A natural electric field data inversion method and system based on a white blood cell optimization algorithm

Abstract

The application provides a natural electric field data inversion method and system based on a white blood cell optimization algorithm, relates to the technical field of geophysical exploration and data inversion, and the white blood cell optimization algorithm is used for simulating the biological immune process of white blood cell identification, chemotaxis, phagocytosis and pathogen elimination, the solution of the optimization problem of the natural electric field data inversion is regarded as a pathogen or an abnormal cell in an immune system, the task of the algorithm main body is to efficiently search, surround and finally remove the targets, the algorithm has a dynamic balance search strategy and multi-mechanism cooperation, can obtain efficient and reliable inversion results, has better balance exploration and development capacity, can adaptively jump out of a local optimal solution to obtain a global optimal solution, can guarantee sufficient exploration of the whole solution space, meanwhile, the good development capacity can guarantee accurate search in a local area, so that accurate inversion results are obtained, and the precision of inversion interpretation is improved.

Description

Technical Field

[0001] This invention relates to the fields of geophysical exploration and data inversion technology, and in particular to a method and system for inverting natural electric field data based on a leukocyte optimization algorithm. Background Technology

[0002] The natural electric field method boasts significant advantages, including high cost-effectiveness, sensitivity to targets, and environmental friendliness, making it widely used in environmental and hydrogeological fields. It is extremely sensitive to groundwater flow; by measuring the natural potential, the groundwater flow field can be delineated, aiding in the clarification of hydrogeological conditions. It is also sensitive to the redox reactions and ion migration of pollutants, enabling rapid location of leak channels and delineation of contamination areas. Therefore, it is highly suitable for large-scale surveys and rapid delineation of anomaly zones for targets such as leaks or contamination.

[0003] However, the interpretation of natural electric field data inversion still faces many challenges. Various underground electrical structures may generate similar spontaneous potential anomalies on the surface. Natural electric field signals are typically weak, easily affected by interference, and have low signal-to-noise ratios. Furthermore, the origins of natural electric fields are diverse; the geometry, spatial distribution, and intensity of the field sources are unknown, increasing the difficulty of forward modeling. The relationship between underground field source distribution and surface spontaneous potential is highly nonlinear and often lacks reliable prior information and constraints. Traditional linear inversion methods have limited effectiveness, while nonlinear inversion is computationally expensive and prone to getting trapped in local extrema.

[0004] Therefore, there is an urgent need for a more efficient, accurate, and generalizable method and system for inverting natural electric field data based on the leukocyte optimization algorithm to overcome the above problems. Summary of the Invention

[0005] To address the aforementioned technical problems in related technologies, this invention proposes a method and system for inverting natural electric field data based on a leukocyte optimization algorithm.

[0006] In a first aspect, the present invention provides a method for inverting natural electric field data based on a leukocyte optimization algorithm, comprising the following steps:

[0007] S1. Obtain the observation data of the natural electric field measurement points to be inverted, and set the inversion search parameters of the white blood cell optimization algorithm;

[0008] S2. Initialize the inversion search space and the white blood cell population and pathogen population of the white blood cell optimization algorithm according to the inversion search parameters;

[0009] S3. Input the natural electric field measurement point observation data into the leukocyte optimization algorithm for inversion iteration to obtain model parameters; the inversion iteration includes leukocyte chemotaxis and recruitment, leukocyte recognition and adhesion of selected pathogens, leukocyte phagocytosis and internalization of selected pathogens, leukocyte degranulation, leukocyte apoptosis and population chemotactic factor update.

[0010] S4. Perform forward modeling calculation on the theoretical natural electric field anomaly response based on the model parameters. If the fitting accuracy between the theoretical natural electric field anomaly response and the observation data of the natural electric field measuring point meets the preset fitting accuracy, then output the model parameters obtained in the current inversion iteration as the final inversion result. Otherwise, adjust the white blood cell optimization algorithm parameters and return to step S3 to continue the inversion iteration until the preset fitting accuracy is met.

[0011] S5. Based on the final inversion results, determine the specific location of the natural electric field anomaly target and obtain relevant information about the target's shape.

[0012] Specifically, the inversion search parameters include the number of white blood cells (nWBC), the number of pathogens (nPathogens), the maximum number of iterations (maxIter), chemokine sensitivity, phagocytic radius, degranulation probability (degranulateProb), the number of white blood cell apoptosis triggers (netosisThresh), the restart probability (restartProb), geoelectric model constraints, the location of white blood cells, the location of pathogens, the resultant force on each white blood cell (total_force), and the current iteration number (iter). The chemokine sensitivity includes the chemokine step size factor alpha for white blood cell movement and the stochastic exploration factor beta for white blood cell movement. The geoelectric model constraints include the geoelectric model parameter dimension dim, the lower bound of the geoelectric model parameter search space lb, and the upper bound of the geoelectric model parameter search space ub.

[0013] Specifically, step S2 includes:

[0014] Based on the white blood cell population size nWBC in the inversion search parameters, the positions of each white blood cell WBCs are randomly generated in the inversion search space to form a white blood cell population WBC_pop.

[0015] Based on the pathogen population size nPathogens in the inversion search parameters, a pathogen population Pathogen_pop is formed by randomly generating or customizing the location Pathogens of each pathogen in the inversion search space.

[0016] The inversion search space is defined by the lower bound lb of the geoelectric model parameter search space and the upper bound ub of the geoelectric model parameter search space.

[0017] Specifically, the inversion iteration in step S3 includes:

[0018] S31, Leukocyte Chemotaxis and Recruitment: The position of leukocytes is updated by calculating the chemokine concentration obtained from the fit difference of pathogens and the average distance between each leukocyte and all pathogens in the solution space.

[0019] S32. Leukocytes recognize and adhere to the selected pathogens: Each leukocyte selects the pathogen closest to it as the selected pathogen. If there are multiple equally spaced nearest pathogens, the selection factor of each pathogen by the leukocyte is calculated, and the pathogen with the smallest factor is selected as the leukocyte's selected pathogen.

[0020] S33, phagocytosis and internalization of selected pathogens by leukocytes: leukocytes are combined with their selected pathogens to produce a new pathogen;

[0021] S34, Leukocyte degranulation: Leukocytes randomly generate multiple new leukocytes in the vicinity based on the degranulation probability. Then, the fit difference of all new leukocytes is evaluated, and the new leukocyte with the smallest fit difference is retained to replace the original leukocyte.

[0022] S35, Leukocyte Autoapoptosis: If leukocytes are not updated after the number of leukocyte autoapoptosis triggers, multiple new leukocytes are randomly generated in the inversion search space according to the restart probability, and the new leukocytes with the smallest fitting difference are retained to replace the original leukocytes.

[0023] S36, Population Chemotaxis Factor Update: If the current iteration number has not reached the maximum iteration number, jump to step S31 to continue the next iteration; if the maximum iteration number has been reached, terminate the iteration and output the pathogen location with the smallest fit difference as the model parameter.

[0024] Specifically, step S31 includes the following steps:

[0025] S311. First, calculate the concentration of the pathogen's chemokine, chemo_conc:

[0026] chemo_conc=1 / (1+exp(-(fitness_Pathogens-mean(fitness_Pathogens)) / std(fitness_Pathogens))),

[0027] Where fitness_Pathogens is the fit difference for each pathogen, mean(fitness_Pathogens) is the mean of the fit differences for all pathogens, std(fitness_Pathogens) is the standard deviation of the fit differences for each pathogen, and exp() is the natural exponential function.

[0028] S312. Calculate the average distance avg_distance between each white blood cell and all pathogens in the solution space, and then update the resultant force total_force on each white blood cell according to the chemokinase concentration chemo_conc of the pathogens and the average distance avg_distance: total_force = total_force + chemo_conc / avg_distance;

[0029] S313. Update the position (WBCs) of each white blood cell based on the resultant force (total_force) acting on each white blood cell, the current iteration number (iter), and the dimension of the geoelectric model parameter (dim):

[0030] WBCs = WBCs + alpha * total_force + beta * (1 - iter / maxIter) * (rand(1, dim) - 0.5) * (ub - lb),

[0031] Here, rand() is a random function, used in this formula to generate a random number between [1, dim].

[0032] Specifically, the formula for the fit difference is as follows:

[0033] Fitness= ,

[0034] Where Fitness represents the fit difference; obdata i caldata represents the actual observed value at the i-th measuring point along the survey line. i nd represents the theoretical response value of the i-th measurement point calculated using the analytical formula or numerical calculation method of the natural electric field, where nd is the number of measurement points.

[0035] Specifically, step S33 includes:

[0036] The white blood cells are merged with the selected pathogens sel_Pathogens to generate a new pathogen new_Pathogens, as shown in the following formula:

[0037] new_Pathogens=0.7 * WBCs + 0.3 * sel_Pathogens,

[0038] Then, the fitting difference of the new pathogens new_Pathogens is calculated. If the fitting difference of the new pathogens new_Pathogens is less than the fitting difference of the selected pathogens sel_Pathogens, the new pathogens new_Pathogens replace the selected pathogens sel_Pathogens; otherwise, the selected pathogens sel_Pathogens are retained. The original leukocytes proceed to the next step of degranulation.

[0039] Specifically, step S34 includes:

[0040] After leukocytes merge with pathogens to produce new individuals, multiple new leukocytes are randomly generated nearby according to the set degranulation probability (degranulateProb). (New_solutions)

[0041] Generate a random number Random between [0,1]. If Random < degranulateProb, then:

[0042] New_solutions = WBCs + Random2*(ub-lb) / maxIter,

[0043] Where Random2 is a random number between [0,1]; New_solutions represents a randomly generated new white blood cell;

[0044] Then, evaluate the fit difference of all new leukocytes and retain the new leukocytes with the smallest fit difference to replace the original leukocytes; if Random>= degranulateProb, then retain the original leukocytes.

[0045] Specifically, in step S35, multiple new white blood cells are randomly generated again within the inversion search space according to the restart probability restartProb, including:

[0046] Generate a random number Random3 between [0,1]. If Random3 < restartProb, then:

[0047] New_solutions2 = WBCs + Random4* (ub-lb);

[0048] Where Random4 is a random number between [0,1]; New_solutions2 represents a randomly generated new white blood cell; restartProb is the restart probability; then the fit difference of all new white blood cells is evaluated, and the new white blood cell with the smallest fit difference is retained to replace the original white blood cell; if Rondom3>= restartProb, then the original white blood cell is retained.

[0049] Secondly, the present invention provides a natural electric field data inversion system based on a leukocyte optimization algorithm, which, based on the natural electric field data inversion method based on the leukocyte optimization algorithm described in the first aspect above, includes the following modules:

[0050] The data acquisition module is used to acquire observation data of the natural electric field measurement points to be inverted and to set the inversion search parameters of the white blood cell optimization algorithm;

[0051] An initialization module is used to initialize the inversion search space and the white blood cell population and pathogen population of the white blood cell optimization algorithm according to the inversion search parameters.

[0052] The inversion iteration module is used to input the natural electric field measurement point observation data into the leukocyte optimization algorithm for inversion iteration to obtain model parameters; the inversion iteration includes leukocyte chemotaxis and recruitment, leukocyte recognition and adhesion of selected pathogens, leukocyte phagocytosis and internalization of selected pathogens, leukocyte degranulation, leukocyte apoptosis and population chemotactic factor update;

[0053] The accuracy verification module is used to perform forward modeling calculations on the theoretical natural electric field anomaly response based on the model parameters. If the fitting accuracy between the theoretical natural electric field anomaly response and the observation data of the natural electric field measuring point meets the preset fitting accuracy, the model parameters obtained in the current inversion iteration are output as the final inversion result. Otherwise, the white blood cell optimization algorithm parameters are adjusted and returned to the inversion iteration module to continue the inversion iteration until the preset fitting accuracy is met.

[0054] The prediction output module is used to determine the specific location of the natural electric field anomaly target and obtain relevant information about the target's shape based on the final inversion result.

[0055] The natural electric field data inversion method based on the leukocyte optimization algorithm provided in this invention has a better balance between exploration and exploitation capabilities compared with conventional methods, and can adaptively escape local optima to obtain the global optimum. When used for natural electric field data inversion, the good exploration capability can ensure that the entire solution space is fully explored, while the good exploitation capability can ensure accurate local searches to obtain accurate inversion results and improve the accuracy of inversion interpretation.

[0056] Natural electric field data is susceptible to various noise interferences. Noise data increases the difficulty of inversion interpretation and may even lead to inversion failure. The leukocyte-optimized inversion algorithm, with its multi-mechanism synergistic strategy that simulates the process of biological leukocytes engulfing pathogens, has greatly improved its resistance to noise interference and improved the reliability of inversion interpretation.

[0057] The natural electric field data inversion method and system based on the leukocyte optimization algorithm provided by this invention can still accurately find search results even with a large search space, and can largely overcome the problem that conventional inversion methods rely heavily on prior information.

[0058] This embodiment provides a natural electric field data inversion method based on a leukocyte optimization algorithm, which has the following beneficial effects:

[0059] (1) Dynamic role switching: White blood cells simultaneously undertake the tasks of exploration (chemotaxis) and development (phagocytosis), and dynamically adjust according to the environment;

[0060] (2) Multi-mechanism collaboration: It integrates random walk, gradient guidance, local search and global restart, and has good balance;

[0061] (3) Adaptability: The NETosis mechanism is used to adaptively handle stagnation problems, which is highly robust;

[0062] (4) Parallelism: Multiple white blood cells can work independently and in parallel, which is suitable for distributed computing;

[0063] (5) It can perform the inversion of natural electric field data without prior constraints and has good noise resistance. Attached Figure Description

[0064] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0065] Figure 1 A schematic diagram of a natural electric field data inversion method based on a leukocyte optimization algorithm provided in an embodiment of the present invention;

[0066] Figure 2 This is a schematic diagram of the inversion and fitting results of natural electric field data with different levels of noise provided in an embodiment of the present invention; Figure 2 The circles in (a) indicate that the observed data contains 10% noise. Figure 2 The circles in (b) indicate that the observed data contains 20% noise;

[0067] Figure 3 This is a schematic diagram of dam seepage detection provided in an embodiment of the present invention;

[0068] Figure 4 This is a schematic diagram of a natural electric field data inversion system based on a leukocyte optimization algorithm, provided as an embodiment of the present invention. Detailed Implementation

[0069] The present invention will be explained in detail through the following embodiments. The purpose of this invention is to protect all technical improvements within its scope. In the description of this invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0070] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0071] Example 1

[0072] refer to Figure 1 To address the need for inversion and interpretation of natural electric field data and the limitations of existing technologies, this embodiment provides a natural electric field data inversion method based on the leukocyte optimization algorithm to quickly and accurately invert natural electric field data, including the following steps:

[0073] S1. Obtain the observation data obdata of the natural electric field measurement points to be inverted, and set the inversion search parameters of the white blood cell optimization algorithm;

[0074] A polarized sphere exists within a uniform Earth. Its actual model parameters (electric dipole moment k = 20 C·m, ​​polarization angle θ = 60°, burial depth of the sphere's center h = 3 m, horizontal position of the sphere's center projected onto the Earth's surface x = 0 m) and the natural electric field observation data along a horizontal survey line on the Earth's surface above it are as follows: Figure 2 (a) Figure 2 The circled dots in (b) are shown.

[0075] In this embodiment, the observation data (obdata) of the natural electric field measurement points to be inverted is obtained through field observation.

[0076] This embodiment proposes a leukocyte optimization algorithm (LOA) that simulates the process by which leukocytes discover and engulf pathogens. LOA is a novel metaheuristic optimization algorithm inspired by the biological immune system. It searches for optimal solutions in a complex search space by simulating the biological immune process of leukocytes recognizing, chemotactically attracting, engulfing, and eliminating pathogens. The algorithm treats the solution to the problem to be optimized as a "pathogen" or "abnormal cell" in the immune system. The task of the algorithm's main body (leukocyte population) is to efficiently search, surround, and ultimately "clear" (optimize) these targets. Its core advantage lies in its dynamically balanced search strategy and multi-mechanism synergy. Applying this algorithm to the inversion of natural electric field data can yield efficient and reliable inversion results.

[0077] To achieve the above technical objectives, the technical solution adopted in this invention is as follows, and the mapping relationship between the white blood cell sterilization process and the algorithm is shown in Table 1.

[0078] Table 1. Mapping relationship between leukocyte sterilization process and algorithm

[0079] Fundraising and chemotaxis Global exploration Leukocytes are attracted by chemokines (target function gradient / pheromone) and move toward potential regions. Identification and Adhesion Evaluation and Selection The receptor identifies non-self targets through pattern recognition (fitness assessment) and decides whether to attack. Phagocytosis Local development Encircle and internalize the pathogen (perform a fine search near the current optimal solution). Degranulation / killing Solution update and optimization Release antibacterial substances (generate new solutions and disturb the current solution) to completely eliminate the target. NETosis Breaking out of local optima Release neutrophil extracellular traps (NETs) to capture pathogens on a large scale (global reboot / diversity introduction).

[0080] The inversion search parameters include leukocyte population size, pathogen population size, maximum number of iterations, chemokine sensitivity, phagocytic radius, degranulation probability, number of leukocyte apoptosis triggers, restart probability, geoelectric model constraints, leukocyte location, pathogen location, resultant force on each leukocyte, current iteration number, and maximum number of iterations.

[0081] The chemokine sensitivity includes the chemotactic step size factor for leukocyte migration and the stochastic exploration factor for leukocyte migration; the geoelectric model constraints include the geoelectric model parameter dimension, the lower bound of the geoelectric model parameter search space, and the upper bound of the geoelectric model parameter search space.

[0082] For example, initializing the parameters in the inversion search parameters:

[0083] Set the white blood cell population size nWBC=30;

[0084] Set the pathogen population size nPathogens = 30;

[0085] Set the maximum number of iterations for the algorithm to maxIter = 100;

[0086] The chemokine step size factor for leukocyte migration was set to alpha = 0.7;

[0087] Set the random exploration factor for white blood cell migration to beta = 0.5;

[0088] Set the degranulation probability degranulateProb = 0.3;

[0089] Set the number of leukocyte apoptosis (NETosis) triggers to netosisThresh = 10;

[0090] Set the restart probability to restartProb = 0.1;

[0091] Set the dimension of the geoelectric model parameter dim = 4;

[0092] Set the lower bound of the geoelectric model parameter search space to lb = [1, 1, 1, -50];

[0093] Set the upper bound of the geoelectric model parameter search space to ub = [50, 90, 10, 50];

[0094] Set the total force (total_force) for each white blood cell to 0;

[0095] Set the current iteration count to 0;

[0096] S2. Initialize the inversion search space and the white blood cell population and pathogen population of the white blood cell optimization algorithm according to the inversion search parameters;

[0097] The inversion search parameters in step S1 set the dimensions of the model parameters to be inverted and the search range of each dimension, thereby obtaining the inversion search space.

[0098] Based on the white blood cell population size nWBC in the inversion search parameters, the positions of each white blood cell WBCs are randomly generated in the inversion search space to form a white blood cell population WBC_pop.

[0099] Based on the pathogen population size nPathogens in the inversion search parameters, a pathogen population Pathogen_pop is formed by randomly generating or customizing the location Pathogens of each pathogen in the inversion search space.

[0100] The inversion search space is defined by the lower bound lb of the geoelectric model parameter search space and the upper bound ub of the geoelectric model parameter search space.

[0101] All individuals in the population must be traversed (such as calculating the fit difference of individuals), and the population size is the implicit number of iterations when traversing all individuals.

[0102] A white blood cell population (WBC_pop) is randomly generated within the feasible solution space of the inversion problem to be solved. Then, an initial "pathogen population" (i.e., the candidate solution set to be evaluated) (Pathogen_pop) is randomly generated or defined. The positions of white blood cells and pathogens in the solution space can be regarded as possible solutions. Their quality is determined by the fitting difference between the forward calculation data of the possible solutions they represent and the observation data to be inverted; the smaller the fitting difference, the better. The feasible solution space is the inversion search space; the dimension of the feasible solution space is the number of parameters in the inversion problem.

[0103] S3. Input the natural electric field measurement point observation data obdata into the leukocyte optimization algorithm for inversion iteration to obtain model parameters; the inversion iteration includes leukocyte chemotaxis and recruitment, leukocyte recognition and adhesion of selected pathogens, leukocyte phagocytosis and internalization of selected pathogens, leukocyte degranulation, leukocyte apoptosis and population chemotactic factor update.

[0104] Step S3 is an iterative process of inversion loop search, specifically including the following steps:

[0105] S31, Leukocyte Chemotaxis and Recruitment (Global Search): The position of leukocytes is updated by calculating the chemokine concentration obtained from the fit difference of pathogens and the average distance between each leukocyte and all pathogens in the solution space.

[0106] Each white blood cell calculates the attractive force (resultant force) based on the concentration of chemokines released by the current "pathogen" (related to the fit difference; the larger the fit difference, the higher the concentration may be, indicating that the problem is more urgent to solve); white blood cells randomly or deterministically move towards areas of high concentration, updating their positions;

[0107] S311. First, calculate the concentration of the pathogen's chemokine, chemo_conc:

[0108] chemo_conc=1 / (1+exp(-(fitness_Pathogens-mean(fitness_Pathogens)) / std(fitness_Pathogens))),

[0109] Where fitness_Pathogens is the fit difference for each pathogen, mean(fitness_Pathogens) is the mean of the fit differences for all pathogens, std(fitness_Pathogens) is the standard deviation of the fit differences for each pathogen, and exp() is the natural exponential function.

[0110] The fit difference value (fitness_Pathogens) for each pathogen is calculated according to the fit difference formula, which is shown in the following formula:

[0111] Fitness= ,

[0112] Where Fitness represents the fit difference; obdata i caldata represents the observation data of the i-th measuring point along the survey line, i.e., the actual observed value. i The calculated data for the i-th measurement point is obtained by forward modeling using the analytical formula or numerical calculation method of the natural electric field, i.e., the theoretical response value, where nd is the number of measurement points;

[0113] Pathogens and leukocytes can essentially be viewed as models within an inversion search space, with different model parameters at different locations. For a given model, its theoretical response values ​​(caldata) at the surface observation points can be obtained through forward modeling based on its model parameters. If a model calculates theoretical response values ​​(caldata) at a series of observation points... i Compared with actual observations (obdata) i If the fit difference between the two is very small, then the model can be considered to be the true model corresponding to the observed data.

[0114] The purpose of geophysical inversion is to calculate the corresponding geophysical model parameters based on actual observation data.

[0115] When calculating the fit difference for each pathogen, replace the actual observed values ​​at each measurement point in the fit difference formula. Replace the theoretical response values ​​of each pathogen in the fit difference formula. Then perform the calculation;

[0116] Similarly, when calculating the fit difference for each white blood cell count, the actual observed values ​​at each measurement point will replace the values ​​in the fit difference formula. Replace the theoretical response values ​​of each white blood cell in the fit difference formula. Then perform the calculation;

[0117] Understandably, there is only one set of actual observations. The purpose of the entire algorithm is to calculate and search for the model parameters that best match this set of actual observations (generally geophysical data recorded at multiple measuring points on the surface of the survey area, such as natural electric field values).

[0118] The analytical formula for the natural electric field is referenced in the following references: Zhu Xiaoxiong, Cui Yian, Li Xiyang, Tong Tiegang, Ji Tongxin. Inversion of natural electric field data based on particle swarm optimization [J]. Journal of Central South University (Natural Science Edition), 2015, 46(02): 579-585. and Zhang Pengfei, Cui Yi-an, Xie Jing, Guo Youjun, Liu Jianxin. Three-dimensional analytical solution of self-potential from regularly polarized bodies in a layered seafloor model [J]. Geoscientific Model Development, 2024, 17(23): 8521-8533.

[0119] The numerical calculation method is referenced in: Zhu Xiaoxiong, Cui Yian, Chen Zhixue. Inversion imaging of natural electric field sources based on least squares regularization [J]. Progress in Geophysics, 2016, 31(05):2313-2318.

[0120] Understandably, in geophysical exploration, observation points are generally arranged at equal intervals along a straight line, which is called a survey line. There are usually several discrete observation points on the survey line, which are called survey points. The survey points themselves have coordinate data, providing spatial location. The natural electric field value observed at the survey point is called the observation data of the survey point.

[0121] S312. Calculate the average distance avg_distance between each white blood cell and all pathogens in the solution space, and then update the total force total_force on each white blood cell based on the chemoconc concentration of the pathogens and the average distance avg_distance.

[0122] Calculate the average distance avg_distance between each white blood cell and all pathogens in the solution space, and define the resultant force total_force = total_force + chemo_conc / avg_distance;

[0123] S313. Update the position (WBCs) of each white blood cell based on the resultant force (total_force) acting on each white blood cell, the current iteration number (iter), and the dimension of the geoelectric model parameter (dim):

[0124] WBCs = WBCs + alpha * total_force + beta * (1 - iter / maxIter) * (rand(1, dim) - 0.5) * (ub - lb),

[0125] Where rand() is a random function, used in this formula to generate a random number between [1, dim];

[0126] The `alpha * total_force` part defines the concentration of the pathogen's chemokines based on the pathogen's fit difference and its distance from leukocytes. The concentration is directly proportional to the fit difference and inversely proportional to the distance, and is used to guide leukocytes to move towards areas with higher concentrations nearby. The `beta * (1 - iter / maxIter) * (rand(1, dim) - 0.5) * (ub - lb)` part is a random vector used to increase the randomness of leukocyte movement.

[0127] The values ​​of alpha and beta are used to balance the movement of leukocytes toward areas with high concentrations of pathogen chemokines and random movement. Generally, the values ​​are taken in the range of [1, 2] according to the actual search situation. If the value of alpha is greater than that of beta, the movement of leukocytes tends to be toward areas with high concentrations of chemokines, and vice versa, random movement is dominant.

[0128] S32. Leukocytes recognize and adhere to selected pathogens: Each leukocyte selects the pathogen closest to it as the selected pathogen sel_Pathogens. If there are multiple equidistant nearest pathogens, the selection factor score of each pathogen is calculated, and the pathogen with the smallest score is selected as the leukocyte's selected pathogen sel_Pathogens.

[0129] The pathogen selection factor scores are calculated according to the following formula:

[0130] scores=distances-0.5*fitness_Pathogens / max(fitness_Pathogens),

[0131] Where fitness_Pathogens represents the fit difference of each pathogen, and max(fitness_Pathogens) represents the maximum fit difference of each pathogen; the pathogen with the smallest score is selected, thus selecting pathogens that are close in distance and have a large fit difference; distances represents the distance of white blood cells or individual pathogens in the inversion search space.

[0132] S33. Leukocytes phagocytose and internalize selected pathogens: Leukocytes are merged with their selected pathogens (sel_Pathogens) to generate a new pathogen (new_Pathogens), as shown in the following formula:

[0133] new_Pathogens=0.7 * WBCs + 0.3 * sel_Pathogens,

[0134] Then, the fitting difference of the new pathogens new_Pathogens is calculated. If the fitting difference of the new pathogens new_Pathogens is better than (less than) the fitting difference of the selected pathogens sel_Pathogens, the new pathogens new_Pathogens replace the selected pathogens sel_Pathogens; otherwise, the selected pathogens sel_Pathogens are retained. The original leukocytes proceed to the next step of degranulation.

[0135] It is understood that in this embodiment, pathogens or white blood cells are characterized by their location. In this embodiment, the generation of new pathogens or white blood cells means that pathogens or white blood cells are generated at a certain location.

[0136] S34, Leukocyte degranulation: Leukocytes randomly generate multiple new leukocytes in the vicinity based on the degranulation probability. Then, the fit difference of all new leukocytes is evaluated, and the new leukocytes with the best fit difference are retained to replace the original leukocytes.

[0137] After leukocytes merge with pathogens to create new individuals, they randomly generate multiple new leukocytes (New_solutions) in the vicinity according to the set degranulateProb probability (leukocytes "release antimicrobial granules" at their current location (performing local perturbation)).

[0138] The specific process of randomly generating multiple new white blood cells (New_solutions) is as follows:

[0139] Generate a random number Random between [0,1]. If Random < degranulateProb, then:

[0140] New_solutions = WBCs + Random2*(ub-lb) / maxIter,

[0141] Where Random2 is a random number between [0,1]; New_solutions represents a randomly generated new white blood cell;

[0142] Then, the fit difference of all new leukocytes is evaluated, and the best individual (with the smallest fit difference) is retained to replace the original leukocytes;

[0143] If Random >= degranulateProb, then the original leukocytes are retained.

[0144] S35, Leukocyte autoapoptosis (population diversity maintenance and local optimum escape): If the number of leukocyte autoapoptosis triggers exceeds the number of times netosisThresh has not been updated, multiple new leukocytes New_solutions2 will be randomly generated in the inversion search space according to the restart probability restartProb.

[0145] The specific process is as follows:

[0146] Generate a random number Random3 between [0,1]. If Random3 < restartProb, then:

[0147] New_solutions2 = WBCs + Random4* (ub-lb),

[0148] Random4 is a random number between [0,1]; New_solutions2 represents a randomly generated new white blood cell;

[0149] The search space is the range formed by the lower bound lb and the upper bound ub;

[0150] Then, the fit difference of all new leukocytes is evaluated, and the best individual (with the smallest fit difference) is retained to replace the original leukocytes;

[0151] If Rondom3>= restartProb, then retain the original white blood cells.

[0152] If white blood cells fail to improve their target for a long time (several iterations), or if the population diversity is below the threshold, "NETosis" (white blood cell self-apoptosis) is triggered. The white blood cell disintegrates, releasing a large number of "DNA networks" (randomly generating a batch of new solutions in the search space), and regenerates itself at a random location. This mechanism can effectively escape local optima.

[0153] S36. Population chemotactic factor update:

[0154] If the current iteration number iter has not reached the maximum iteration number maxIter, jump to step S31 to continue the next round of iteration; if the maximum iteration number maxIter has been reached, terminate the iteration and output the optimal (minimum fit difference) pathogen location as the inversion result, i.e., the model parameters.

[0155] If the current iteration number iter has not reached the maximum iteration number maxIter, update the chemokine field in the whole space based on the distribution and quality (fitness) of the "pathogen" after this round of iteration, guide the recruitment of white blood cells in the next round, and jump to step S31 to continue the next round of iteration;

[0156] S4. Perform forward modeling calculation on the theoretical natural electric field anomaly response based on the model parameters. If the fitting accuracy between the theoretical natural electric field anomaly response and the natural electric field measurement point observation data obdata meets the preset fitting accuracy, then output the model parameters obtained in the current inversion iteration as the final inversion result. Otherwise, adjust the white blood cell optimization algorithm parameters and return to step S3 to continue the inversion iteration until the preset fitting accuracy is met.

[0157] The specific conditions for determining whether the preset fitting accuracy is met are as follows:

[0158] The fitting accuracy is calculated by using the fitting difference formula to determine the fitting difference between the theoretical natural electric field anomaly response and the observed data obdata from the natural electric field measuring point. If the fitting accuracy is ≤1%, it is considered to meet the preset fitting accuracy; otherwise, it is considered not to meet the preset fitting accuracy.

[0159] In geophysical measurement data inversion, a fitting error of less than or equal to 1% is generally considered to meet the fitting accuracy requirement, i.e., the preset fitting accuracy is 1%, but it can be modified according to actual needs.

[0160] S5. Based on the final inversion results, determine the specific location of the natural electric field anomaly target (such as the seepage point of the dam) and obtain relevant information such as the target shape (such as the seepage channel situation).

[0161] The output of the leukocyte optimization algorithm's inversion loop search iteration is used as model parameters. The theoretical natural electric field anomalous response is then calculated using the analytical formula for the natural electric field or numerical calculation methods. The fit between the calculated parameters and the observed data is compared. If the fit meets the accuracy requirements, the model parameters are output as the final inversion result. If the fit does not meet the requirements, the leukocyte optimization algorithm parameters are adjusted, and the inversion loop search iteration continues until the fit meets the accuracy requirements, at which point the result is output.

[0162] The final output model parameters are: electric dipole moment k = 20.2 C·m, ​​polarization angle θ = 60.5°, burial depth of the sphere center h = 3.0 m, and horizontal position of the sphere center projected onto the ground x = 0.0 m.

[0163] Based on the example data and algorithm inversion fitting results, as follows: Figure 2 As shown. Figure 2 In (a), the circled dots represent the observed data containing 10% noise, the solid line represents the response curve calculated by the inversion model, and the dashed line represents the response curve calculated by the real model. Figure 2In (b) of the diagram, the circled area represents the observed data containing 20% ​​noise. The solid line represents the response curve calculated by the inversion model, and the dashed line represents the response curve calculated by the actual model. The results show that the inversion algorithm achieves good inversion fits for data with varying levels of noise, demonstrating the effectiveness of the algorithm in inverting natural electric field data. Furthermore, the noise level of the observed data has little impact on the inversion fitting results, indicating that the algorithm has good noise resistance. This suggests that the algorithm is well-suited for inverting geophysical data such as natural electric fields, which are easily affected by noise.

[0164] refer to Figure 3 If a dam has a potential piping or seepage channel, the natural electric field data observed at the center of the dam crest and... Figure 2 If the circled points in (b) are the same, then the specific locations of possible piping and seepage points below the survey line can be quickly obtained through the above inversion algorithm process (e.g., Figure 3 (As shown in the solid circle in the image). If multiple measuring lines are arranged, the above inversion algorithm can be used to invert the natural electric field data observed on each measuring line to obtain a series of possible leakage point locations (e.g., ...). Figure 3 (As shown in the hollow circle). Connecting all the seepage points provides a complete picture of the dam's seepage channels, providing a reliable basis for timely reinforcement and seepage prevention measures to avoid disasters.

[0165] Compared to conventional methods, the inversion algorithm in this embodiment has a better balance between exploration and exploitation capabilities, and can adaptively escape local optima to obtain the global optimum. When used for inversion of natural electric field data, good exploration capability ensures a thorough exploration of the entire solution space, while good exploitation capability ensures accurate local searches to obtain precise inversion results, thus improving the accuracy of inversion interpretation. (Conventional methods with good exploration capabilities often lack exploitation capabilities, and those with good exploitation capabilities often lack exploration capabilities, leading to a tendency to get trapped in local optima).

[0166] Natural electric field data is susceptible to various noise interferences. Noise data increases the difficulty of inversion interpretation and may even lead to inversion failure. The leukocyte-optimized inversion algorithm, with its multi-mechanism synergistic strategy that simulates the process of biological leukocytes engulfing pathogens, has greatly improved its resistance to noise interference and improved the reliability of inversion interpretation.

[0167] The natural electric field inversion method based on the white blood cell optimization algorithm can still provide accurate search results even with a large search space, and can largely overcome the problem of conventional inversion methods that rely heavily on prior information.

[0168] This embodiment provides a natural electric field data inversion method based on a leukocyte optimization algorithm, which has the following beneficial effects:

[0169] (1) Dynamic role switching: White blood cells simultaneously undertake the tasks of exploration (chemotaxis) and development (phagocytosis), and dynamically adjust according to the environment;

[0170] (2) Multi-mechanism collaboration: It integrates random walk, gradient guidance, local search and global restart, and has good balance;

[0171] (3) Adaptability: The NETosis mechanism is used to adaptively handle stagnation problems, which is highly robust;

[0172] (4) Parallelism: Multiple white blood cells can work independently and in parallel, which is suitable for distributed computing;

[0173] (5) It can perform the inversion of natural electric field data without prior constraints and has good noise resistance.

[0174] Example 2

[0175] refer to Figure 4 This embodiment provides a natural electric field data inversion system based on the leukocyte optimization algorithm. The natural electric field data inversion method based on the leukocyte optimization algorithm described in this embodiment includes the following modules:

[0176] The data acquisition module is used to acquire observation data of the natural electric field measurement points to be inverted and to set the inversion search parameters of the white blood cell optimization algorithm;

[0177] An initialization module is used to initialize the inversion search space and the white blood cell population and pathogen population of the white blood cell optimization algorithm according to the inversion search parameters.

[0178] The inversion iteration module is used to input the natural electric field measurement point observation data into the leukocyte optimization algorithm for inversion iteration to obtain model parameters; the inversion iteration includes leukocyte chemotaxis and recruitment, leukocyte recognition and adhesion of selected pathogens, leukocyte phagocytosis and internalization of selected pathogens, leukocyte degranulation, leukocyte apoptosis and population chemotactic factor update;

[0179] The accuracy verification module is used to perform forward modeling calculations on the theoretical natural electric field anomaly response based on the model parameters. If the fitting accuracy between the theoretical natural electric field anomaly response and the observation data of the natural electric field measuring point meets the preset fitting accuracy, the model parameters obtained in the current inversion iteration are output as the final inversion result. Otherwise, the white blood cell optimization algorithm parameters are adjusted and returned to the inversion iteration module to continue the inversion iteration until the preset fitting accuracy is met.

[0180] The prediction output module is used to determine the specific location of the natural electric field anomaly target and obtain relevant information about the target's shape based on the final inversion result.

[0181] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 A process, multiple processes, and / or boxes Figure 1 Devices that specify the functions in one or more boxes.

[0182] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction device, which is implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0183] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0184] The parts of this invention not described in detail are prior art. It will be apparent to those skilled in the art that this invention is not limited to the details of the above exemplary embodiments, and that the invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects, and are intended to encompass all changes falling within the meaning and scope of equivalents within this invention.

Claims

1. A method for inverting natural electric field data based on a leukocyte optimization algorithm, characterized in that, Includes the following steps: S1. Obtain the observation data of the natural electric field measurement points to be inverted, and set the inversion search parameters of the white blood cell optimization algorithm; The leukocyte optimization algorithm is a novel metaheuristic optimization algorithm inspired by the biological immune system. It finds the optimal solution in a complex search space by simulating the biological immune process of leukocytes recognizing, chemotactically attracting, phagocytizing and eliminating pathogens. S2. Initialize the inversion search space and the white blood cell population and pathogen population of the white blood cell optimization algorithm according to the inversion search parameters; The inversion search parameters include the number of white blood cells (nWBC), the number of pathogens (nPathogens), the maximum number of iterations (maxIter), chemokine sensitivity, phagocytic radius, degranulation probability (degranulateProb), number of white blood cell apoptosis triggers (netosisThresh), restart probability (restartProb), geoelectric model constraints, the location of white blood cells, the location of pathogens, the resultant force on each white blood cell (total_force), and the current iteration number (iter). The chemokine sensitivity includes the chemokine step size factor alpha for white blood cell movement and the random exploration factor beta for white blood cell movement. The geoelectric model constraints include the geoelectric model parameter dimension dim, the lower bound of the geoelectric model parameter search space lb, and the upper bound of the geoelectric model parameter search space ub. S3. Input the natural electric field measurement point observation data into the leukocyte optimization algorithm for inversion iteration to obtain model parameters; the inversion iteration includes leukocyte chemotaxis and recruitment, leukocyte recognition and adhesion of selected pathogens, leukocyte phagocytosis and internalization of selected pathogens, leukocyte degranulation, leukocyte apoptosis and population chemotactic factor update. The inversion iteration specifically includes: S31, Leukocyte Chemotaxis and Recruitment: The position of leukocytes is updated by calculating the chemokine concentration obtained from the fit difference of pathogens and the average distance between each leukocyte and all pathogens in the solution space. Step S31 specifically includes the following steps: S311. First, calculate the concentration of the pathogen's chemokine, chemo_conc: chemo_conc=1 / (1+exp(-(fitness_Pathogens-mean(fitness_Pathogens)) / std(fitness_Pathogens))), Where fitness_Pathogens is the fit difference for each pathogen, mean(fitness_Pathogens) is the mean of the fit differences for all pathogens, std(fitness_Pathogens) is the standard deviation of the fit differences for each pathogen, and exp() is the natural exponential function. S312. Calculate the average distance avg_distance between each white blood cell and all pathogens in the solution space, and then update the resultant force total_force on each white blood cell according to the chemokinase concentration chemo_conc of the pathogens and the average distance avg_distance: total_force = total_force + chemo_conc / avg_distance; S313. Update the position (WBCs) of each white blood cell based on the resultant force (total_force) acting on each white blood cell, the current iteration number (iter), and the dimension of the geoelectric model parameter (dim): WBCs = WBCs + alpha * total_force + beta * (1 - iter / maxIter) * (rand(1,dim) - 0.5) * (ub - lb), Where rand() is a random function, used in this formula to generate a random number between [1, dim]; S32. Leukocytes recognize and adhere to the selected pathogens: Each leukocyte selects the pathogen closest to it as the selected pathogen. If there are multiple equally spaced nearest pathogens, the selection factor of each pathogen by the leukocyte is calculated, and the pathogen with the smallest factor is selected as the leukocyte's selected pathogen. S33. Leukocytes phagocytose and internalize the selected pathogen: The leukocytes are merged with the selected pathogen to generate a new pathogen. The fitting difference of the new pathogen new_Pathogens is then calculated. If the fitting difference of the new pathogen new_Pathogens is less than the fitting difference of the selected pathogen sel_Pathogens, the new pathogen new_Pathogens replaces the selected pathogen sel_Pathogens. Otherwise, the selected pathogen sel_Pathogens is retained. The original leukocytes proceed to the next degranulation operation. S34, Leukocyte degranulation: Leukocytes randomly generate multiple new leukocytes in the vicinity based on the degranulation probability. Then, the fit difference of all new leukocytes is evaluated, and the new leukocyte with the smallest fit difference is retained to replace the original leukocyte. Step S34 specifically includes: After leukocytes merge with pathogens to produce new individuals, multiple new leukocytes are randomly generated nearby according to the set degranulation probability (degranulateProb). (New_solutions) Generate a random number Random between [0,1]. If Random < degranulateProb, then: New_solutions = WBCs + Random2*(ub-lb) / maxIter, Where Random2 is a random number between [0,1]; New_solutions represents a randomly generated new white blood cell; Then, the fit difference of all new leukocytes is evaluated, and the new leukocytes with the smallest fit difference are retained to replace the original leukocytes; if Random>= degranulateProb, then the original leukocytes are retained. S35, Leukocyte Autoapoptosis: If leukocytes are not updated after the number of leukocyte autoapoptosis triggers, multiple new leukocytes are randomly generated in the inversion search space according to the restart probability, and the new leukocytes with the smallest fitting difference are retained to replace the original leukocytes. In step S35, multiple new white blood cells are randomly generated again within the inversion search space according to the restart probability restartProb, specifically including: Generate a random number Random3 between [0,1]. If Random3 < restartProb, then: New_solutions2 = WBCs + Random4* (ub-lb); Where Random4 is a random number between [0,1]; New_solutions2 represents a randomly generated new white blood cell; restartProb is the restart probability; then the fit difference of all new white blood cells is evaluated, and the new white blood cell with the smallest fit difference is retained to replace the original white blood cell; if Rondom3>= restartProb, then the original white blood cell is retained; S36, Population Chemotaxis Factor Update: If the current iteration number has not reached the maximum iteration number, jump to step S31 to continue the next round of iteration; if the maximum iteration number has been reached, terminate the iteration and output the pathogen location with the smallest fit difference as the model parameter. S4. Perform forward modeling calculation on the theoretical natural electric field anomaly response based on the model parameters. If the fitting accuracy between the theoretical natural electric field anomaly response and the observation data of the natural electric field measuring point meets the preset fitting accuracy, then output the model parameters obtained in the current inversion iteration as the final inversion result. Otherwise, adjust the white blood cell optimization algorithm parameters and return to step S3 to continue the inversion iteration until the preset fitting accuracy is met. S5. Based on the final inversion results, determine the specific location of the natural electric field anomaly target and obtain relevant information about the target's shape.

2. The method according to claim 1, characterized in that, Step S2 specifically includes: Based on the white blood cell population size nWBC in the inversion search parameters, the positions of each white blood cell WBCs are randomly generated in the inversion search space to form a white blood cell population WBC_pop. Based on the pathogen population size nPathogens in the inversion search parameters, a pathogen population Pathogen_pop is formed by randomly generating or customizing the location Pathogens of each pathogen in the inversion search space. The inversion search space is defined by the lower bound lb of the geoelectric model parameter search space and the upper bound ub of the geoelectric model parameter search space.

3. The method according to claim 1, characterized in that, The formula for the fit difference is as follows: Fitness= , Where Fitness represents the fit difference; obdata i caldata represents the actual observed value at the i-th measuring point along the survey line. i nd represents the theoretical response value of the i-th measurement point calculated using the analytical formula or numerical calculation method of the natural electric field, where nd is the number of measurement points.

4. The method according to claim 1, characterized in that, Step S33 specifically includes: The white blood cells are merged with the selected pathogens sel_Pathogens to generate a new pathogen new_Pathogens, as shown in the following formula: new_Pathogens=0.7 * WBCs + 0.3 * sel_Pathogens.

5. A natural electric field data inversion system based on a leukocyte optimization algorithm, characterized in that, The natural electric field data inversion method based on the leukocyte optimization algorithm according to any one of claims 1-4 includes the following modules: The data acquisition module is used to acquire observation data of the natural electric field measurement points to be inverted and to set the inversion search parameters of the white blood cell optimization algorithm; An initialization module is used to initialize the inversion search space and the white blood cell population and pathogen population of the white blood cell optimization algorithm according to the inversion search parameters. The inversion iteration module is used to input the natural electric field measurement point observation data into the leukocyte optimization algorithm for inversion iteration to obtain model parameters; the inversion iteration includes leukocyte chemotaxis and recruitment, leukocyte recognition and adhesion of selected pathogens, leukocyte phagocytosis and internalization of selected pathogens, leukocyte degranulation, leukocyte apoptosis and population chemotactic factor update; The accuracy verification module is used to perform forward modeling calculations on the theoretical natural electric field anomaly response based on the model parameters. If the fitting accuracy between the theoretical natural electric field anomaly response and the observation data of the natural electric field measuring point meets the preset fitting accuracy, the model parameters obtained in the current inversion iteration are output as the final inversion result. Otherwise, the white blood cell optimization algorithm parameters are adjusted and returned to the inversion iteration module to continue the inversion iteration until the preset fitting accuracy is met. The prediction output module is used to determine the specific location of the natural electric field anomaly target and obtain relevant information about the target's shape based on the final inversion result.

Images