Method for evaluating corrosion state of grounding grid of transformer substation based on hybrid driving of data model and simulation model

Abstract

The invention discloses a transformer substation grounding grid corrosion state evaluation method based on hybrid driving of a data model and a simulation model, which is used for detecting a grounding grid laid in a transformer substation and judging and evaluating the corrosion state of the grounding grid. Comprising the following steps: firstly, performing simulation modeling on a grounding grid in a magnetic field environment under specific conditions by adopting electromagnetic field simulation software; secondly, performing data modeling on the corrosion state of the grounding grid by adopting a neural network based on field magnetic field measurement data; thirdly, fusing the simulation model and the data model; and finally, evaluating the corrosion state of the grounding grid according to the fused model. According to the method, the simulation model and the data model are fused, the problem that evaluation is inaccurate due to the fact that a single model is adopted is solved, in addition, the idea of double-layer fusion of model fusion and result fusion is adopted, and the evaluation accuracy is further improved.

Description

Technical Field

[0001] The invention relates to the electrical field, and more specifically to a method for evaluating the corrosion state of a transformer substation grounding grid based on a hybrid drive of a data model and a simulation model. Background Art

[0002] The assessment of the corrosion status of the substation grounding grid is crucial to ensure the safety of substation equipment and personnel. The existing assessment of the corrosion status of the grounding grid often uses the method of magnetic field simulation, or the method of establishing an assessment model based on the collected magnetic field data. Of these two methods, the simulation method is greatly affected by the complexity and uncertainty of the actual operating environment, resulting in deviations in the analysis and decision-making results, and the more complex the environment, the more obvious the deviation; while the method based on data analysis lacks the support of the operating mechanism of the mechanism model, and cannot accurately express the objective physical laws, resulting in non-universality and unsatisfactory results. Summary of the invention

[0003] In view of the above-mentioned deficiencies in the prior art, this patent proposes a method for evaluating the corrosion status of the substation grounding grid based on a hybrid drive of a data model and a simulation model. First, the electromagnetic field simulation software is used to simulate and model the magnetic field environment of the grounding grid under specific conditions; secondly, based on the on-site magnetic field measurement data, a neural network is used to perform data modeling on the corrosion status of the grounding grid; next, the simulation model and the data model are fused; finally, the corrosion status of the grounding grid is evaluated based on the fused model.

[0004] The technical solution adopted by the present invention is:

[0005] A method for evaluating the corrosion state of a substation grounding grid based on a hybrid drive of a data model and a simulation model, the method being used to detect a grounding grid laid in a substation and to identify and evaluate the corrosion state of the grounding grid; comprising:

[0006] Step 1, inject current into the grounding grid, establish a simulation model of the grounding grid corrosion state, set the grounding grid simulation environment parameters, calculate the magnetic induction intensity distribution, and output the grounding grid corrosion state x1 based on the simulation environment;

[0007] Step 2: Based on the magnetic field measurement data, a data model of the grounding grid corrosion state is established, and the grounding grid corrosion state x2 based on the data environment is output;

[0008] Step 3: weighted fusion of the judgment results of the simulation model and the data model is performed as the result output of the corrosion state of the current grounding network;

[0009] Step 4: Use the fused model to evaluate the corrosion status of the grounding grid.

[0010] The current injected into the ground grid is a sinusoidal excitation current with different frequencies.

[0011] The injected heterodyne sinusoidal excitation current has a frequency range of 20-2000 Hz, an amplitude of 10-20A, and an injection position of the diagonal line.

[0012] The simulation environment parameters include: the number of grounding conductors, the length and diameter of each grounding conductor, the resistivity of the grounding conductor, the buried depth of the grounding conductor and the resistivity of the soil.

[0013] The data model adopts a neural network, takes the collected magnetic field data as a sample set, and establishes a data model for evaluating the corrosion state of the grounding grid through iterative training.

[0014] The collected magnetic field data includes: dividing the surface area of ​​the ground grid into regions and measuring the distribution of magnetic induction intensity.

[0015] The simulation model and the data model both output binary classification results: the ground grid is healthy or unhealthy.

[0016] The weighted fusion of the discrimination results of the simulation model and the data model is performed by using a weight distribution method to obtain the fused ground network discrimination result y:

[0017] y=λ1x1+λ2x2

[0018] Among them, x1 is the output result of the simulation model, λ1 is the weight coefficient of the simulation result; x2 is the output result of the data model, λ2 is the weight coefficient of the data analysis result; λ1+λ2=1.

[0019] The λ1 and λ2 are obtained through multiple measurements.

[0020] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0021] The method proposed in the present invention integrates the simulation model with the data model, which makes up for the problem of inaccurate evaluation caused by using a single model. In addition, the method proposed in the present invention adopts the idea of ​​double-layer fusion of model fusion and result fusion, which further improves the accuracy of the evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is an overall flow chart of the method of the present invention;

[0023] Figure 2 This is a schematic diagram of the fusion principle of the simulation model and the data model of the method of the present invention. DETAILED DESCRIPTION

[0024] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation method of the present invention is described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the invention, so the present invention is not limited by the specific implementation disclosed below.

[0025] like Figure 1 As shown, it is a flow chart of the method of the present invention, a method for assessing the corrosion state of a substation grounding grid based on a hybrid drive of a data model and a simulation model, and the specific implementation steps are as follows:

[0026] Step 1: Simulate and model the magnetic field environment of the grounding grid under specific conditions;

[0027] Wherein, in step 1, based on the electromagnetic field simulation software, the magnetic induction intensity distribution on the ground surface of the grounding grid injected with heterofrequency sinusoidal excitation current is simulated to establish a magnetic field simulation model.

[0028] The electromagnetic field simulation software uses Ansoft finite element software.

[0029] The simulation environment parameters for building the magnetic field simulation model include the number of grounding conductors, the length and diameter of each grounding conductor, the resistivity of the grounding conductor, the buried depth of the grounding conductor and the resistivity of the soil.

[0030] The frequency range of the injected heterodyne sinusoidal excitation current can be selected to be 20-2000 Hz, the amplitude can be selected to be 10-20A, and the injection position is the diagonal line.

[0031] The simulation model provides a physical basis and judgment basis for data analysis for the data model.

[0032] Step 2: Based on the magnetic field measurement data, data modeling is performed on the corrosion state of the grounding grid;

[0033] Among them, in step 2, the magnetic field data collection on site adopts the magnetic field detection device, the environment of the grounding grid is consistent with the simulation environment, and the injection signal condition is also consistent with the simulation environment.

[0034] The data model adopts a neural network, takes the collected magnetic field data as samples, and establishes a data model for evaluating the corrosion state of the grounding grid through training.

[0035] The neural network can adopt a network structure such as CNN or RNN, and the network parameters select a 3-layer network with 64 nodes in each layer.

[0036] The collected magnetic field data includes dividing the surface area of ​​the ground grid into regions and measuring the distribution of magnetic induction intensity. The region division is to divide the ground grid surface into grids or concentric circles or rings along the center of the ground grid and mark the position coordinates.

[0037] The analysis results of the data model can be used to correct the simulation model.

[0038] Step 3: Fusion the simulation model and the data model;

[0039] The specific implementation principle of step 3 is as follows: Figure 2 As shown:

[0040] The fusion of the simulation model and the data model is divided into two levels. The first level is the mutual calibration and correction of the simulation model and the data model; the second level is the fusion of the output results of the simulation model and the output results of the data model.

[0041] Furthermore, the simulation model provides physical constraints and simulation data samples for the data model, and the data model calibrates the structural parameters of the simulation model.

[0042] Furthermore, the physical constraints provided by the simulation model for the data model include the boundary range of the magnetic field value. The simulated data samples refer to data samples obtained by simulating the corrosion of the conductor, which are used as inputs of the data model to enhance the training effect of the data model.

[0043] The fusion of the simulation model output results and the data model output results is performed by using a weight distribution method. The specific formula is as follows:

[0044] y=λ1x1+λ2x2

[0045] Among them, y is the evaluation result of the final output of the fusion model; x1 is the output result of the simulation model, λ1 is the weight coefficient of the simulation result; x2 is the output result of the data model, λ2 is the weight coefficient of the data analysis result. λ1+λ2=1. λ1 and λ2 are the empirical weight values ​​set according to the actual measurement of the grounding grid.

[0046] Step 4: Evaluate the corrosion status of the grounding grid based on the fused model.

[0047] After the above steps, a double-layer iterative algorithm model can be established, in which the inner layer is the synchronous iterative optimization of the simulation model and the data model, and the outer layer is the fusion analysis of the decision results. On this basis, the corrosion status of the grounding grid can be diagnosed and analyzed.

[0048] The present invention is described above by way of example in conjunction with the accompanying drawings. Obviously, the present invention is not limited to the above embodiments, and may also have many variations. As long as various non-substantial improvements are made using the method concept and technical solution of the present invention, or the concept and technical solution of the invention are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

Claims

1. A method for assessing the corrosion status of a substation grounding grid based on a hybrid drive of a data model and a simulation model, characterized in that: The method is used to detect the grounding grid laid in the substation and to identify and evaluate the corrosion state of the grounding grid; it includes: Step 1, inject current into the grounding grid, establish a simulation model of the grounding grid corrosion state, set the grounding grid simulation environment parameters, calculate the magnetic induction intensity distribution, and output the grounding grid corrosion state x1 based on the simulation environment; Step 2: Based on the magnetic field measurement data, a data model of the grounding grid corrosion state is established, and the grounding grid corrosion state x2 based on the data environment is output; Step 3: weighted fusion of the judgment results of the simulation model and the data model is performed as the result output of the corrosion state of the current grounding network; Step 4: Use the fused model to evaluate the corrosion status of the grounding grid.

2. According to claim 1, a method for assessing the corrosion status of a substation grounding grid based on a hybrid drive of a data model and a simulation model, characterized in that: The current injected into the ground grid is a sinusoidal excitation current with different frequencies.

3. According to claim 2, a method for assessing the corrosion status of a substation grounding grid based on a hybrid drive of a data model and a simulation model is characterized in that: The injected heterodyne sinusoidal excitation current has a frequency range of 20-2000 Hz, an amplitude of 10-20A, and an injection position of the diagonal line.

4. According to claim 1, a method for assessing the corrosion status of a substation grounding grid based on a hybrid drive of a data model and a simulation model is characterized in that: The simulation environment parameters include: the number of grounding conductors, the length and diameter of each grounding conductor, the resistivity of the grounding conductor, the buried depth of the grounding conductor and the resistivity of the soil.

5. According to claim 1, a method for assessing the corrosion status of a substation grounding grid based on a hybrid drive of a data model and a simulation model, characterized in that: The data model adopts a neural network, takes the collected magnetic field data as a sample set, and establishes a data model for evaluating the corrosion state of the grounding grid through iterative training.

6. The method for assessing the corrosion status of a substation grounding grid based on a hybrid drive of a data model and a simulation model according to claim 1, characterized in that: The collected magnetic field data includes: dividing the surface area of ​​the ground grid into regions and measuring the distribution of magnetic induction intensity.

7. The method for assessing the corrosion status of a substation grounding grid based on a hybrid drive of a data model and a simulation model according to claim 1, characterized in that: The simulation model and the data model both output binary classification results: the ground grid is healthy or unhealthy.

8. The method for assessing the corrosion status of a substation grounding grid based on a hybrid drive of a data model and a simulation model according to claim 1, characterized in that: The weighted fusion of the discrimination results of the simulation model and the data model is performed by using a weight distribution method to obtain the fused ground network discrimination result y: y=λ1x1+λ2x2 Among them, x1 is the output result of the simulation model, λ1 is the weight coefficient of the simulation result; x2 is the output result of the data model, λ2 is the weight coefficient of the data analysis result; λ1+λ2=1.

9. The method for assessing the corrosion status of a substation grounding grid based on a hybrid drive of a data model and a simulation model according to claim 1, characterized in that: The λ1 and λ2 are obtained through multiple measurements.

Images