A corrosion diagnosis method based on fusion of grounding grid ground potential and ground surface potential

Abstract

The application discloses a corrosion diagnosis method based on fusion of ground potential of grounding net and ground surface potential, and specifically comprises the following steps: acquiring parameters of a grounding net to be diagnosed to establish a corrosion diagnosis calculation model of the grounding net; constructing a fusion observation vector in a healthy state and a to-be-diagnosed state, and obtaining a fusion potential response increment caused by corrosion through difference; setting a candidate corrosion degree set, calculating a theoretical response increment of the ground surface potential and a theoretical response increment of the ground potential of the grounding net, and constructing a fusion corrosion response dictionary matrix; constructing a weighting matrix, establishing a weighted Lasso corrosion positioning model, and determining a corrosion branch set; calculating a theoretical fusion response increment corresponding to different candidate corrosion degree vectors and a weighted residual sum of squares between the theoretical fusion response increment and a measured fusion potential response increment, selecting a candidate corrosion degree vector with the minimum weighted residual sum of squares as a corrosion degree estimation result, and calculating a remaining radius of a conductor. The application realizes corrosion branch positioning and corrosion degree estimation of the grounding net, and can improve the accuracy of corrosion diagnosis of the grounding net in an early slight corrosion state.

Description

Technical Field

[0001] This invention belongs to the technical field of corrosion diagnosis of power grounding grids, and particularly relates to a corrosion diagnosis method based on the fusion of grounding grid potential and ground surface potential. Background Technology

[0002] Grounding grids are crucial infrastructure in power facilities such as substations and converter stations, ensuring the safety of equipment and personnel. They are primarily used to discharge fault currents and lightning currents, and to reduce contact voltage and step voltage. Because grounding grids are buried underground for extended periods, they are susceptible to corrosion due to factors such as soil moisture, resistivity, pH, oxygen content, stray current, and electrochemical reactions. Corrosion reduces the cross-sectional area of ​​the grounding conductor, and in severe cases, can cause conductor breakage or partial failure, thereby affecting the current dissipation performance of the grounding grid and threatening the safe operation of the power system. Therefore, accurately diagnosing the location and extent of corrosion in the grounding grid is of great significance for extending its service life and developing targeted protective measures.

[0003] Currently, commonly used trenchless detection methods for grounding grid corrosion diagnosis include electrical network analysis, electromagnetic field analysis, and surface potential-based diagnostic methods. Electrical network analysis, by treating the grounding grid as a resistive network model and using port measurement data to invert branch resistance changes, achieves corrosion diagnosis. However, the actual number of branches within the grounding grid is usually much greater than the number of measurable down conductors, easily leading to insufficient measurable information and ill-conditioned inversion equations. This is especially true in the early, slight corrosion stage, where branch resistance changes are small and external port responses are weak, making it difficult to reliably identify corroded branches. Electromagnetic field analysis determines conductor corrosion or fracture status by measuring the surface magnetic field distribution after current injection into the grounding grid. However, the electromagnetic environment at substation sites is complex; operating equipment, external lines, environmental noise, conductor burial depth, and soil conditions can all weaken or submerge weak abnormal signals caused by corrosion, affecting diagnostic accuracy. Diagnostic methods based on surface potential are convenient and low-cost, and can reflect the overall potential field changes of the grounding grid through the distribution of surface potential. However, surface potential mainly reflects the macroscopic diffusion characteristics of leakage current in the soil medium. Potential disturbances caused by local minor corrosion are easily weakened during the propagation to the surface, and the sensitivity to early minor corrosion is insufficient. Summary of the Invention

[0004] To overcome the shortcomings of existing grounding grid corrosion diagnosis methods in terms of the ability to identify minor corrosion, the completeness of observation information, and the reliability of quantitative assessment of corrosion degree, this invention provides a corrosion diagnosis method based on the fusion of grounding grid potential and ground surface potential.

[0005] The present invention provides a corrosion diagnosis method based on the fusion of ground potential and surface potential of the grounding grid, comprising the following steps:

[0006] Step 1: Obtain the topology, conductor material parameters, initial conductor radius, conductor branch number, soil parameters, injected current magnitude, injected current location, ground surface potential measurement point location, and ground potential observation point location of the grounding grid to be diagnosed, and establish a grounding grid corrosion diagnosis calculation model.

[0007] The grounding grid to be diagnosed includes N conductor branches to be diagnosed. The k-th conductor branch to be diagnosed is numbered k, k=1, 2, …, N. The initial radius of the conductor in the healthy state is denoted as r0, and the remaining radius of the k-th conductor branch after corrosion is denoted as r. k The corrosion degree of the k-th conductor branch to be diagnosed is δ. k And satisfy:

[0008] (1)

[0009] Where, when δ k When δ = 0, it indicates that the k-th conductor branch to be diagnosed has not experienced corrosion; when δ k When the value is greater than 0, it indicates that corrosion has occurred in the k-th conductor branch to be diagnosed.

[0010] Step 2: Based on the grounding grid corrosion diagnosis calculation model established in Step 1, obtain the ground surface potential observation vector and the ground grid-to-ground potential observation vector under the healthy state of the grounding grid, and construct the fused observation vector under the healthy state. :

[0011] (2)

[0012] In the formula, This represents the observed surface potential vector under healthy conditions. This is the ground potential observation vector of the grounding grid under healthy conditions.

[0013] Step 3: Obtain the surface potential observation vector and the grounding grid-to-ground potential observation vector under the condition to be diagnosed, and construct the fused observation vector under the condition to be diagnosed. :

[0014] (3)

[0015] In the formula, This represents the surface potential observation vector under the condition to be diagnosed. This is the ground potential observation vector of the grounding grid under the condition to be diagnosed.

[0016] Step 4: Based on the fused observation vectors of the health status obtained in Step 2 The fused observation vector obtained in step 3 under the state to be diagnosed Calculate the increment of fusion potential response caused by corrosion. :

[0017] (4)

[0018] In the formula, This is the surface potential response increment vector, used to characterize the spatial potential field changes in the surface area surrounding the grounding grid due to corrosion disturbance. This is the ground potential response increment vector of the grounding grid, used to characterize the effect of corrosion disturbance on the potential distribution of the grounding grid conductors themselves.

[0019] Step 5: Define the candidate corrosion degree set The set of candidate corrosion degrees There are M candidate corrosion degrees, and the m-th candidate corrosion degree is denoted as... m=1, 2, …, M, and based on the candidate corrosion degree Determine the corresponding candidate remaining radius :

[0020] (5)

[0021] Step 6: Initialize the branch number to be diagnosed and the candidate corrosion level number. Induce the current branch to be diagnosed to undergo corrosion at the current candidate corrosion level, and calculate the theoretical surface potential response increment of this candidate corrosion state relative to the healthy state. and the theoretical response increment of the grounding grid to ground potential The two are then combined to form the corresponding corrosion response atom; if the current candidate corrosion degree has not been completely traversed, the candidate corrosion degree number is updated and the corrosion response atom is calculated again. If the current candidate corrosion level has been traversed but the current conductor branch to be diagnosed has not been traversed, then update the conductor branch number to be diagnosed and traverse the candidate corrosion level again; if both the candidate corrosion level and the conductor branch to be diagnosed have been traversed, then proceed to step 7.

[0022] Step 7: Combine all the corrosion response atoms obtained in Step 6 Arrange the conductor branches to be diagnosed (k) and the candidate corrosion severity numbers (m) to construct a fusion corrosion response dictionary matrix. .

[0023] Step 8: Based on the number of surface potential measurement points Number of ground potential observation points of the grounding grid Standard deviation of noise in ground potential measurement Standard deviation of noise in grounding grid to ground potential measurement Surface potential fusion weighting coefficient and the grounding grid's weighting coefficient for ground potential Construct a weighted matrix W, and use the weighted matrix W to adjust the fusion potential response increment. and fusion corrosion response dictionary matrix Weighting is performed to obtain the weighted fusion potential response increment and the weighted fusion corrosion response dictionary matrix.

[0024] Step 9: Based on the weighted fusion potential response increment and the weighted fusion corrosion response dictionary matrix obtained in Step 8, establish a weighted Lasso corrosion localization model and define the dictionary coefficient vector. And solve for the dictionary coefficient vector estimate. ; where, dictionary coefficient vector This is used to characterize the degree of matching between the corrosion response atoms corresponding to different conductor branches to be diagnosed and different candidate corrosion degrees, and the weighted measured fusion potential response increment.

[0025] Step 10: Estimate the dictionary coefficient vector based on the values ​​obtained in Step 9. The dictionary coefficients for multiple candidate corrosion levels corresponding to the same conductor branch to be diagnosed are synthesized to obtain the comprehensive response index S for each conductor branch to be diagnosed. k And according to the comprehensive response index S k Determine the set of corrosion branches .

[0026] Step 11: Set of corrosion branches determined in Step 10 Within the specified range, a set of candidate corrosion degree vectors is defined, and the theoretical fusion response increment corresponding to each candidate corrosion degree vector and its weighted sum of squared residuals between the theoretical and measured fusion potential response increments are calculated sequentially. If the candidate corrosion degree vectors are not fully traversed, the candidate corrosion degree vectors are updated and the weighted sum of squared residuals is calculated again. After the traversal is complete, the candidate corrosion degree vector with the smallest weighted sum of squared residuals is selected as the corrosion degree estimation vector for the located corrosion branch. .

[0027] Step 12: Estimate the corrosion degree vector obtained in Step 11 Calculate the estimated vector of the remaining conductor radius of the located corroded branch. It outputs an estimate of the degree of corrosion and an estimate of the remaining radius of the conductor.

[0028] Furthermore, in step 6, corrosion response atoms are generated by splicing single-branch corrosion perturbation with dual-potential response.

[0029] When constructing the m-th candidate corrosion state corresponding to the k-th conductor branch to be diagnosed, simply change the radius of that branch from r0 to r m If the radii of the remaining branches remain r0, then the branch radius vector under this candidate corrosion state is expressed as:

[0030] (6)

[0031] Based on the branch radius vector r (k,m) The corresponding theoretical response vector of the land surface potential was calculated. The theoretical response vector of the grounding grid to ground potential Then the corrosion response atoms Represented as:

[0032] (7)

[0033] Furthermore, in step 8, the weighting matrix W is determined jointly based on the number of measurement points for surface potential and grounding grid-to-ground potential, the standard deviation of measurement noise, and the fusion weighting coefficient. It is used to balance the differences between surface potential and grounding grid-to-ground potential in terms of the number of measurement points, the standard deviation of measurement noise, and the numerical scale.

[0034] The weighted matrix W is represented as:

[0035] (8)

[0036] In the formula, for 3D identity matrix for 3D identity matrix This is the surface potential weighting coefficient. This is the grounding grid's weighting coefficient for ground potential.

[0037] The weighted fusion potential response increment and the weighted fusion corrosion response dictionary matrix are as follows:

[0038] (9)

[0039] Furthermore, in step 9, the weighted Lasso erosion localization model constrains the dictionary coefficient vector through a norm regularization term. The sparsity of the coefficients causes the dictionary coefficients corresponding to the non-corroded conductor branches to approach zero, and the weighted fusion corrosion response dictionary matrix is ​​derived from this. The incremental potential response of the weighted measured fusion potential was selected from the data. Matching corrosion response atoms.

[0040] The weighted Lasso erosion localization model is expressed as:

[0041] (10)

[0042] In the formula, These are estimates of the dictionary coefficient vector. For regularization parameters, It is a norm 2. It is a norm.

[0043] Furthermore, steps 9 to 11 employ a two-stage corrosion diagnosis process. First, a weighted Lasso corrosion localization model is used to identify the set of corrosion branches in a high-dimensional candidate space consisting of all N conductor branches to be diagnosed and M candidate corrosion degrees. Then, the corrosion branches are assembled. Within the range, a weighted least squares model is used to quantitatively estimate the degree of corrosion, and the remaining radius estimation vector of the conductor of the located corrosion branch is calculated based on the corrosion degree estimation vector.

[0044] The set of eroded branches is represented as:

[0045] (11)

[0046] The set of candidate corrosion degree vectors for the located corrosion branches is represented as follows:

[0047] (12)

[0048] The q-th candidate corrosion degree vector is represented as:

[0049] (13)

[0050] Considering the current set of measurable points Ω, the candidate corrosion degree vector The measurable theoretical fusion response increment is expressed as follows:

[0051] (14)

[0052] In the formula, For the first The identified corrosion branch is in the candidate corrosion level. The single-branch fusion response increment can be obtained from the corresponding corrosion response atoms constructed in step 6. The weighted least squares model for corrosion degree assessment is expressed as:

[0053] (15)

[0054] In the formula, Let Q be the q-th candidate corrosion degree vector, and let Q be the number of candidate corrosion degree vectors. For corrosion branch collection In the candidate corrosion degree vector Theoretical fusion response increment over the set of measurable points Ω For the corresponding weighted sum of squared residuals, The candidate corrosion degree vector index that minimizes the weighted sum of squared residuals. This is the estimated corrosion level vector for the located corrosion branches.

[0055] The beneficial technical effects of this invention compared to the prior art are as follows:

[0056] This invention utilizes both the grounding grid's ground potential and the ground surface potential as diagnostic features for corrosion. It comprehensively leverages the advantages of the grounding grid's ground potential's sensitivity to local branch corrosion and the rich spatial distribution information of the ground surface potential, overcoming the problem of insufficient diagnostic information from single observations. By constructing a fused corrosion response dictionary, establishing a weighted Lasso corrosion localization model, and combining it with the weighted least squares method for estimating corrosion degree and conductor remaining radius, this invention improves the accuracy and stability of locating corroded branches in early, minor corrosion conditions. It addresses the weakness of existing methods in identifying minor corrosion, providing a valid basis for grounding grid condition assessment, operation and maintenance, and the formulation of protective measures. Attached Figure Description

[0057] Figure 1 This is a schematic diagram of the corrosion diagnosis method based on the fusion of ground potential and surface potential of the grounding grid according to the present invention. Detailed Implementation

[0058] The present invention will now be described in further detail with reference to the accompanying drawings and specific implementation methods.

[0059] The process of a corrosion diagnosis method based on the fusion of ground potential and surface potential of the grounding grid according to the present invention is as follows: Figure 1 As shown, the specific steps include:

[0060] Step 1: Obtain the topology, conductor material parameters, initial conductor radius, conductor branch number, soil parameters, injected current magnitude, injected current location, ground surface potential measurement point location, and ground potential observation point location of the grounding grid to be diagnosed, and establish a grounding grid corrosion diagnosis calculation model.

[0061] The grounding grid to be diagnosed includes N conductor branches to be diagnosed. The k-th conductor branch to be diagnosed is numbered k, k=1, 2, …, N. The initial radius of the conductor in the healthy state is denoted as r0, and the remaining radius of the k-th conductor branch after corrosion is denoted as r. k The corrosion degree of the k-th conductor branch to be diagnosed is δ. k And satisfy:

[0062] (1)

[0063] Where, when δ k When δ = 0, it indicates that the k-th conductor branch to be diagnosed has not experienced corrosion; when δ k When the value is greater than 0, it indicates that corrosion has occurred in the k-th conductor branch to be diagnosed.

[0064] The corrosion state vector δ of the grounding grid branch is represented as:

[0065] (2)

[0066] Since grounding grid corrosion typically occurs only on some branches, the corrosion state vector δ satisfies the sparsity condition:

[0067] (3)

[0068] In the formula, The number of non-zero elements in the corrosion state vector is represented by s, and s represents the number of conductor branches that actually undergo corrosion.

[0069] Step 2: Based on the grounding grid corrosion diagnosis calculation model established in Step 1, obtain the ground surface potential observation vector and the ground grid-to-ground potential observation vector under the healthy state of the grounding grid, and construct the fused observation vector under the healthy state. :

[0070] (4)

[0071] In the formula, This represents the observed surface potential vector under healthy conditions. This is the ground potential observation vector of the grounding grid under healthy conditions.

[0072] Step 3: Obtain the surface potential observation vector and the grounding grid-to-ground potential observation vector under the condition to be diagnosed, and construct the fused observation vector under the condition to be diagnosed. :

[0073] (5)

[0074] In the formula, This represents the surface potential observation vector under the condition to be diagnosed. This is the ground potential observation vector of the grounding grid under the condition to be diagnosed.

[0075] Step 4: Based on the fused observation vectors of the health status obtained in Step 2 The fused observation vector obtained in step 3 under the state to be diagnosed Calculate the increment of fusion potential response caused by corrosion. :

[0076] (6)

[0077] In the formula, This is the surface potential response increment vector, used to characterize the spatial potential field changes in the surface area surrounding the grounding grid due to corrosion disturbance. This is the ground potential response increment vector of the grounding grid, used to characterize the effect of corrosion disturbance on the potential distribution of the grounding grid conductors themselves.

[0078] Step 5: Define the candidate corrosion degree set The set of candidate corrosion degrees There are M candidate corrosion degrees, and the m-th candidate corrosion degree is denoted as... m=1, 2, …, M, and based on the candidate corrosion degree Determine the corresponding candidate remaining radius .

[0079] The candidate corrosion degree set Represented as:

[0080] (7)

[0081] The candidate remaining radius corresponding to the m-th candidate corrosion degree Represented as:

[0082] (8)

[0083] Step 6: Initialize the branch number to be diagnosed, k=1, and the candidate corrosion level number, m=1. Iterate through the branch numbers to be diagnosed, k and m, assuming that the k-th branch to be diagnosed has the m-th candidate corrosion level. For corrosion, calculate the theoretical surface potential increment of the candidate corrosion state relative to the healthy state. and the theoretical response increment of the grounding grid to ground potential And combine the two to form the corresponding corrosion-response atoms. Specifically, when constructing the m-th candidate corrosion state corresponding to the k-th conductor branch to be diagnosed, only the radius of that branch is set to... Become The radius of the remaining branch roads remains the same. Then the branch radius vector under this candidate corrosion state is expressed as:

[0084] (9)

[0085] Based on branch radius vector The corresponding theoretical response vector of the land surface potential is calculated. The theoretical response vector of the grounding grid to ground potential Then, the corrosion response atoms of the k-th conductor branch to be diagnosed at the m-th candidate corrosion level... Represented as:

[0086] (10)

[0087] In the formula, Let be the theoretical surface potential increment of the k-th conductor branch to be diagnosed relative to the healthy state at the m-th candidate corrosion level. This represents the theoretical response increment of the grounding grid to ground potential relative to the healthy state for the k-th conductor branch to be diagnosed at the m-th candidate corrosion level. During the construction of corrosion response atoms, if the current candidate corrosion level number m is less than M, then m = m + 1, and the process returns to continue calculating the corrosion response atoms for the same conductor branch at the next candidate corrosion level; if m = M and the current conductor branch number k is less than N, then k = k + 1 and m = 1, and the process returns to continue calculating the corrosion response atoms for the next conductor branch; if k = N and m = M, then the calculation of all corrosion response atoms is complete.

[0088] Step 7: Combine all the corrosion response atoms obtained in Step 6 Arrange the conductor branches to be diagnosed (k) and the candidate corrosion severity numbers (m) to construct a fusion corrosion response dictionary matrix. :

[0089] (11)

[0090] In the formula, the fusion corrosion response dictionary matrix Each column corresponds to a corrosion response atom determined by the branch number k of the conductor to be diagnosed and the candidate corrosion degree number m.

[0091] Step 8: Based on the number of surface potential measurement points Number of ground potential observation points of the grounding grid Standard deviation of noise in ground potential measurement Standard deviation of noise in grounding grid to ground potential measurement Surface potential fusion weighting coefficient and the grounding grid's weighting coefficient for ground potential Construct a weighted matrix W, and use the weighted matrix W to adjust the fusion potential response increment. and fusion corrosion response dictionary matrix Weighting is performed to obtain the weighted fusion potential response increment and the weighted fusion corrosion response dictionary matrix.

[0092] The weighting matrix W is used to balance the differences between the surface potential and the ground potential of the grounding grid in terms of the number of measuring points, the standard deviation of measurement noise, and the numerical scale. It is expressed as follows:

[0093] (12)

[0094] In the formula, for 3D identity matrix for 3D identity matrix This is the surface potential weighting coefficient. Let be the ground potential weighting coefficient of the grounding grid, and satisfy:

[0095] (13)

[0096] Among them, the weighted fusion potential response increment and weighted fusion corrosion response dictionary matrix They are respectively:

[0097] (14)

[0098] Step 9: Based on the weighted fusion potential response increment obtained in Step 8 and weighted fusion corrosion response dictionary matrix A weighted Lasso corrosion localization model is established. To establish this model, the coefficient vector x is defined according to the column order of the fused corrosion response dictionary matrix Φ, and then its estimated values ​​are solved. The dictionary coefficient vector x is represented as:

[0099] (15)

[0100] The weighted fusion potential response increment, the weighted fusion corrosion response dictionary matrix, and the dictionary coefficient vector satisfy the following weighted observation relationship:

[0101] (16)

[0102] In the formula, This is the weighted observation noise term.

[0103] The weighted Lasso erosion localization model is expressed as follows:

[0104] (17)

[0105] In the formula, These are estimates of the dictionary coefficient vector. For regularization parameters, It is a norm 2. The first norm is used. The sparsity of the dictionary coefficient vector x is constrained by the first norm regularization term, causing the dictionary coefficients corresponding to the non-corroded conductor branches to approach zero. This is then applied to the weighted fused corrosion response dictionary matrix. The incremental potential response of the weighted measured fusion potential was selected from the data. Matching corrosion response atoms.

[0106] Step 10: Estimate the dictionary coefficient vector based on the values ​​obtained in Step 9. The dictionary coefficients for multiple candidate corrosion levels corresponding to the same conductor branch to be diagnosed are synthesized to obtain the comprehensive response index S for each conductor branch to be diagnosed. k And according to the comprehensive response index S k Determine the set of corrosion branches .

[0107] The comprehensive response index S of the k-th conductor branch to be diagnosed k Represented as:

[0108] (18)

[0109] In the formula, This represents the dictionary coefficient estimate for the k-th conductor branch to be diagnosed at the m-th candidate corrosion level. If the preset number of corrosion branches is f, then the set of corrosion branches... Represented as:

[0110] (19)

[0111] In the formula, This indicates the selection of the top f conductor branches with the highest overall response index. If the number of corrosion branches is not preset, it can also be based on a preset threshold S. th Determine the set of corrosion branches:

[0112] (20)

[0113] In the formula, S th The threshold for the preset comprehensive response index.

[0114] Step 11: Set of corrosion branches determined in Step 10 Within the specified range, a set of candidate corrosion degree vectors is defined, and the candidate corrosion degree vector numbers are initialized to q=1. The theoretical fusion response increment and its weighted residual sum of squares corresponding to each candidate corrosion degree vector are calculated sequentially. The candidate corrosion degree vector with the smallest weighted residual sum of squares is selected as the corrosion degree estimation result. The set of corrosion branches is represented as follows:

[0115] (twenty one)

[0116] The set of candidate corrosion degree vectors for the located corrosion branches is represented as follows:

[0117] (twenty two)

[0118] The q-th candidate corrosion degree vector Represented as:

[0119] (twenty three)

[0120] In the formula, Q represents the number of candidate corrosion degree vectors. For the q-th candidate corrosion degree vector, the first... The candidate corrosion levels corresponding to the identified corrosion branches have been determined.

[0121] During the corrosion degree assessment process, each candidate corrosion degree vector in the candidate corrosion degree vector set Z is traversed sequentially. Calculate the set of corrosion branches The theoretical fusion response increment under this candidate corrosion degree vector. Considering the current measurable point set Ω, the candidate corrosion degree vector... The measurable theoretical fusion response increment is expressed as follows:

[0122] (twenty four)

[0123] In the formula, Select a matrix for the measurement points. For the first The identified corrosion branch is in the candidate corrosion level. The single-branch fusion response increment. The weighted sum of squared residuals for the corrosion assessment is expressed as:

[0124] (25)

[0125] In the formula, This represents the measured fusion potential response increment on the current set of measurable points Ω. This is the weighting matrix corresponding to the current set of measurable points Ω. For corrosion branch collection In the candidate corrosion degree vector The theoretical fusion response increment over the measurable point set Ω. The corrosion degree estimation vector is determined by the candidate corrosion degree vector that minimizes the weighted sum of squared residuals:

[0126] (26)

[0127] In the formula, The candidate corrosion degree vector index that minimizes the weighted sum of squared residuals. This is the corrosion degree estimation vector for the located corrosion branches. In the corrosion degree estimation loop, if the current candidate corrosion degree vector number q is less than Q, then let q = q + 1, and continue to calculate the weighted residual sum of squares corresponding to the next candidate corrosion degree vector; if q = Q, then complete the traversal of all candidate corrosion degree vectors, and determine the corrosion degree estimation vector that minimizes the weighted residual sum of squares.

[0128] Step 12: Estimate the corrosion degree vector obtained in Step 11 Calculate the estimated vector of the remaining conductor radius of the located corroded branch. It outputs an estimate of the degree of corrosion and an estimate of the remaining radius of the conductor.

[0129] The corrosion degree estimation vector of the located corrosion branch is represented as follows:

[0130] (27)

[0131] The estimated vector of the remaining radius of the conductor in the located corrosion branch is expressed as:

[0132] (28)

[0133] In the formula, This is the corrosion degree estimation vector for the located corrosion branches. This is the vector for estimating the remaining radius of the conductor in the located corrosion branch. Let f be an f-dimensional column vector of all 1s. The initial radius of the conductor in a healthy state. For the first The estimated corrosion level of the located corrosion branches. The final output includes the estimated corrosion level of the located corrosion branches and the estimated remaining conductor radius.

[0134] This invention can comprehensively utilize the advantages of grounding grids being sensitive to local corrosion and having rich information on the spatial distribution of ground potential, thereby improving the accuracy of locating corrosion branches and assessing the degree of corrosion in grounding grids. It also solves the problem of insufficient diagnostic stability of existing methods in the early stage of slight corrosion, and provides an effective basis for the diagnosis and maintenance of grounding grid corrosion.

Claims

1. A corrosion diagnosis method based on the fusion of ground potential and surface potential of a grounding grid, characterized in that, Includes the following steps: Step 1: Obtain the topology, conductor material parameters, initial conductor radius, conductor branch number, soil parameters, injected current magnitude, injected current location, ground surface potential measurement point location, and ground potential observation point location of the grounding grid to be diagnosed, and establish a grounding grid corrosion diagnosis calculation model; The grounding grid to be diagnosed includes N conductor branches to be diagnosed. The k-th conductor branch to be diagnosed is numbered k, k=1,2,…,N. The initial radius of the conductor in the healthy state is denoted as r0, and the remaining radius of the k-th conductor branch to be diagnosed after corrosion is denoted as r. k The corrosion degree of the k-th conductor branch to be diagnosed is δ. k And satisfy: （1） Where, when δ k When δ = 0, it indicates that the k-th conductor branch to be diagnosed has not experienced corrosion; when δ k When the value is greater than 0, it indicates that corrosion has occurred in the k-th conductor branch to be diagnosed; Step 2: Based on the grounding grid corrosion diagnosis calculation model established in Step 1, obtain the ground surface potential observation vector and the ground grid-to-ground potential observation vector under the healthy state of the grounding grid, and construct the fused observation vector under the healthy state. : （2） In the formula, This represents the observed surface potential vector under healthy conditions. This represents the ground potential observation vector of the grounding grid under healthy conditions. Step 3: Obtain the surface potential observation vector and the grounding grid-to-ground potential observation vector under the condition to be diagnosed, and construct the fused observation vector under the condition to be diagnosed. : （3） In the formula, This represents the surface potential observation vector under the condition to be diagnosed. This is the ground potential observation vector of the grounding grid under the condition to be diagnosed; Step 4: Based on the fused observation vectors of the health status obtained in Step 2 The fused observation vector obtained in step 3 under the state to be diagnosed Calculate the increment of fusion potential response caused by corrosion. : （4） In the formula, This is the surface potential response increment vector, used to characterize the spatial potential field changes in the surface area surrounding the grounding grid due to corrosion disturbance. This is the ground potential response increment vector of the grounding grid, used to characterize the effect of corrosion disturbance on the potential distribution of the grounding grid conductors themselves; Step 5: Define the candidate corrosion degree set The set of candidate corrosion degrees There are M candidate corrosion degrees, and the m-th candidate corrosion degree is denoted as... m=1, 2, …, M, and based on the candidate corrosion degree Determine the corresponding candidate remaining radius : （5） Step 6: Initialize the branch number to be diagnosed and the candidate corrosion level number. Induce the current branch to be diagnosed to undergo corrosion at the current candidate corrosion level, and calculate the theoretical surface potential response increment of this candidate corrosion state relative to the healthy state. and the theoretical response increment of the grounding grid to ground potential The two are then combined to form the corresponding corrosion response atom; if the current candidate corrosion degree has not been completely traversed, the candidate corrosion degree number is updated and the corrosion response atom is calculated again. If the current candidate corrosion level has been traversed but the current conductor branch to be diagnosed has not been traversed, then update the conductor branch number to be diagnosed and traverse the candidate corrosion level again; if both the candidate corrosion level and the conductor branch to be diagnosed have been traversed, then proceed to step 7. Step 7: Combine all the corrosion response atoms obtained in Step 6 Arrange the conductor branches to be diagnosed (k) and the candidate corrosion severity numbers (m) to construct a fusion corrosion response dictionary matrix. ; Step 8: Based on the number of surface potential measurement points Number of ground potential observation points of the grounding grid Standard deviation of noise in ground potential measurement Standard deviation of noise in grounding grid to ground potential measurement Surface potential fusion weighting coefficient and the grounding grid's weighting coefficient for ground potential Construct a weighted matrix W, and use the weighted matrix W to adjust the fusion potential response increment. and fusion corrosion response dictionary matrix The weighted processing is performed to obtain the weighted fusion potential response increment. and weighted fusion corrosion response dictionary matrix ; Step 9: Based on the weighted fusion potential response increment obtained in Step 8 and weighted fusion corrosion response dictionary matrix A weighted Lasso erosion localization model is established, and a dictionary coefficient vector is defined. And solve for the dictionary coefficient vector estimate. ; where, dictionary coefficient vector The corrosion response atoms and weighted fusion potential response increments used to characterize different conductor branches under diagnosis and different candidate corrosion degrees. The degree of matching between them; Step 10: Estimate the dictionary coefficient vector based on the values ​​obtained in Step 9. The dictionary coefficients for multiple candidate corrosion levels corresponding to the same conductor branch to be diagnosed are synthesized to obtain the comprehensive response index S for each conductor branch to be diagnosed. k And according to the comprehensive response index S k Determine the set of corrosion branches ; Step 11: Set of corrosion branches determined in Step 10 Within the specified range, a set of candidate corrosion degree vectors is defined, and the theoretical fusion response increment corresponding to each candidate corrosion degree vector and the weighted sum of squared residuals between it and the measured fusion potential response increment are calculated sequentially. If the candidate corrosion degree vectors are not traversed completely, the candidate corrosion degree vectors are updated and the weighted sum of squared residuals is calculated again. After the traversal is completed, the candidate corrosion degree vector with the smallest weighted sum of squared residuals is selected as the corrosion degree estimation vector for the located corrosion branch. ; Step 12: Estimate the corrosion degree vector obtained in Step 11 Calculate the remaining radius estimation vector of the located corrosion branch. It outputs an estimate of the degree of corrosion and an estimate of the remaining radius of the conductor.

2. The corrosion diagnosis method based on the fusion of grounding grid potential and surface potential according to claim 1, characterized in that, In step 6, corrosion response atoms are generated by splicing single-branch corrosion perturbation with dual-potential response. When constructing the m-th candidate corrosion state corresponding to the k-th conductor branch to be diagnosed, simply change the radius of that branch from r0 to r m If the radii of the remaining branches remain r0, then the branch radius vector under this candidate corrosion state is expressed as: （6） Based on the branch radius vector r (k,m) The corresponding theoretical response vector of the land surface potential was calculated. The theoretical response vector of the grounding grid to ground potential The difference between these values ​​and the corresponding observation vectors under healthy conditions is used to obtain the theoretical response increment of the ground potential and the theoretical response increment of the grounding grid to the ground potential. Then, the corrosion response atoms... Represented as: （7）。 3. The corrosion diagnosis method based on the fusion of grounding grid potential and surface potential according to claim 1, characterized in that, In step 8, the weighting matrix W is determined based on the number of measurement points for surface potential and grounding grid-to-ground potential, the standard deviation of measurement noise, and the fusion weighting coefficient. It is used to balance the differences between surface potential and grounding grid-to-ground potential in terms of the number of measurement points, the standard deviation of measurement noise, and the numerical scale. The weighted matrix W is represented as: （8） In the formula, for 3D identity matrix for 3D identity matrix This is the surface potential weighting coefficient. This is the ground potential weighting coefficient of the grounding grid; The weighted fusion potential response increment and the weighted fusion corrosion response dictionary matrix are as follows: （9）。 4. The corrosion diagnosis method based on the fusion of grounding grid potential and surface potential according to claim 1, characterized in that, In step 9, the weighted Lasso corrosion localization model constrains the dictionary coefficient vector through a norm regularization term. The sparsity of the coefficients causes the dictionary coefficients corresponding to the non-corroded conductor branches to approach zero, and the weighted fusion corrosion response dictionary matrix is ​​derived from this. The incremental potential response of the weighted measured fusion potential was selected from the data. Matched corrosion response atoms; The weighted Lasso erosion localization model is expressed as: （10） In the formula, These are estimates of the dictionary coefficient vector. For regularization parameters, It is a norm 2. It is a norm.

5. The corrosion diagnosis method based on the fusion of grounding grid potential and surface potential according to claim 1, characterized in that, Steps 9 to 11 employ a two-stage corrosion diagnosis process. First, a weighted Lasso corrosion localization model is used to identify the set of corrosion branches in a high-dimensional candidate space consisting of all N conductor branches to be diagnosed and M candidate corrosion degrees. Then, the corrosion branches are assembled. Within the range, a weighted least squares model is used to quantitatively estimate the degree of corrosion, and the remaining radius estimation vector of the conductor of the located corrosion branch is calculated based on the corrosion degree estimation vector. The set of eroded branches is represented as: （11） The set of candidate corrosion degree vectors for the located corrosion branches is represented as follows: （12） The q-th candidate corrosion degree vector is represented as: （13） Considering the current set of measurable points Ω, the candidate corrosion degree vector The measurable theoretical fusion response increment is expressed as follows: （14） The weighted least squares model for corrosion assessment is expressed as follows: （15） In the formula, Let Q be the q-th candidate corrosion degree vector, and let Q be the number of candidate corrosion degree vectors. For corrosion branch collection In the candidate corrosion degree vector Theoretical fusion response increment over the set of measurable points Ω For the corresponding weighted sum of squared residuals, The candidate corrosion degree vector index that minimizes the weighted sum of squared residuals. This is the estimated corrosion level vector for the located corrosion branches.

Images