Training method, prediction method, device and equipment of grounding grid corrosion prediction model and medium
By training the grounding net corrosion prediction model, using the sample data set to characterize the influence and grading of corrosion factors, the problem of complex corrosion detection process and low accuracy in the existing technology is solved, and efficient and accurate corrosion prediction and early warning are achieved to ensure that The safety of power equipment and personnel.
Patent Information
- Application Number
- CN202411113962.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-14
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art cannot effectively detect the corrosion degree of the grounding grid of the power steel tower, resulting in complex inspection processes and low accuracy, endangering the safety of power equipment and personal safety.
The grounding net corrosion prediction model training method is adopted. By obtaining sample data sets for different years, the first model parameters are trained to characterize the degree of influence of corrosion factors, and the second model parameters are used to characterize the grading of corrosion factors to improve detection accuracy and efficiency. .
The corrosion detection process of grounding grid is simplified, the accuracy and efficiency of grounding grid corrosion prediction is improved, manpower and material waste is reduced, and the normal operation of power equipment and the safety of operation and maintenance personnel is ensured.
Smart Images

Figure CN118940907A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of electric power technology, and in particular to a training method, a prediction method, a device, equipment and a medium for predicting a grounding grid corrosion. Background Art
[0002] Grounding is an important part of the reliable operation of electric power steel towers, and it is of great significance for the protection against lightning current, surge, and fault current impact. Among them, the grounding grid materials of electric power steel towers in mountainous areas, such as hot-dip galvanized round steel, will be affected by the electrochemical corrosion of the soil environment. The grounding grid materials (especially the lap welding area) are prone to poor grounding performance due to corrosion, endangering power equipment and personal safety. However, the corrosion of the grounding grid is inevitable. The existing grounding resistance test of the grounding grid can only be used to evaluate the reliability of the grounding, and it is impossible to intuitively evaluate the degree of corrosion of the grounding grid; at the same time, on-site operations can only use large-scale excavation and destruction of existing grounding grid connections to evaluate the degree of corrosion of the grounding grid materials, which will cause a lot of manpower and material resources. Waste, and the detection accuracy is not high. Therefore, the existing technology has the problem of complicated detection process and low detection accuracy of grounding grid corrosion. Summary of the invention
[0003] The present invention provides a training method, prediction method, device, equipment and medium for a grounding grid corrosion prediction model to simplify the detection process of grounding grid corrosion and improve the accuracy and efficiency of grounding grid corrosion prediction.
[0004] According to one aspect of the present invention, a method for training a grounding grid corrosion prediction model is provided, the method comprising:
[0005] Obtain the grounding grid corrosion prediction model to be trained and sample data sets of different years;
[0006] The first model parameter of the grounding grid corrosion prediction model is trained according to the sample data set of each year to obtain the grounding grid corrosion prediction basic model and the first model parameter of the corresponding year; the first model parameter is used to characterize the influence degree of each corrosion factor in the grounding grid corrosion prediction model;
[0007] The second model parameters of the grounding grid corrosion prediction basic model are trained according to the sample data set of each year and the first model parameters to obtain the final grounding grid corrosion prediction model and the second model parameters of the corresponding year; the second model parameters are used to characterize the classification of each corrosion factor.
[0008] According to another aspect of the present invention, a grounding grid corrosion prediction method is provided, the method comprising:
[0009] Obtain sample data for the year to be predicted; the sample data includes sampling values of various corrosion factors;
[0010] The sample data is input into the grounding grid corrosion prediction model to obtain the grounding grid corrosion prediction result for the year to be predicted; wherein the grounding grid corrosion prediction model is obtained by the training method of the grounding grid corrosion prediction model of any embodiment of the present invention.
[0011] According to another aspect of the present invention, a training device for a grounding grid corrosion prediction model is provided, the device comprising:
[0012] A prediction model and sample data acquisition module is used to obtain the grounding grid corrosion prediction model to be trained and sample data sets of different years;
[0013] The first model training module is used to train the first model parameters of the grounding grid corrosion prediction model according to the sample data set of each year, and obtain the grounding grid corrosion prediction basic model and the first model parameters of the corresponding year; the first model parameters are used to characterize the influence degree of each corrosion factor in the grounding grid corrosion prediction model;
[0014] The second model training module is used to train the second model parameters of the grounding grid corrosion prediction basic model according to the sample data set of each year and the first model parameters to obtain the final grounding grid corrosion prediction model and the second model parameters of the corresponding year; the second model parameters are used to characterize the classification of each corrosion factor.
[0015] According to another aspect of the present invention, a grounding grid corrosion prediction device is provided, the device comprising:
[0016] The data acquisition module is used to obtain sample data of the year to be predicted; the sample data includes sampling values of various corrosion factors;
[0017] The corrosion prediction module is used to input sample data into the grounding grid corrosion prediction model to obtain the grounding grid corrosion prediction result for the year to be predicted; wherein the grounding grid corrosion prediction model is obtained by the training method of the grounding grid corrosion prediction model of any embodiment of the present invention.
[0018] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:
[0019] at least one processor; and
[0020] a memory communicatively connected to the at least one processor; wherein,
[0021] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the training method of the grounding grid corrosion prediction model described in any embodiment of the present invention, or execute the grounding grid corrosion prediction method described in any embodiment of the present invention.
[0022] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the training method of the grounding grid corrosion prediction model described in any embodiment of the present invention, or to execute the grounding grid corrosion prediction method described in any embodiment of the present invention when executed.
[0023] The technical solution of the embodiment of the present invention obtains the grounding grid corrosion prediction model to be trained and the sample data sets of different years; trains the first model parameters of the grounding grid corrosion prediction model according to the sample data sets of each year, and obtains the grounding grid corrosion prediction basic model and the first model parameters of the corresponding year; the first model parameters are used to characterize the influence degree of each corrosion factor in the grounding grid corrosion prediction model; trains the second model parameters of the grounding grid corrosion prediction basic model according to the sample data sets of each year and the first model parameters, and obtains the final grounding grid corrosion prediction model and the second model parameters of the corresponding year; the second model parameters are used to characterize the classification of each corrosion factor. This technical solution adopts model training of sample data in two dimensions, and continuously trains and corrects the first model parameters and the second model parameters respectively, thereby effectively improving the model accuracy of the grounding grid corrosion prediction model, thereby improving the accuracy and efficiency of grounding grid corrosion prediction, and simplifying the detection process of grounding grid corrosion.
[0024] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0026] Figure 1 is a flow chart of a training method for a grounding grid corrosion prediction model provided according to Embodiment 1 of the present invention;
[0027] Figure 2 is a flow chart of a training method for a grounding grid corrosion prediction model provided according to the second embodiment of the present invention;
[0028] Figure 3 is a schematic diagram of a round steel test area provided according to Embodiment 2 of the present invention;
[0029] Figure 4is a schematic diagram of a model training process according to the second embodiment of the present invention;
[0030] Figure 5 is a schematic diagram of the process of secondary model training provided according to the second embodiment of the present invention;
[0031] Figure 6 is a flow chart of a grounding grid corrosion prediction method provided according to Embodiment 3 of the present invention;
[0032] Figure 7 2 is a schematic diagram of the structure of a training device for a grounding grid corrosion prediction model provided in accordance with a fourth embodiment of the present invention;
[0033] Figure 8 is a schematic diagram of the structure of a grounding grid corrosion prediction device provided according to Embodiment 5 of the present invention;
[0034] Fig. 9 It is a structural schematic diagram of an electronic device provided according to Embodiment 6 of the present invention. DETAILED DESCRIPTION
[0035] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0036] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0037] Embodiment 1
[0038] Figure 1The flowchart of a method for training a grounding grid corrosion prediction model provided in the first embodiment of the present invention is applicable to the case of training a grounding grid corrosion prediction model for a power steel tower. The method can be executed by a training device for a grounding grid corrosion prediction model. The training device for a grounding grid corrosion prediction model can be implemented in the form of hardware and / or software. The training device for a grounding grid corrosion prediction model can be configured in an electronic device. The electronic device can include but is not limited to computers, terminals, servers and other devices with data processing capabilities. Figure 1 As shown, the training method of a grounding grid corrosion prediction model provided in the first embodiment specifically includes the following steps:
[0039] S110, obtaining a grounding grid corrosion prediction model to be trained and sample data sets of different years.
[0040] In an embodiment of the present invention, a pre-created grounding grid corrosion prediction model to be trained can be obtained from a data storage location not limited to a local or remote server, and a sample data set collected in different years, i.e., with a natural year as a cycle, can be obtained. Among them, the grounding grid corrosion prediction model can be constructed based on the relationship between the equivalent resistivity ρ of the grounding grid material and the corrosion coefficient a corresponding to each corrosion factor X. In addition, this embodiment uses a natural year as a cycle to collect sample data in each natural year. In practical applications, other time granularities can also be used as the sample data collection cycle, such as 6 months, 2 years, etc., which is not specifically limited in this embodiment. The sample data set may include but is not limited to the following sample data: equivalent resistivity observation values and sampled values of each corrosion factor.
[0041] S120. Train the first model parameter of the grounding grid corrosion prediction model according to the sample data set of each year to obtain the grounding grid corrosion prediction basic model and the first model parameter of the corresponding year; the first model parameter is used to characterize the influence degree of each corrosion factor in the grounding grid corrosion prediction model.
[0042] Among them, the corrosion factors can be understood as various factors that may cause corrosion to the grounding grid (material), and the corrosion factors may include at least one of the following: soil resistivity, water content, gas content, pH, redox potential, soluble salt content, chloride ion content and sulfate ion content. The first model parameter can be used to characterize the influence of each corrosion factor in the grounding grid corrosion prediction model, that is, the weight of each corrosion factor on the corrosion effect of the grounding grid.
[0043] In the embodiment of the present invention, the sample data set A of each year j can be obtained. jThe first model parameters of the grounding grid corrosion prediction model are trained to obtain the grounding grid corrosion prediction basic model and the first model parameters corresponding to year j. Specifically, the first model parameters of the grounding grid corrosion prediction model can be trained using the sample data set A1 of the first year (i.e., j=1) to obtain the grounding grid corrosion prediction model and the first model parameters of the first year, and then the sample data set A1 of the j (j>1) year can be used to train the first model parameters of the grounding grid corrosion prediction model. j The first model parameter of the grounding grid corrosion prediction model for year j-1 is trained to obtain the trained grounding grid corrosion prediction basic model and the first model parameter for year j. It should be understood that this embodiment does not specifically limit the training method of the model, and the existing general neural network training process can be used for training. For example, the grounding grid corrosion prediction model and the first model parameter can be iteratively trained through the forward propagation process and the back propagation process in sequence.
[0044] S130. The second model parameters of the grounding grid corrosion prediction basic model are trained according to the sample data set of each year and the first model parameters to obtain the final grounding grid corrosion prediction model and the second model parameters of the corresponding year; the second model parameters are used to characterize the classification of each corrosion factor.
[0045] Among them, the second model parameter can be used to characterize the classification of each corrosion factor, that is, the classification of the corrosion score corresponding to each corrosion factor is subdivided. Taking the corrosion factor of soil resistivity (unit: Ω·m) as an example, and assuming that the classification is divided into 3 levels, 3 value intervals and corresponding associated corrosion scores can be set. For example, it can be set as follows: when the soil resistivity is ≥50, the corresponding corrosion score is 1; when the soil resistivity is within (20, 50), the corresponding corrosion score is 2; when the soil resistivity is ≤20, the corresponding corrosion score is 3.
[0046] In an embodiment of the present invention, the grounding grid corrosion prediction (basic) model may include a first model parameter and a second model parameter. In S120, the second model parameter will be frozen (using the initial default), and only the first model parameter will be trained and optimized to obtain a grounding grid corrosion prediction basic model after one training and the best first model parameter; and in S130, the first model parameter (after one training) is frozen, and only the second model parameter is trained and optimized to obtain the final grounding grid corrosion prediction model. Specifically, taking the training process of the grounding grid corrosion prediction basic model for year j as an example, the sample data set A corresponding to that year can be jThe data is input into the basic model for grounding grid corrosion prediction for continuous iterative training. At the same time, this process will continuously optimize and correct the parameters of the second model until the model training termination conditions are met, such as: reaching the preset number of training rounds, the loss function value is lower than the preset threshold, etc., and then the final grounding grid corrosion prediction model is obtained. Subsequently, the grounding grid corrosion prediction and risk warning can be realized based on the grounding grid corrosion prediction model.
[0047] The technical solution of the embodiment of the present invention obtains the grounding grid corrosion prediction model to be trained and the sample data sets of different years; trains the first model parameters of the grounding grid corrosion prediction model according to the sample data sets of each year, and obtains the grounding grid corrosion prediction basic model and the first model parameters of the corresponding year; the first model parameters are used to characterize the influence degree of each corrosion factor in the grounding grid corrosion prediction model; trains the second model parameters of the grounding grid corrosion prediction basic model according to the sample data sets of each year and the first model parameters, and obtains the final grounding grid corrosion prediction model and the second model parameters of the corresponding year; the second model parameters are used to characterize the classification of each corrosion factor. This technical solution adopts model training of sample data in two dimensions, and continuously trains and corrects the first model parameters and the second model parameters respectively, thereby effectively improving the model accuracy of the grounding grid corrosion prediction model, thereby improving the accuracy and efficiency of grounding grid corrosion prediction, and simplifying the detection process of grounding grid corrosion.
[0048] Embodiment 2
[0049] Figure 2 The flowchart of a training method for a grounding grid corrosion prediction model provided in the second embodiment of the present invention is further optimized and expanded based on the above implementation, and can be combined with various optional technical solutions in the above implementation. Figure 2 As shown, the training method of a grounding grid corrosion prediction model provided in the second embodiment specifically includes the following steps:
[0050] S210, obtaining a grounding grid corrosion prediction model to be trained and sample data sets of different years.
[0051] In an embodiment of the present invention, a grounding grid corrosion prediction model can be constructed based on corrosion factor X, corrosion factor Y, corrosion score Z, weight coefficient k, corrosion coefficient a and equivalent resistivity ρ.
[0052] Among them, the matrix relationship corresponding to the grounding grid corrosion prediction model is shown in the following table:
[0053]
[0054] Among them, the threshold value (boundary value Y1 and Y2) of the corrosion factor Y and the original value of the corrosion score Z are shown in the table above. The classification of each corrosion factor is divided into 3 levels. The weight coefficient k can be obtained through the sample training data in the subsequent steps. The corrosion coefficient a can be calculated through the relationship between the various parameters of the grounding grid corrosion prediction model given in the table above.
[0055] The annual corrosion coefficient a represents the rate and degree of grounding grid corrosion, which can be further expressed by the change in the equivalent resistivity value, that is, ρ j =ρ(a j )(j is the natural year, ρ j is the equivalent resistivity value of the grounding grid material in the jth natural year, and ρ0 (i.e. j=0) represents the equivalent resistivity of the grounding grid material before corrosion).
[0056] In the embodiment of the present invention, the process of collecting the sample data set specifically includes:
[0057] Take natural year j as the cycle and collect sample data for each natural year. The value of j should be ≥5, and the sample data set collected for each natural year is marked as A j , i.e., the sample data set A1 for the first natural year, the sample data set A2 for the second natural year, and so on. In addition, the sample data set for the first 3 years of service must meet the requirement that the annual sample volume is not less than 50, of which the sample data with the annual corrosion score Z≠1 is not less than 10; the sample data set for 4-5 years of service must meet the requirement that the annual sample volume is not less than 80, of which the sample data with the annual corrosion score Z≠1 is not less than 20; the sample data set for more than 5 years of service must meet the requirement that the annual abnormal sample sampling is not less than 20. It should be understood that the sample data volume mentioned above is only an example, and the specific value can be set accordingly according to the actual application scenario, which is not limited in this embodiment.
[0058] In this embodiment, the grounding grid material can be galvanized round steel. The sample data set of each year includes the following observation data: equivalent resistivity observation value ρ and sampling values of each corrosion factor (X1-X8). Among them, the corrosion factor X is the average value of the periodic sampling results within a natural year; the equivalent resistivity ρ can be obtained by testing the resistance R of the galvanized round steel, and the calculation constraints are: Where l is the test length of the round steel resistance, S0 is the original cross-sectional area of the round steel, and b is the correction coefficient of the round steel test area. Figure 3 As shown, when the test area is a round steel area ( Figure 3 AB and CD areas in the test), the coefficient b = 1; when the test area is the lap weld area ( Figure 3 The value of coefficient b can be calculated by the original resistance R0 of the uncorroded round steel.
[0059] For a certain grounding grid structure (completed by many galvanized round steel overlap welding), when testing the resistance of galvanized round steel in this scenario, the randomly sampled resistance test data must include the round steel area and the overlap welding area. Among them, the resistance test points in each scenario shall not be less than 5, and the resistance test points in the overlap welding area shall not be less than 3. Finally, the equivalent resistivity maximum value max{ρ m}(m represents the number of test points), recorded as the equivalent resistivity observation value ρ at this moment and in this scenario.
[0060] This solution does not require large-scale grounding grid excavation when acquiring sample data. It only needs to test the resistance of a few sections of round steel (including the lap welding area) in the sampling to complete the collection of sample data. This method has less damage to the existing grounding grid and is easy to apply in engineering.
[0061] S220, removing abnormal data from the sample data set of each year, and dividing the sample data set after removing the abnormal data into a training sample set and a verification sample set according to a preset data set segmentation ratio.
[0062] In the embodiment of the present invention, after obtaining the sample data set A of each year, j After that, we first need to remove the abnormal sample data in each year, that is, to weaken the disturbance effect brought by the sample data with corrosion score Z≠1, so as to ensure that a more accurate weight coefficient k can be obtained after subsequent model training. The specific abnormal data removal process is as follows: In the process of removing abnormal sample data, the equivalent resistivity ρ result in the sample data set can be used as the criterion for removing abnormal data. The removal condition of abnormal sample data is Among them, ρ i,j A is the sample dataset of year j j The equivalent resistivity observation value contained in the i-th sample data, A is the sample dataset of year j j The average value of the equivalent resistivity observations contained in all sample data, S ρ is the error result, which can be expressed as n is the sample data set A in year j j By eliminating the above abnormal sample data, we can get a new sample data set A corresponding to each year. 1j .
[0063] Then, the new sample dataset A is split into two datasets according to the preset dataset split ratio of 8:2. 1j Divided into 0.8A 1j The training sample set and 0.2A 1j The validation sample set.
[0064] S230. Input the sample data in the training sample set into the grounding grid corrosion prediction model for continuous iterative training until the obtained grounding grid corrosion prediction basic model minimizes the equivalent resistivity prediction error obtained by verifying the verification data in the verification sample set, and at the same time obtains the optimal weight coefficient corresponding to the grounding grid corrosion prediction basic model.
[0065] In the embodiment of the present invention, Figure 4 As shown, it can be based on 0.8A 1j The training sample set and 0.2A 1j The weight coefficient k (k1-k8) of the grounding grid corrosion prediction model for the corresponding year j is trained based on the validation sample set to optimize and correct the weight coefficient k. Specifically, the corrosion factor Y and corrosion score Z originally set in the above table can be used by default. First, the 0.8A corresponding to the first year (i.e., j = 1) is used. 1j The training sample set and 0.2A 1j The weight coefficient of the initial grounding grid corrosion prediction model to be trained is trained and verified, and the grounding grid corrosion prediction model and weight coefficient of the first year are obtained. Then, the 0.8A corresponding to the j (j>1) year is used in turn. 1j The training sample set and 0.2A 1j The grounding grid corrosion prediction model and weight coefficient for year j-1 are trained and verified based on the validation sample set, so as to obtain the trained grounding grid corrosion prediction basic model for year j and the optimal weight coefficient k.
[0066] Taking year j as an example, the training process of the grounding grid corrosion prediction model and weight coefficient is described as follows: 1j The training sample set can be represented as a data set Q = {(j, ρ(j)), ...}. In the identification and optimization of the weight coefficient k, a mathematical relationship including the intermediate hidden layer is established: H = f(W h ·X+b h ), after 0.8A 1j Iterative training of the training sample set to obtain the model prediction result output Among them, W h and b h are two key parameters that need to be identified in the hidden layer. is the predicted value of equivalent resistivity output by the model, f represents the activation function, such as Sigmoid activation function, etc., W h and W ρ represents the weight of the model during training, b h and b ρ Represents the bias term.
[0067] Then use 0.2A 1jThe grounding grid corrosion prediction model trained above is verified by the validation sample set, and the weight coefficient k is corrected again through residual evaluation until the obtained grounding grid corrosion prediction basic model makes the equivalent resistivity prediction error obtained by validating the validation data in the validation sample set minimum, thus obtaining the grounding grid corrosion prediction basic model after one training and the corresponding optimal weight coefficient. The specific model verification process is as follows: First, 0.2A 1j The validation sample set is input into the trained ground grid corrosion prediction model, and the residual y between the equivalent resistivity prediction value output by the model and the equivalent resistivity observation value in the original sample data set is determined. j , and the residual sequence P = {(j,y j ),...} are analyzed for their size and distribution; then, the residual sequence P is divided into a residual training set and a residual validation set, where the residual training set is used for hyperparameter optimization, and the residual validation set is used to evaluate the quality of the modified model; the iterative solutions for residual training and residual validation are as follows: P next =P i +step rand(0,1), where step is the step size and rand(0,1) is the random function. These two parameters can be used to construct the direction control of the iterative solution.
[0068] Pass above 0.8A 1j Iterative training of training sample sets and 0.2A 1j By correcting the model of the validation sample set, the basic model for predicting grounding grid corrosion in the corresponding year and the weight coefficient k can be established. Combined with the constraint relationship between the equivalent resistivity ρ and the corrosion coefficient a, the corrosion characteristic characterization relationship ρ(a,j) or ρ(X1,X2,X3,X4,X5,X6,X7,X8,j) in natural year j can be established.
[0069] S240. According to the classification of each corrosion factor, the sample data set is divided into a first-class sample set, a second-class sample set, and a third-class sample set.
[0070] In the embodiment of the present invention, the sample data set A can be classified into the following categories according to the classification of each corrosion factor shown in the above table, that is, the original corrosion factor Y and corrosion score Z in the table, according to max{Z}=1, max{Z}=2 and max{Z}=3: j Divided into a sample set A j1 , two-class sample set A j2 And three types of sample set A j3 , where a sample set A j1 It will be mainly used to verify the basic model and calibrate the corrosion characteristics of normal samples. j2The judgment thresholds (Y1 and Y2) for the corrosion factor Y and the corrosion score parameter Z i2 Correction; Three types of sample sets A j3 Will be used for the corrosion score parameter Z i3 correction.
[0071] S250, according to the equivalent resistivity observation values corresponding to each sample data in the first sample set, the second sample set and the third sample set, respectively obtain the corresponding first model observation result set, the second model observation result set and the third model observation result set.
[0072] In the embodiment of the present invention, Figure 5 As shown, a class of sample sets A can be extracted respectively j1 , two-class sample set A j2 And three types of sample set A j3 The equivalent resistivity observation value corresponding to each sample data in , thus obtaining a class of sample set A j1 The corresponding first model observation set ρ j1 , two-class sample set A j2 The corresponding second model observation set ρ j2 And three types of sample sets A j3 The corresponding third model observation set ρ j3 .
[0073] S260, determining a first error between the first model observation result set and the equivalent resistivity prediction value corresponding to a class of sample sets, and taking the sample data in the class of sample sets that meet the first error as a normal sample set, and taking the sample data in the class of samples that do not meet the first error as a low-risk sample set.
[0074] In the embodiment of the present invention, Figure 5 As shown, for a class of sample set A j1 For this scenario, the corrosion score Z is 1. Combined with the above-mentioned grounding grid corrosion prediction basic model ρ(a, j), the equivalent resistivity prediction value output by the model can be obtained: The equivalent resistivity predicted value As a reference value, we can get its difference with the first model observation result set ρ j1 The first error between
[0075] Then, using the error criterion: Perform error analysis and abnormal data extraction, where: A is a sample set j1 The equivalent resistivity observation value corresponding to the i-th sample data in , n is a class of sample set A j1The total number of samples included. Through the above error criterion, a class of sample set A j1 Divided into normal sample set A j4 (i.e., satisfying the error criterion) and low-risk sample set A j5 (i.e. the error criterion is not met).
[0076] S270, taking the average value of the equivalent resistivity observation value corresponding to each sample data in the normal sample set as the normal sample corrosion calibration value, and determining the value range of each corrosion factor based on the normal samples.
[0077] In the embodiment of the present invention, after obtaining the normal sample set A j4 Afterwards, the corresponding normal sample corrosion calibration value can be obtained And the value range of each corrosion factor X in the normal sample (i.e., maximum value + minimum value). Specifically, the normal sample set A can be used j4 The equivalent resistivity observation value ρ corresponding to each sample data j4 The average value of the normal sample corrosion calibration value is determined Expressed as n is the normal sample set A j4 The total number of samples included; then, for the normal sample set A j4 Error analysis is performed to determine the value range (i.e. maximum value + minimum value) of each corrosion factor X corresponding to the normal sample.
[0078] S280. Taking the normal sample corrosion calibration value as a reference value, determine the second error corresponding to the second class sample set and the third error corresponding to the third class sample set, and divide the second class sample set and the third class sample set into a small error sample set, a high risk second class sample set and a high risk third class sample set according to the second error and the third error.
[0079] In the embodiment of the present invention, the normal sample corrosion calibration value can be As the second class sample set A j2 And three types of sample set A j3 The reference value of the normal sample set and the low-risk sample set is then determined by referring to the division method of the normal sample set and the low-risk sample set in S260. j2 And three types of sample set A j3 The corresponding second error and third error are then used to divide the second and third class sample sets into small error sample sets A j6 , High-risk Class II Sample Set A j7 and high risk three-category sample set A j8 .
[0080] Further, based on the above-mentioned embodiment of the invention, S280 specifically includes the following steps:
[0081] S2801, respectively determining a second error between the normal sample corrosion calibration value and the equivalent resistivity observation value corresponding to the second type of sample set, and determining a third error between the normal sample corrosion calibration value and the equivalent resistivity observation value corresponding to the third type of sample set;
[0082] S2802. The sample data that meet the second error in the second category sample set are regarded as the small error sample set, the sample data that do not meet the second error in the second category sample set are regarded as the high-risk second category sample set, and the sample data that do not meet the third error in the third category sample set are regarded as the high-risk third category sample set.
[0083] Specifically, in the second model observation set ρ j2 and the third model observation set ρ j3 In the error analysis, the normal sample corrosion calibration value of year j can be As reference values, the following error criteria are used: as well as The second class sample set A j2 Divide into small error sample set A j6 (i.e., satisfying the error criterion) and the high-risk second-class sample set A j7 (i.e., does not meet the error criterion), and the three-class sample set A j3 Divided into three types of high risk sample set A j8 (i.e. the error criterion is not met).
[0084] S290. The corrosion factor is corrected using the low-risk sample set and the small-error sample set, and the corrosion score is corrected using the high-risk Class II sample set and the high-risk Class III sample set, thereby obtaining a trained grounding grid corrosion prediction model.
[0085] In the embodiment of the present invention, continue to refer to Figure 5 , we can use the low-risk sample set A j5 Correct the judgment threshold Y1 in the corrosion factor Y; use the small error sample set A j6 Modify the judgment thresholds Y1 and Y2 in the corrosion factor Y (if the number of judgment thresholds is greater than 2, modify the multiple judgment thresholds); use the high-risk two-category sample set A j7 and high risk three-category sample set A j8 The corrosion score parameter Z i2 and Z i3 It should be understood that the corrosion score parameter Z i1=1 as the model reference value without correction. In addition, this embodiment does not impose any specific restrictions on the correction method of the above parameters, for example, statistical analysis and other methods are used to achieve parameter correction. Taking the correction method of soil resistivity X1 corresponding to judgment threshold Y1 as an example, the low-risk sample set A can be j5 The minimum value of each soil resistivity sampling value is used as the judgment threshold Y1 after a correction, and then the small error sample set A is determined. j6 The minimum value of each soil resistivity sampling value is then taken as the average value of the minimum value and the previously corrected judgment threshold value Y1 as the final corrected judgment threshold value Y1.
[0086] After the secondary training of the grounding grid corrosion prediction model, that is, after the corrosion factor Y and the corrosion score Z are corrected, the grounding grid corrosion prediction model under any natural year can be obtained, which can be used to predict the corrosion degree of the grounding grid in actual projects, and the risk sample data A can be realized. j5 (Low Risk), A j7 (Medium Risk) and A j8 (High risk) warning output.
[0087] At the same time, the trained grounding grid corrosion prediction model supports dynamic value assignment correction of random sampling. That is, in the actual application of the grounding grid corrosion prediction model, the prediction model can be verified at any time using the sampled historical sample data or the newly collected sample data, and the weight coefficient, corrosion factor and corrosion score corresponding to the prediction model can be dynamically corrected to achieve the update and iteration of the prediction model, thereby improving the prediction accuracy of the grounding grid corrosion prediction model.
[0088] The training method of the grounding grid corrosion prediction model provided in this embodiment uses the change of equivalent resistivity value to characterize the grounding grid corrosion characteristics, and establishes the weight relationship between corrosion factors and corrosion effects by collecting the evolution law of round steel resistance during the service life of the existing grounding grid; at the same time, through the classification and training of annual sample data, the factor division criteria of the annual round steel corrosion evolution are refined, and a rapid corrosion hazard investigation method and criterion suitable for engineering applications are established. This method can accurately predict the degree of corrosion of round steel samples in that year by analyzing the annual corrosion factors and their weight factors; at the same time, in the establishment and parameter optimization of the model, it does not need to destroy the existing grounding grid connection method. Based on the training of sample data, it can realize the online early warning of grounding grid round steel corrosion, which is easy to quickly investigate the grounding grid corrosion in engineering applications.
[0089] The technical solution of the embodiment of the present invention has the following beneficial effects:
[0090] ① A precise training method for the corrosion prediction model of galvanized round steel of grounding grid is proposed. The model is trained with sample data of two dimensions, which can realize the rapid and accurate prediction and correction of corrosion factors and their weight coefficients.
[0091] ② In terms of the processing of training data, it is more targeted. For the multi-objective parameter optimization problem of corrosion factors in multiple dimensions, the sample classification processing is carried out. The designed parameter optimization method is more targeted to ensure the accuracy of the algorithm and the reliability of engineering application;
[0092] ③ In-depth mining of sample data. Through targeted sample data classification, more optimized sample training result output can be obtained. Only the resistance sample data of the grounding grid area needs to be collected to establish a high-reliability corrosion model under small sample collection;
[0093] ④ Based on a natural year round steel corrosion prediction method, it is easy to collect and update sample data, output high-precision prediction results, and online early warning of high-risk samples;
[0094] ⑤ Based on random sampling feedback in the project, the model and parameters can be updated and iterated, which is easy for engineering applications of grounding grid corrosion prediction and early warning.
[0095] Embodiment 3
[0096] Figure 6 The flowchart of a grounding grid corrosion prediction method provided in the third embodiment of the present invention is applicable to the case of predicting the corrosion of the grounding grid of the power steel tower. The method can be executed by a grounding grid corrosion prediction device, which can be implemented in the form of hardware and / or software. The grounding grid corrosion prediction device can be configured in an electronic device, which can include but is not limited to computers, computers, terminals, servers and other devices with data processing capabilities. Figure 6 As shown, a grounding grid corrosion prediction method provided in this embodiment 3 specifically includes the following steps:
[0097] S310, obtaining sample data for the year to be predicted; the sample data includes sampling values of various corrosion factors.
[0098] In an embodiment of the present invention, when corrosion prediction of a power steel tower grounding grid is required, sample data of the year to be predicted may be obtained, for example, sample data of the year j+1 including sampling values of various corrosion factors may be obtained.
[0099] S320, inputting the sample data into a grounding grid corrosion prediction model to obtain a grounding grid corrosion prediction result for the year to be predicted; wherein the grounding grid corrosion prediction model is obtained by the training method of the grounding grid corrosion prediction model of any embodiment of the present invention.
[0100] In an embodiment of the present invention, the sample data of the year to be predicted can be input into the grounding grid corrosion prediction model to obtain the corresponding grounding grid corrosion prediction result. The following takes the grounding grid corrosion prediction of year j+1 as an example to describe the grounding grid corrosion prediction process in detail, wherein the grounding grid corrosion prediction result can be determined in the following two ways, including but not limited to the following two ways: ① Based on the trained grounding grid corrosion prediction model for year j provided in the above embodiment and Figure 5 The flowchart of the secondary model training shown in the figure determines whether the sample data of year j+1 falls into A j5 (Low Risk), A j7 (Medium Risk) and A j8 (High risk) Which of the three sample sets can determine the grounding grid corrosion prediction result corresponding to the sample data; ② The sample data of year j+1 can be input into the grounding grid corrosion prediction model of year j to obtain the equivalent resistivity prediction value output by the model, and at the same time obtain the normal sample corrosion calibration value corresponding to the grounding grid corrosion prediction model of year j And determine the equivalent resistivity prediction value and the normal sample corrosion calibration value The ratio between them is then compared with the preset warning threshold to determine the grounding grid corrosion prediction result corresponding to the sample data. For example, if the ratio is less than 1.5, it is regarded as a low corrosion risk; if the ratio is greater than 1.5 and less than 2, it is regarded as a moderate corrosion risk; if the ratio is greater than 2, it is regarded as a high corrosion risk.
[0101] This embodiment uses the grounding grid corrosion prediction model to predict the degree of grounding grid corrosion, which can achieve online early warning of grounding grid round steel corrosion, improve the accuracy and efficiency of grounding grid corrosion prediction, and facilitate the rapid investigation of grounding grid corrosion in engineering applications, thereby ensuring the normal operation of power equipment and the personal safety of power operation and maintenance personnel.
[0102] Embodiment 4
[0103] Figure 7 This is a schematic diagram of the structure of a training device for a grounding grid corrosion prediction model provided by the fourth embodiment of the present invention. Figure 7 As shown, the device comprises:
[0104] A prediction model and sample data acquisition module 41 is used to obtain a grounding grid corrosion prediction model to be trained and sample data sets of different years;
[0105] The first model training module 42 is used to train the first model parameters of the grounding grid corrosion prediction model according to the sample data set of each year, and obtain the grounding grid corrosion prediction basic model and the first model parameters of the corresponding year; the first model parameters are used to characterize the influence degree of each corrosion factor in the grounding grid corrosion prediction model;
[0106] The second model training module 43 is used to train the second model parameters of the grounding grid corrosion prediction basic model according to the sample data set of each year and the first model parameters to obtain the final grounding grid corrosion prediction model and the second model parameters of the corresponding year; the second model parameters are used to characterize the classification of each corrosion factor.
[0107] This technical solution adopts model training of sample data in two dimensions to continuously train and correct the first model parameters and the second model parameters respectively, thereby effectively improving the model accuracy of the grounding grid corrosion prediction model, thereby improving the accuracy and efficiency of grounding grid corrosion prediction and simplifying the detection process of grounding grid corrosion.
[0108] Further, based on the above-mentioned embodiment of the invention, the grounding grid corrosion prediction model is constructed based on corrosion factors, corrosion factors, corrosion scores, weight coefficients, corrosion coefficients and equivalent resistivity, and the first model parameters include: weight coefficients; accordingly, the first model training module 42 includes:
[0109] The first sample set division unit is used to remove abnormal data from the sample data set of each year, and divide the sample data set after the abnormality is removed into a training sample set and a verification sample set according to a preset data set division ratio;
[0110] The model training unit is used to input the sample data in the training sample set into the grounding grid corrosion prediction model for continuous iterative training until the obtained grounding grid corrosion prediction basic model minimizes the equivalent resistivity prediction error obtained by verifying the verification data in the verification sample set, and at the same time obtains the optimal weight coefficient corresponding to the grounding grid corrosion prediction basic model.
[0111] Further, based on the above-mentioned embodiment of the invention, the second model parameters include: corrosion factor and corrosion score; the classification of each corrosion factor is divided into 3 levels; accordingly, the second model training module 43 includes:
[0112] The second sample set division unit is used to divide the sample data set into a first-class sample set, a second-class sample set and a third-class sample set according to the classification of each corrosion factor;
[0113] The observation result set determination unit is used to obtain the corresponding first model observation result set, second model observation result set and third model observation result set according to the equivalent resistivity observation values corresponding to each sample data in the first sample set, the second sample set and the third sample set;
[0114] The third sample set division unit is used to determine a first error between the first model observation result set and the equivalent resistivity prediction value corresponding to the first sample set, and respectively regard the sample data in the first sample set that meet the first error as a normal sample set, and regard the sample data in the first sample set that do not meet the first error as a low-risk sample set;
[0115] A calibration unit, used to take the average value of the equivalent resistivity observation value corresponding to each sample data in the normal sample set as the normal sample corrosion calibration value, and determine the value range of each corrosion factor according to the normal sample;
[0116] The fourth sample set division unit is used to use the normal sample corrosion calibration value as a reference value, respectively determine the second error corresponding to the second sample set and the third error corresponding to the third sample set, and divide the second sample set and the third sample set into a small error sample set, a high risk second sample set and a high risk third sample set according to the second error and the third error;
[0117] The parameter correction unit is used to correct the corrosion factor using the low-risk sample set and the small-error sample set, and to correct the corrosion score using the high-risk Class II sample set and the high-risk Class III sample set, so as to obtain the trained grounding grid corrosion prediction model.
[0118] Further, based on the above-mentioned embodiment of the invention, the fourth sample set division unit is specifically used for:
[0119] respectively determining a second error between the normal sample corrosion calibration value and the equivalent resistivity observation value corresponding to the second type of sample set, and determining a third error between the normal sample corrosion calibration value and the equivalent resistivity observation value corresponding to the third type of sample set;
[0120] The sample data that meet the second error in the second category sample set are taken as the small error sample set, the sample data that do not meet the second error in the second category sample set are taken as the high-risk second category sample set, and the sample data that do not meet the third error in the third category sample set are taken as the high-risk third category sample set.
[0121] Further, based on the above-mentioned embodiments of the invention, the sample data set includes: equivalent resistivity observation values and sampling values of various corrosion factors; wherein the equivalent resistivity observation values are determined by the resistance values of the grounding grid materials in different test areas; the corrosion factors include at least one of the following: soil resistivity, water content, gas content, pH, redox potential, soluble salt content, chloride ion content and sulfate ion content.
[0122] The training device for the grounding grid corrosion prediction model provided in the embodiment of the present invention can execute the training method for the grounding grid corrosion prediction model provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.
[0123] Embodiment 5
[0124] Figure 8 This is a schematic diagram of the structure of a grounding grid corrosion prediction device provided in Embodiment 5 of the present invention. Figure 8 As shown, the device comprises:
[0125] The data acquisition module 51 is used to acquire sample data of the year to be predicted; the sample data includes the sampling values of each corrosion factor;
[0126] The corrosion prediction module 52 is used to input sample data into the grounding grid corrosion prediction model to obtain the grounding grid corrosion prediction result for the year to be predicted; wherein the grounding grid corrosion prediction model is obtained by the training method of the grounding grid corrosion prediction model of any embodiment of the present invention.
[0127] This embodiment uses the grounding grid corrosion prediction model to predict the degree of grounding grid corrosion, which can achieve online early warning of grounding grid round steel corrosion, improve the accuracy and efficiency of grounding grid corrosion prediction, and facilitate the rapid investigation of grounding grid corrosion in engineering applications, thereby ensuring the normal operation of power equipment and the personal safety of power operation and maintenance personnel.
[0128] The grounding grid corrosion prediction device provided in the embodiment of the present invention can execute the grounding grid corrosion prediction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.
[0129] Embodiment 6
[0130] Fig. 9 A schematic diagram of the structure of an electronic device 60 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.
[0131] like Fig. 9As shown, the electronic device 60 includes at least one processor 61, and a memory connected to the at least one processor 61, such as a read-only memory (ROM) 62, a random access memory (RAM) 63, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 61 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 62 or the computer program loaded from the storage unit 68 to the random access memory (RAM) 63. In the RAM 63, various programs and data required for the operation of the electronic device 60 can also be stored. The processor 61, the ROM 62, and the RAM 63 are connected to each other via a bus 64. An input / output (I / O) interface 65 is also connected to the bus 64.
[0132] A number of components in the electronic device 60 are connected to the I / O interface 65, including: an input unit 66, such as a keyboard, a mouse, etc.; an output unit 67, such as various types of displays, speakers, etc.; a storage unit 68, such as a disk, an optical disk, etc.; and a communication unit 69, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 69 allows the electronic device 60 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.
[0133] The processor 61 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 61 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 61 executes the various methods and processes described above, such as the training method of the grounding grid corrosion prediction model and the grounding grid corrosion prediction method.
[0134] In some embodiments, the training method of the grounding grid corrosion prediction model and the grounding grid corrosion prediction method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 68. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 60 via the ROM 62 and / or the communication unit 69. When the computer program is loaded into the RAM 63 and executed by the processor 61, one or more steps of the training method of the grounding grid corrosion prediction model and the grounding grid corrosion prediction method described above can be executed. Alternatively, in other embodiments, the processor 61 can be configured to execute the training method of the grounding grid corrosion prediction model and the grounding grid corrosion prediction method in any other appropriate manner (for example, by means of firmware).
[0135] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
[0136] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.
[0137] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0138] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).
[0139] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.
[0140] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.
[0141] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.
[0142] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A training method for a grounding grid corrosion prediction model, characterized in that: The method comprises: Obtain the grounding grid corrosion prediction model to be trained and sample data sets of different years; The first model parameters of the grounding grid corrosion prediction model are trained according to the sample data sets of each year to obtain the grounding grid corrosion prediction basic model and the first model parameters corresponding to the year; the first model parameters are used to characterize the influence degree of each corrosion factor in the grounding grid corrosion prediction model; The second model parameters of the grounding grid corrosion prediction basic model are trained according to the sample data sets of each of the years and the first model parameters to obtain the final grounding grid corrosion prediction model and the second model parameters corresponding to the year; the second model parameters are used to characterize the classification of each of the corrosion factors.
2. The method according to claim 1, characterized in that: The grounding grid corrosion prediction model is constructed based on corrosion factors, corrosion factors, corrosion scores, weight coefficients, corrosion coefficients and equivalent resistivity, and the first model parameters include: weight coefficients; Correspondingly, the first model parameter of the grounding grid corrosion prediction model is trained according to the sample data set of each year to obtain the grounding grid corrosion prediction basic model and the first model parameter corresponding to the year, including: Eliminate abnormal data from the sample data sets of each year, and divide the sample data sets after eliminating abnormal data into a training sample set and a validation sample set according to a preset data set segmentation ratio; The sample data in the training sample set is input into the grounding grid corrosion prediction model for continuous iterative training until the obtained grounding grid corrosion prediction basic model minimizes the equivalent resistivity prediction error obtained by verifying the verification data in the verification sample set, and at the same time, the optimal weight coefficient corresponding to the grounding grid corrosion prediction basic model is obtained.
3. The method according to claim 1, characterized in that: The second model parameters include: corrosion factor and corrosion score; the classification of each corrosion factor is divided into 3 levels; Correspondingly, the second model parameters of the grounding grid corrosion prediction basic model are trained according to the sample data sets of each year and the first model parameters to obtain the final grounding grid corrosion prediction model and the second model parameters corresponding to the year, including: According to the classification of each of the corrosion factors, the sample data set is divided into a first-class sample set, a second-class sample set and a third-class sample set; According to the equivalent resistivity observation values corresponding to each sample data in the first sample set, the second sample set and the third sample set, respectively obtain the corresponding first model observation result set, second model observation result set and third model observation result set; Determine a first error between the first model observation result set and the equivalent resistivity prediction value corresponding to the first type of sample set, and respectively use the sample data in the first type of sample set that meets the first error as a normal sample set, and use the sample data in the first type of sample that does not meet the first error as a low-risk sample set; Taking the average value of the equivalent resistivity observation value corresponding to each sample data in the normal sample set as the normal sample corrosion calibration value, and determining the value range of each corrosion factor according to the normal sample; Taking the normal sample corrosion calibration value as a reference value, determining the second error corresponding to the second sample set and the third error corresponding to the third sample set, and dividing the second sample set and the third sample set into a small error sample set, a high risk second sample set and a high risk third sample set according to the second error and the third error; The corrosion factor is corrected using the low-risk sample set and the small-error sample set, and the corrosion score is corrected using the high-risk Class II sample set and the high-risk Class III sample set, thereby obtaining the trained grounding grid corrosion prediction model.
4. The method according to claim 3, characterized in that The method uses the normal sample corrosion calibration value as a reference value, determines the second error corresponding to the second sample set and the third error corresponding to the third sample set, and divides the second sample set and the third sample set into a small error sample set, a high risk second sample set and a high risk third sample set according to the second error and the third error, including: Respectively determining the second error between the normal sample corrosion calibration value and the equivalent resistivity observation value corresponding to the second type of sample set, and determining the third error between the normal sample corrosion calibration value and the equivalent resistivity observation value corresponding to the third type of sample set; The sample data in the two-category sample set that meets the second error is taken as the small-error sample set, the sample data in the two-category sample set that does not meet the second error is taken as the high-risk two-category sample set, and the sample data in the three-category sample set that does not meet the third error is taken as the high-risk three-category sample set.
5. The method according to claim 1, characterized in that The sample data set includes: equivalent resistivity observation values and sampling values of various corrosion factors; wherein the equivalent resistivity observation values are determined by the resistance values of the grounding grid materials in different test areas; and the corrosion factors include at least one of the following: soil resistivity, water content, gas content, pH, redox potential, soluble salt content, chloride ion content, and sulfate ion content.
6. A method for predicting grounding grid corrosion, characterized in that: The method comprises: Obtaining sample data for the year to be predicted; the sample data includes sampling values of various corrosion factors; The sample data is input into a grounding grid corrosion prediction model to obtain a grounding grid corrosion prediction result for the year to be predicted; wherein the grounding grid corrosion prediction model is obtained by the training method of the grounding grid corrosion prediction model described in any one of claims 1-5.
7. A training device for a grounding grid corrosion prediction model, characterized in that: The device comprises: A prediction model and sample data acquisition module is used to obtain the grounding grid corrosion prediction model to be trained and sample data sets of different years; A first model training module is used to train the first model parameters of the grounding grid corrosion prediction model according to the sample data set of each year, so as to obtain the grounding grid corrosion prediction basic model and the first model parameters corresponding to the year; the first model parameters are used to characterize the influence degree of each corrosion factor in the grounding grid corrosion prediction model; The second model training module is used to train the second model parameters of the grounding grid corrosion prediction basic model according to the sample data set of each year and the first model parameters to obtain the final grounding grid corrosion prediction model and the second model parameters corresponding to the year; the second model parameters are used to characterize the classification of each corrosion factor.
8. A grounding grid corrosion prediction device, characterized in that: The device comprises: A data acquisition module is used to acquire sample data of the year to be predicted; the sample data includes sampling values of various corrosion factors; A corrosion prediction module is used to input the sample data into a grounding grid corrosion prediction model to obtain a grounding grid corrosion prediction result for the year to be predicted; wherein the grounding grid corrosion prediction model is obtained by the training method of the grounding grid corrosion prediction model described in any one of claims 1-5.
9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the training method of the grounding grid corrosion prediction model described in any one of claims 1-5, or execute the grounding grid corrosion prediction method described in claim 6.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the training method of the grounding grid corrosion prediction model described in any one of claims 1-5, or to execute the grounding grid corrosion prediction method described in claim 6 when executed.
