Grounding grid corrosion detection method based on improved teaching and learning optimization algorithm

Abstract

The invention discloses a grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm, relates to the technical field of grounding grid corrosion detection, and solves the problems that in the prior art, early-stage tiny corrosion or complex corrosion forms are difficult to reliably recognize, and high-precision and high-reliability detection results of power grid operation and maintenance cannot be met. According to the method, an updating mechanism based on dominant subject reservation is adopted, the dominant dimensions of the optimal and suboptimal individuals are subjected to refined cross reservation, and the local development capability is enhanced. In addition, a self-adaptive t distribution variation strategy is introduced, the search step size is dynamically adjusted according to the iteration process, global exploration is strengthened in the early stage, local refinement is focused in the later stage, and self-adaptive balance of the search precision is achieved. The multi-strategy collaborative optimization framework not only improves the convergence stability and the solving precision of the inversion algorithm, but also can accurately identify the fine electrical change of the shallow low-resistance mutant, and provides reliable technical support for the accurate positioning of the corrosion defect of the grounding grid.

Description

Technical Field

[0001] This invention relates to the field of grounding grid corrosion detection technology, and in particular to a grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm. Background Technology

[0002] As a critical safety facility in the power system, the grounding grid plays a vital role in providing a discharge channel for fault currents, lightning currents, and operational overvoltages, serving as the foundation for ensuring the safety of personnel and equipment and maintaining the stable operation of the power grid. However, grounding grids are buried in complex soil environments for extended periods, and are subject to the combined effects of soil physicochemical properties, stray currents, and electrochemical corrosion. This makes the grounding conductors highly susceptible to corrosion, leading to reduced cross-sectional area, decreased structural strength, and even fracture, severely weakening the discharge capacity and directly threatening the reliability of the power system. Traditional detection methods, such as localized excavation inspection, while intuitive, are inefficient and costly. Non-excavation methods, such as power network analysis and electrochemical analysis, suffer from problems such as requiring power outages, complex procedures, or inaccurate location. In recent years, transient electromagnetic methods, as a geophysical exploration technology that requires no excavation and does not interfere with operating equipment, have demonstrated significant advantages in detecting shallow underground targets, providing a new technical approach for grounding grid corrosion detection. This method applies a pulsed current through a transmitting coil, receives the secondary magnetic field response generated by induced eddy currents in the underground conductor, and then inverts the resistivity distribution to determine the corrosion status. It is particularly suitable for online detection needs under uninterrupted power grid conditions.

[0003] Although transient electromagnetic methods have potential applications in grounding grid inspection, the limitations of their inversion algorithms severely restrict their detection accuracy and engineering practicality. Traditional linear inversion algorithms, such as the least squares method, OCCAM inversion, and Gauss-Newton method, are essentially local optimization methods relying on the initial model. They approximate the optimal solution through mathematical iteration, making them prone to local optima traps. These methods require regularization constraints during computation to improve ill-conditioned behavior, but for special targets like grounding grids with shallow burial depths and abrupt low-resistance changes, the inversion results are often unsatisfactory, failing to accurately characterize the boundary morphology of corrosion defects and quantify the degree of corrosion. More importantly, the inversion results are highly sensitive to the choice of the initial model; different initial guesses can lead to drastically different inversion conclusions, lacking robustness and stability. This fundamental deficiency makes it difficult for existing technologies to reliably identify early-stage micro-corrosion or complex corrosion morphologies in practical engineering, failing to meet the urgent needs of power grid operation and maintenance for high-precision, high-reliability detection results. Therefore, it is imperative to break through the constraints of traditional algorithmic frameworks.

[0004] Therefore, a grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm is needed. Summary of the Invention

[0005] To address the limitations of existing technologies in reliably identifying early-stage minute corrosion or complex corrosion morphologies, and thus failing to meet the demands of power grid operation and maintenance for high-precision and high-reliability detection results, this invention provides a grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm. This method, utilizing a multi-strategy collaborative optimization framework, significantly improves the convergence stability and solution accuracy of the inversion algorithm. Furthermore, it accurately identifies subtle electrical changes in shallow, low-resistivity aberrations, providing a reliable technical guarantee for the precise location and quantification of grounding grid corrosion defects. The specific technical solution is as follows: A grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm includes the following steps: S1: A one-dimensional forward model of the grounding grid is constructed based on the transient electromagnetic method. The theoretical electromagnetic response data is obtained by injecting pulse current into the transmitting coil and receiving the time-domain response signal of the induced electromotive force. S2: Construct an improved teaching and learning optimization algorithm to perform inversion optimization with the goal of minimizing the fitting error between transient electromagnetic observation data and forward modeling data; S3: Based on the resistivity distribution characteristics obtained from the inversion, locate the corrosion fracture location of the grounding grid and quantify the degree of corrosion; The improved teaching and learning optimization algorithm includes the following improvement strategies: (1) Introduce an update mechanism based on the retention of advantageous disciplines, and perform cross-operations on different dimensions of the most and second best individuals; (2) An adaptive t-distribution mutation strategy is adopted. When the optimal fitness does not improve in multiple iterations, the population is perturbed by the t-distribution in which the degree of freedom parameter changes adaptively with the number of iterations, so as to balance the global and local search accuracy.

[0006] The preferred and improved teaching and learning optimization algorithm uses Hammersley low-discrepancy sequences and an improved reverse learning strategy for population initialization, as follows: For an optimization problem with D decision variables and a population size of Np, the first dimension is generated by uniform distribution, and the remaining dimensions are generated using Van der Corput sequences based on prime numbers, resulting in an initial population of uniformly distributed individuals.

[0007] Preferably, an elite selection process is performed after Hammersley sequence initialization: For each individual in the initial population X i Calculate the inverse solution within the upper and lower bounds of the parameter. The formula for calculating the inverse solution is: In the formula, The solution is the reverse. is a random number that follows a uniform distribution between [1, 5], and [lb, ub] are the upper and lower bounds of each variable.

[0008] Preferred updating mechanisms based on the retention of advantageous disciplines include: In each iteration, the components of each dimension of the best-fit individual Xbest and the second-best individual Xsecond are crossed sequentially to generate a cross individual Xcross(j), where the j-th dimension is taken from the j-th dimension of Xsecond and the other dimensions are taken from Xbest. Calculate the fitness of the crossover individual. If it is better than the original best individual, then update the best individual with the crossover individual; otherwise, leave the original best individual unchanged.

[0009] Preferably, the adaptive t-distribution mutation strategy includes: When the optimal fitness is detected to have not improved in three consecutive iterations, the mutation operation is activated. The mutation formula is: Where t(iter) represents a t-distributed random number related to the degree of freedom parameter and the current iteration number iter.

[0010] Preferably, the fitness function for the inversion optimization is: in, Mean absolute percentage error is used to measure the relative error between the forward model and the observed value. This is the standard deviation term for the data, used to increase the weight of the early transient electromagnetic signal.

[0011] Preferably, the one-dimensional forward model is based on a horizontal layered geoelectric model. The frequency domain electromagnetic field response is numerically solved by Hankel transformation, and then converted into a time domain induced electromotive force response by cosine transformation and piecewise linear approximation.

[0012] A computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the grounding grid corrosion detection method based on the improved teaching and learning optimization algorithm as described above.

[0013] A processor for running a program, wherein the program executes the grounding grid corrosion detection method based on the improved teaching and learning optimization algorithm as described above.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention fundamentally overcomes the shortcomings of traditional linear methods that are prone to getting trapped in local optima by introducing an improved teaching and learning optimization algorithm into transient electromagnetic inversion. The scheme utilizes the global search characteristics of metaheuristic algorithms to explore a wide range of solutions, effectively avoiding dependence on the initial model. A high-quality initial population is constructed using Hammersley low-discrepancy sequences and an improved back-learning strategy, significantly enhancing the diversity of the starting point for solutions. An update mechanism based on the retention of advantageous disciplines is adopted to refine the cross-preservation of the advantageous dimensions of optimal and suboptimal individuals, enhancing local development capabilities. An adaptive t-distribution mutation strategy is introduced to dynamically adjust the search step size according to the iteration process, strengthening global exploration in the early stages and focusing on local refinement in the later stages, achieving an adaptive balance in search accuracy. This multi-strategy collaborative optimization framework not only significantly improves the convergence stability and solution accuracy of the inversion algorithm but also enables it to accurately identify subtle electrical changes in shallow low-resistivity mutants, providing reliable technical support for the precise location and quantification of grounding grid corrosion defects. It is particularly suitable for engineering applications in complex power grid environments requiring trenchless, high-precision detection. Attached Figure Description

[0015] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0016] Figure 1 This is a forward-modeled layered model; Figure 2 For different degrees of freedom t distributed; Figure 3 Here is the algorithm flowchart; Figure 4 This is a comparison chart of the inversion results; Figure 5 The algorithm error reduction curve; Figure 6 This is a comparison chart of the forward response. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0019] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0020] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0021] In one embodiment of the present invention, an improved teaching and learning optimization inversion algorithm is provided for the detection of grounding grid corrosion defects. First, taking a layered geoelectric model as the research object, the transient electromagnetic one-dimensional forward modeling theory is introduced. Furthermore, several improvement strategies are introduced to the standard teaching and learning optimization algorithm to improve inversion accuracy and accurately determine the grounding grid corrosion status. Specifically, Hammersley low-discrepancy sequences and improved reverse learning initialization populations are introduced to improve population diversity and overall population quality; an update mechanism based on the retention of advantageous disciplines is introduced to enhance the algorithm's local search capability; and a mutation strategy based on adaptive t-distribution is introduced to enhance the algorithm's global search capability and improve algorithm accuracy. To verify the effectiveness of the algorithm, a multilayered geoelectric model is used to test the inversion effect. Finally, measured data is used to verify the feasibility of this method in practical engineering applications. The specific process is as follows: Transient electromagnetic method is a detection technique based on the principle of electromagnetic induction. By injecting a pulsed current into a transmitting coil, eddy currents are induced in the underground conductor under a changing primary magnetic field at the instant the current is turned off. By detecting and analyzing the secondary magnetic field response generated during the decay of the eddy currents, the resistivity distribution characteristics of the underground medium can be inferred.

[0022] Since the burial depth of the grounding grid is fixed, we use a horizontal layered geoelectric model for approximate research. For n Layered horizontal geoelectric model, such as Figure 1 As shown. Taking a center loop source as an example, the frequency domain electromagnetic field response is: (1) In the formula I It is the amplitude of the excitation current of the transmitting coil. It is the radius of the transmitting coil. It is angular frequency. The permeability of free space, It is an integral variable. J 1 and J 0 represents the zeroth-order and first-order Bessel functions, respectively. TE This is the reflection coefficient, calculated using the following formula: (2) In the formula It can be obtained recursively from equation (3): (3) The integral in equation (1) is numerically solved using the Hankel transform. This algorithm uses 140-point Hankel filter coefficients to obtain the following transformation: (4) In the formula W i These are the filter coefficients. m =140, a This is the offset coefficient. s The sampling interval is given. Based on spectral analysis theory, the perpendicular component of the magnetic field in the time domain... H z (t) and It can be represented as: (5) After performing a cosine transform on the expression for the vertical component of the magnetic field given in equation (5), the solution is obtained by using the piecewise linear approximation method. This allows the magnetic field strength and rate of change of magnetic field strength in the frequency domain to be converted to the time domain, as shown in equation (6). (6) The expression for the induced electromotive force is further obtained as follows: (7) In the formula S For effective receiving area, μ is the relative permeability.

[0023] Based on the time-domain response signal of the induced electromotive force, subsequent inversion of formation parameters can be performed.

[0024] The teaching and learning optimization algorithm is a metaheuristic algorithm proposed by Rao.

[26] Its core idea originates from the two basic stages of "teaching" and "learning" in a simulated classroom teaching process. In the "teaching" stage, the individual with the highest knowledge level acts as the teacher, raising the average level of the entire class; in the "learning" stage, students discuss and exchange ideas with each other to further improve their knowledge. This algorithm requires very few control parameters, avoiding performance degradation and reduced computational efficiency caused by improper parameter settings. Furthermore, because the algorithm has few steps and efficient global search, it converges quickly. The specific steps of the teaching and learning optimization algorithm are as follows: 1. Population Initialization Define population size N p The maximum number of iterations (Maxiter), the number of decision variables (D), and the upper and lower bounds [lb, ub] of each variable are randomly generated within the feasible region of the decision variables. N p There are 10 candidate solutions.

[0025] 2. Teaching Stage Find the individual with the best fitness in the current population and consider it as the teacher. X teacher The iterative formula for the teaching stage is: (8) In the formula, It is the first i The current location of each individual. This is the updated location. r i It is a random number that follows a uniform distribution within the interval [0,1]. TF i The teaching factor is randomly assigned a value of 1 or 2 to simulate the uncertainty of teaching effectiveness.

[0026] 3 learning stages For the student group within the population, let each individual X i Randomly select another different individual X j Learning. Taking the minimization problem as an example, the iterative formula for the learning stage is: (9) In the formula, r i It is a random number that follows a uniform distribution within the interval [0,1]. and The first i and the j The fitness of an individual.

[0027] Furthermore, the standard teach-and-learn optimization algorithm uses a simple uniform random distribution to generate the initial population. This initialization method leads to an uneven distribution of the initial population, reducing the algorithm's global optimization ability, and the solution results are overly dependent on the initial random process, affecting the algorithm's stability. Therefore, for high-dimensional nonlinear optimization problems with complex solution spaces, such as transient electromagnetic inversion, a uniformly distributed and elite-selected initial population is crucial for accelerating algorithm convergence and finding the global optimum.

[0028] Based on this, this embodiment introduces Hammersley low-dissimilarity sequences. Hammersley low-dissimilarity sequences are very well adapted to high-dimensional fixed-point problems and have better uniformity than traditional initialization methods such as Logistic mapping and Tent mapping.

[0029] The basic idea behind Hammersley sequences is to construct point sets using inverted cardinality representation. For D dimensional vector, N p For each individual, the first dimension of the generated Hammersley sequence is: (10) The remaining dimensions are: (11) In the formula i =1,2,…, N p , j =1,2,…, D By selecting a prime number base b , natural numbers k Convert to b The number is then processed by a reversing function. The Van der Corput sequence was generated, and the final Hammersley low-difference sequence was obtained as follows: (12) After mapping to the actual parameter space, the actual initialized population is obtained as follows: (13) To further improve the overall quality of the population and increase the proportion of elite solutions based on population diversity, an improved reverse learning strategy is introduced to further elitize the initial population after homogenizing it.

[0030] Backward learning expands the search space by calculating the geometrically symmetric points of the current solution, compares the fitness of the current solution with the backward solution, and retains the better solution as the final initial solution. However, the pure geometric symmetry search method used by backward learning is inefficient in high-dimensional nonlinear problems. Therefore, an improved backward learning method is used to achieve population elitism. The specific calculation formula is as follows: (14) In the formula, The solution is the reverse. The population consists of random numbers that follow a uniform distribution between [1, 5]. and Fitness comparison, before selection N p The final initial population is composed of individuals with better fitness.

[0031] Furthermore, since the early and late responses of the transient electromagnetic method are affected by the parameters of the shallow and deep media, respectively, the sensitivity of different dimensional parameters to the fitness function varies during the inversion process. In the teaching and learning optimization algorithm, the student population is updated across all dimensions during the teaching and learning phases. This causes some high-quality dimensional variables of individuals to be destroyed during iteration and cannot be retained in the next generation of the population, reducing the accuracy of the algorithm's local search.

[0032] To address this issue, this embodiment proposes an update mechanism based on the retention of advantageous disciplines. Inspired by the crossover operation of the differential evolution algorithm, each iteration sequentially performs crossover operations on different dimensions of the two individuals with the best and second-best fitness, calculates the fitness value after crossover, and if the fitness value of the second-best individual is better than that of the best individual, then the crossover result is retained, and the best individual is updated; otherwise, the original result is retained. Taking the minimization problem as an example, the iterative formula is shown in equation (15): (15) In the formula The first representing individual new teachers j dimensional components, The first student individual representing the best fitness j dimensional components, f(·) Individual fitness.

[0033] Furthermore, the iterative mechanism of the teaching and learning optimization algorithm struggles to perform refined searches of the elite solution domain, and continuing to use the linear iterative strategy of the teaching phase in the later stages of iteration can easily lead to the population getting trapped in local optima and losing inversion accuracy. The mutation strategy proposed in this embodiment activates the mutation operation when it detects that the optimal fitness has not improved in three consecutive iterations, employing an adaptive... t The distribution perturbs the population, enhancing the algorithm's global search capability.

[0034] t The probability distribution is one of the most important probability distributions in statistics, and its probability density function is: (15) In the formula For degrees of freedom parameters, This is the gamma function. For example... Figure 2 As shown, when hour, t The distribution is similar to the Cauchy distribution, exhibiting heavy-tailed characteristics and possessing good global search capabilities; when hour, t The distribution is similar to that of a Gaussian distribution, but the variable asynchrony length decreases, thus it has good local search capabilities.

[0035] Adaptive t The iterative formula for the distribution mutation strategy is: (16) In the formula, iter refers to the number of iterations. When the number of iterations is small, a mutation strategy with strong global search capability is adopted; as the number of iterations increases, t The distribution gradually tends to a Gaussian distribution, which is beneficial for the algorithm to converge quickly.

[0036] To ensure the accuracy of the algorithm's inversion, a greedy selection principle is adopted for screening mutated individuals. That is, only mutated individuals with fitness greater than the original individuals are retained, thus preventing population degradation.

[0037] Finally, simulation verification was performed: Early transient electromagnetic signals are more sensitive to changes in the electrical properties of shallow media, making the inversion accuracy of these early signals particularly important. Furthermore, the induced electromotive force (EMF) signal decays over time, with early signals decaying by several orders of magnitude more than later signals. To increase the weight of early signals during the inversion process, this embodiment defines the data standard deviation (std) as one term in the fitness function. In addition, to comprehensively evaluate the inversion model, the mean absolute percentage error (MAPE) is used to measure the relative error between the forward values ​​and the observed values. The fitness function finally constructed in this embodiment is: (17) In the formula, y i This is the forward value of the induced voltage. n For time channels, This is the observed value of the induced voltage. Figure 3 This is a flowchart of the algorithm in this embodiment.

[0038] To verify the effectiveness of the method, a four-layer geoelectric model was set up for testing, with a population size of 60. The algorithm inversion results are shown in Table 1.

[0039] Table 1 Results of the algorithm-inverted geoelectric model The inversion results show that the algorithm proposed in this embodiment exhibits significant advantages in shallow medium parameter inversion. Although there is some attenuation in deep resolution, this is an adaptive optimization made by the algorithm to meet the engineering requirements of grounding grid corrosion detection.

[0040] from Figure 5 As can be seen, the algorithm has a very low error during population initialization, thanks to the improved population initialization strategy. The subsequent iteration process sets the maximum number of iterations to 150, at which point the algorithm's inversion accuracy meets the requirements. Figure 6 It can be seen that the forward modeling response error of the inverted and observed geoelectric models is very small at this time. However, due to the inherent ill-conditioning of transient electromagnetic inversion, the results of the inverted geoelectric models are difficult to completely match the observation results.

[0041] In summary, this invention fundamentally overcomes the shortcomings of traditional linear methods that are prone to getting trapped in local optima by introducing an improved teaching and learning optimization algorithm into transient electromagnetic inversion. This scheme utilizes the global search characteristics of metaheuristic algorithms to explore a wide range of solutions, effectively avoiding dependence on the initial model. A high-quality initial population is constructed using Hammersley low-discrepancy sequences and an improved back-learning strategy, significantly enhancing the diversity of the solution starting point. An update mechanism based on the retention of advantageous disciplines is adopted to refine the cross-preservation of the advantageous dimensions of optimal and suboptimal individuals, enhancing local development capabilities. An adaptive t-distribution mutation strategy is introduced to dynamically adjust the search step size according to the iteration process, strengthening global exploration in the early stages and focusing on local refinement in the later stages, achieving an adaptive balance in search accuracy. This multi-strategy collaborative optimization framework not only significantly improves the convergence stability and solution accuracy of the inversion algorithm but also enables it to accurately identify subtle electrical changes in shallow low-resistivity mutants, providing reliable technical support for the precise location and quantification of grounding grid corrosion defects. It is particularly suitable for trenchless, high-precision detection engineering applications in complex power grid environments.

[0042] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of each example have been generally described in terms of function in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0043] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical functional division. In actual implementation, there may be other division methods, such as multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored.

[0044] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0045] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this 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 cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm, characterized in that, Includes the following steps: S1: A one-dimensional forward model of the grounding grid is constructed based on the transient electromagnetic method. The theoretical electromagnetic response data is obtained by injecting pulse current into the transmitting coil and receiving the time-domain response signal of the induced electromotive force. S2: Construct an improved teaching and learning optimization algorithm to perform inversion optimization with the goal of minimizing the fitting error between transient electromagnetic observation data and forward modeling data; S3: Based on the resistivity distribution characteristics obtained from the inversion, locate the corrosion fracture location of the grounding grid and quantify the degree of corrosion; The improved teaching and learning optimization algorithm includes the following improvement strategies: (1) Introduce an update mechanism based on the retention of advantageous disciplines, and perform cross-operations on different dimensions of the most and second best individuals; (2) An adaptive t-distribution mutation strategy is adopted. When the optimal fitness does not improve in multiple iterations, the population is perturbed by the t-distribution in which the degree of freedom parameter changes adaptively with the number of iterations, so as to balance the global and local search accuracy.

2. The grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm according to claim 1, characterized in that, The improved teaching and learning optimization algorithm uses Hammersley low-discrepancy sequences and an improved reverse learning strategy for population initialization, as follows: For an optimization problem with D decision variables and a population size of Np, the first dimension is generated by uniform distribution, and the remaining dimensions are generated using Van der Corput sequences based on prime numbers, resulting in an initial population of uniformly distributed individuals.

3. The grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm according to claim 2, characterized in that, Elite selection was performed after Hammersley sequence initialization: For each individual in the initial population X i Calculate the inverse solution within the upper and lower bounds of the parameter. The formula for calculating the inverse solution is: In the formula, The solution is the reverse. is a random number that follows a uniform distribution between [1, 5], and [lb, ub] are the upper and lower bounds of each variable.

4. The grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm according to claim 1, characterized in that, The updating mechanism based on the retention of advantageous disciplines includes: In each iteration, the components of each dimension of the best-fit individual Xbest and the second-best individual Xsecond are crossed sequentially to generate a cross individual Xcross(j), where the j-th dimension is taken from the j-th dimension of Xsecond and the other dimensions are taken from Xbest. Calculate the fitness of the crossover individual. If it is better than the original best individual, then update the best individual with the crossover individual; otherwise, leave the original best individual unchanged.

5. The grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm according to claim 1, characterized in that, The adaptive t-distribution mutation strategy includes: When the optimal fitness is detected to have not improved in three consecutive iterations, the mutation operation is activated. The mutation formula is: Where t(iter) represents a t-distributed random number related to the degree of freedom parameter and the current iteration number iter.

6. The grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm according to claim 1, characterized in that, The fitness function for the inversion optimization is: in, Mean absolute percentage error is used to measure the relative error between the forward model and the observed value. This is the standard deviation term for the data, used to increase the weight of the early transient electromagnetic signal.

7. The grounding grid corrosion detection method based on an improved teaching and learning optimization algorithm according to claim 1, characterized in that, The one-dimensional forward model is based on a horizontally layered geoelectric model. The frequency domain electromagnetic field response is solved numerically by Hankel transformation, and then converted into the time domain induced electromotive force response by cosine transformation and piecewise linear approximation.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the grounding grid corrosion detection method based on any one of claims 1 to 7.

9. A processor, characterized in that, The processor is used to run a program, wherein the program executes the grounding grid corrosion detection method based on the improved teaching and learning optimization algorithm as described in any one of claims 1 to 7.

Images