 200 Ω·m, h1 = 100 m, h2 = 200 m; the polarizability parameter m2 = 0.3 (only the middle layer is polarized, and the other layers are 0). The basic parameters of the improved shuffled frog leaping algorithm are set as the population size P = 20, the number of meme groups k = 4, the number of frogs in each meme group n = 5, the number of iterations iter l = 10, iter g = 20.

[0171] Table 1 gives the evaluation indexes of the inversion results when the polarized layer is in the middle layer and the Dias model and the Cole-Cole model are adopted in the present invention. The evaluation index results show that the inversion result of the Dias model is better than that of the Cole-Cole model. In addition, Figure 3 the inversion results of the method described in the present invention under the Dias and Cole-Cole models, the fitting results of the vertical magnetic field H z curve and the iteration situation of the proposed algorithm are further given. From Figure 3 it can be seen that for the three-layer type-A geoelectric model, compared with the Cole-Cole model, the Dias model can more accurately realize the inversion extraction of resistivity, layer thickness and polarizability parameters, the vertical magnetic field curve has a higher fitting degree with the observed data, and the algorithm iteration convergence effect is also better. It shows that the TEM nonlinear inversion adopting the Dias model in the present invention is more conducive to reflecting the actual geoelectric model situation containing IP induced polarization information.

[0172] Table 1 Inversion results of induced polarization information extraction for the three-layer type-A geoelectric model

[0173]

[0174] Experiment 2:

[0175] To further verify the effectiveness of the algorithm, a four-layer KH-type geoelectric model was designed. The inversion performance comparison of the improved SFLA method (WTSFLA) of the present invention with the particle swarm optimization PSO algorithm, DE algorithm (Differential Evolution), ABC algorithm (Artificial Bee Colony algorithm), and standard SFLA algorithm was studied. Among them, the geoelectric model parameters were set as ρ1 = 100 Ω·m, ρ2 = 300 Ω·m, ρ3 = 50 Ω·m, ρ4 = 100 Ω·m, h1 = 100 m, h2 = 200 m, h3 = 300 m; the polarizability parameters m2 = m3 = 0.3 (only the second and third layers are polarized), and other parameters are the same as above. For easy comparison, the parameters of different non-linear inversion algorithms were kept as consistent as possible, and other specific parameter settings are shown in Table 2.

[0176] Table 3 gives the performance and calculation time of different non-linear inversion algorithms when using the Dias model and the middle layer is polarized. It can be seen from Table 3 that compared with the performance of the PSO algorithm, DE algorithm, and ABC algorithm, the SFLA-type algorithms (standard SFLA and improved WTSFLA) achieve better results. This is because the inversion process of induced polarization information extraction places higher requirements on the global search ability of non-linear inversion algorithms. The SFLA-type algorithms based on meme group evolution are superior to the PSO algorithm, DE algorithm, and ABC algorithm based on individual evolution in terms of global search performance. The inversion performance of the improved WTSFLA algorithm is significantly better than that of the standard SFLA algorithm. This is because WTSFLA uses the Tent chaotic distribution operator to generate the initial population, randomly increases the search range of the frogs, and enhances the optimization ability of the algorithm; then the adaptive moving step size is applied to the search process to control the relationship between local development and global exploration. These two measures enhance the global search ability, improve the convergence performance in the search optimization of IP information, and largely avoid premature convergence. At the same time, the introduction of these two update strategies also increases the calculation time of the algorithm. In addition, Figure 4 the inversion results of the algorithm in the Dias model and the fitting results of the vertical magnetic field H z curve and the algorithm iteration situation are further given. It can be Figure 4 seen that for the four-layer KH-type geoelectric model, the method of the present invention can accurately realize the inversion extraction of resistivity, layer thickness, and polarizability parameters. The calculated vertical magnetic field data and the observed data have the highest fitting degree, and WTSFLA also has a greater convergence speed and a lower fitness value. It shows that the hybrid frog leaping non-linear inversion based on the comprehensive improvement of Tent chaotic distribution - adaptive moving factor proposed by the present invention has more superior global search ability and higher inversion accuracy, and is suitable for extracting induced polarization information.

[0177] Table 2 Simulation parameter settings of the four-layer KH-type geoelectric model

[0178]

[0179] Table 3 Inversion performance of different non-linear inversion methods

[0180]

[0181] Experiment 3:

[0182] Since the data collected in the field all contain a certain amount of noise, taking the five-layer HKH-type geoelectric model (Dias model) as an example, this invention studies the inversion stability and robustness of the algorithm proposed in this paper under different noise conditions. Among them, the geoelectric model parameters are set as ρ1 = 300 Ω·m, ρ2 = 50 Ω·m, ρ3 = 500 Ω·m, ρ4 = 50 Ω·m, ρ5 = 1000 Ω·m; h1 = 100 m, h2 = 200 m, h3 = 300 m, h4 = 500 m; the polarization rate parameters m3 = m4 = 0.3 (set the polarization layer as the 3rd - 4th layer), and other parameters are the same as above.

[0183] Table 4 shows the inversion performance of the algorithm in this paper when different degrees of noise are added with the Dias model and polarization in the middle layer. Figure 5 The inversion results of the algorithm in this paper under different degrees of noise and the fitting results of the vertical magnetic field H z curve and the algorithm iteration situation are given. From Figure 5 of (a) and (b), it can be seen that in the case of no noise, the algorithm of this invention accurately reconstructs the resistivity parameters, layer thickness parameters, and polarization rate parameters of the HKH-type geoelectric model. Under the interference of 10% and 20% random noise, the inversion algorithm can also well invert the resistivity parameters, layer thickness parameters, and polarization rate parameters of the model, but the degree of coincidence of the inversion results for the model parameters decreases slightly, and the high-resistance layer parameters and the middle layer polarization rate deviate greatly, but basically the change form of the model parameters can be inverted. From Figure 5 of (c) and (d) in the vertical magnetic field H z the results of data fitting and the algorithm fitness curve, although the noise increases, the inversion results are not much different, which reflects the stability of the proposed inversion algorithm. It can be seen from Table 4 that as the noise increases, the inversion errors MSE, APE, and TRE gradually increase, but generally speaking, the algorithm in this paper ensures a low inversion error under the condition of adding noise, because the Tent chaotic distribution and the adaptive moving factor enhance the robustness of the algorithm. Finally, the calculation time for the inversion of the five-layer model is increased compared with the inversion of the three-layer and four-layer models, which is mainly because for the inversion of a more complex geoelectric model structure, the population dimension of the algorithm and the forward calculation amount are increased.

[0184] Table 4 Inversion results of induced polarization information extraction for the five-layer noisy geoelectric model

[0185]

[0186] The simulation results of the theoretical model show that the inversion results of the method described in the present invention are stable, can reconstruct the geoelectric structure and extract induced polarization information well, and have strong robustness in a noisy environment. Compared with other non-linear inversion algorithms, the method described in the present invention has superior convergence performance and inversion accuracy, and can effectively extract the induced polarization information in the TEM signal. At the same time, the results of 3 geoelectric model test examples show that the proposed method can reconstruct the underground model parameters with relatively high resolution, which is of great significance for prospecting or finding groundwater, etc.

[0187] It should be emphasized that the examples described in the present invention are illustrative rather than restrictive. Therefore, the present invention is not limited to the examples described in the specific embodiments. Any other embodiments obtained by those skilled in the art according to the technical solution of the present invention, whether modified or replaced, as long as they do not depart from the purpose and scope of the present invention, also belong to the protection scope of the present invention.

Claims

1. A method for extracting induced polarization information from transient electromagnetic data, characterized in that: It includes the following steps: Step 1: Obtain the parameter search range of the geoelectric model corresponding to the detection area, and form an initial frog population of the shuffled frog leaping algorithm based on the parameter search range; Among them, the model parameters of the geoelectric model include: the resistivity, thickness, and polarizability corresponding to each layer of the earth medium in the geoelectric model, and each frog represents a set of model parameters of the geoelectric model: Step 2: Convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer; Step 3: Substitute the complex resistivity of each layer corresponding to each frog into the transient electromagnetic TEM forward calculation to obtain the vertical magnetic field corresponding to each frog; Step 4: Calculate the fitness of each frog in the frog population, and the fitness is constructed based on the error between the forward vertical magnetic field and the vertical magnetic field observation data of the detection area; Step 5: Perform an iterative search operation of the shuffled frog leaping algorithm with minimizing the fitness as the optimization goal until the iterative termination condition is met, then output the model parameters corresponding to the frog with the minimum fitness, and use the model parameters as the inversion optimal solution, and the polarizabilities of each layer obtained are the extracted induced polarization information.

2. The extraction method according to claim 1, wherein: In the shuffled frog leaping algorithm, the frog population is divided into k meme groups through a grouping operator, and then local search is performed on the frog with the worst fitness in each meme group; Among them, the local search is performed according to the combination of formula 1 and formula 2 or the combination of formula 3 and formula 2 for search and update; Formula 1: Formula 2: X w (t + 1) = X w (t) + Δ w (t + 1), S min ≤ Δ w (t) ≤ S max Formula 3: where Δ w (t + 1) is the moving step size, i represents the current global iteration number, t is the local search number of the current meme group, iter g is the preset maximum global iteration number, X b (t), X w (t) represent the frogs with the best fitness and the worst fitness in the current meme group at the t-th local search respectively; S min and S max are the minimum boundary and the maximum boundary of the distance range allowed for the frogs to move; X g (t) represents the frog with the best fitness in the frog population in the current meme group at the t-th local search; Among them, if the fitness of the frog updated by the combination of formula 1 and formula 2 or the combination of formula 3 and formula 2 is not better than the fitness before update, a new frog is randomly generated to replace the original frog and then local search is performed.

3. The extraction method according to claim 1, wherein: The initial frog population is constructed based on the Tent mapping, and the formula of the Tent mapping is as follows: wherein, the chaotic sequence value x n has a distribution range x n ∈ [0, 1], and a is a control parameter; Among them, according to the mapping relationship between the parameter search range and the distribution range of the chaotic sequence values, the model parameter values of the initially generated geoelectric model randomly within the parameter search range are mapped to the chaotic sequence value x n ; Then update the chaotic sequence value \(x\) according to the Tent mapping n to the chaotic sequence value \(x\) n+1 ; According to the mapping relationship between the parameter search range and the distribution range of the chaotic sequence values, the chaotic sequence value x n+1 is converted into the model parameters of the geoelectric model, and the model parameters of the initial geoelectric model are updated, thereby forming the initial frog population.

4. The extraction method according to claim 1, characterized in that: The execution process of Step 5 is as follows: Execute the grouping operator to divide the frogs in the frog population into k meme groups; Execute local search, and perform local search on the frog with the worst fitness in each meme group respectively. If the fitness of the frog updated by the local search formula is not better than the fitness before update, a new frog is randomly generated to replace the original frog and then local search is performed; Execute global shuffling. After all meme groups complete local search, perform global shuffling on all frogs in the frog population, and then re-execute the grouping operator and local search; Among them, repeat the operations of the grouping operator, local search, and global shuffling until the global search times reach the preset global maximum iteration times, and then output the model parameters corresponding to the frog with the minimum fitness as the inversion optimal solution.

5. The extraction method according to claim 4, wherein: When executing the grouping operator, first arrange the frogs in the frog population in descending order of fitness; then put the first frog into the first meme group, the second frog into the second meme group, until the kth frog is put into the kth meme group; then the k + 1th frog is put into the first meme group again, the k + 2th frog is put into the second meme group, and so on, and the frogs in the frog population are divided into k meme groups.

6. The extraction method according to claim 1, characterized in that: The formula for the fitness is: where Fitness is the fitness, N is the time sampling point, represents the vertical magnetic field observation data at the i-th time sampling point; represents the geoelectric model determined based on the pattern parameters corresponding to the frog, and then the vertical magnetic field at the i-th time sampling point obtained through forward calculation.

7. The extraction method according to claim 1, characterized in that: In step 2, the Dias induced polarization model or the Cole-Cole induced polarization model is used to convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer; Among them, the formula for the complex resistivity of the Dias induced polarization model is: The formula for the complex resistivity of the Cole-Cole induced polarization model is: And there is: μ = iωτ+(iωτ″) 1 / 2 , τ″ = τ 2 η 2 In the formula, ρ(ω) is the complex resistivity, ρ is the resistivity without considering the polarization effect, m is the polarization rate; τ is the time constant; η is the electrochemical parameter; δ is the polarization resistivity coefficient, τ′, τ″, μ are all intermediate variables, i is the complex number symbol, ω is the angular frequency, and c is the frequency-dependent coefficient.

8. A system based on the extraction method according to any one of claims 1-7, characterized in that: An initial frog population generation module, configured to obtain the parameter search range of the geoelectric model corresponding to the detection area, and form an initial frog population of the shuffled frog leaping algorithm based on the parameter search range; Among them, the parameters of the geoelectric model include: the resistivity, thickness, and polarization rate of each layer of the earth medium in the geoelectric model, and each frog represents a set of parameters of the geoelectric model: A complex resistivity calculation module, configured to convert the resistivity of each layer corresponding to each frog in the frog population into the complex resistivity of each layer; A forward modeling module, configured to substitute the complex resistivity of each layer corresponding to each frog into the transient electromagnetic TEM forward modeling to obtain the vertical magnetic field corresponding to each frog; A fitness calculation module, configured to calculate the fitness of each frog in the frog population, and the fitness is constructed based on the error between the forward modeled vertical magnetic field and the vertical magnetic field observation data of the detection area; An extraction module, configured to perform the grouping operator, local search, and global shuffling of the shuffled frog leaping algorithm with minimizing the fitness as the optimization goal, until the iteration termination condition is met, and then output the model parameters corresponding to the frog with the minimum fitness, and use it as the inversion optimal solution, and the polarization rates of each layer obtained are the extracted induced polarization information.

9. A computer device, characterized in that: Including: One or more processors; A memory storing one or more computer programs; Among them, the processor calls the computer program to implement: The steps of the method for extracting induced polarization information from transient electromagnetic data according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: A computer program is stored, and the computer program is called by the processor to implement: The steps of the method for extracting induced polarization information from transient electromagnetic data according to any one of claims 1-7.

Images