Soil inversion and grounding impedance simulation algorithm

Abstract

The invention relates to the field of soil inversion, in particular to a soil inversion and grounding impedance simulation algorithm. The method comprises the steps of collecting hyperspectral remote sensing data and observation data at multiple moments, calculating soil electrical parameters, dividing remote sensing data difficult to invert and normal inversion remote sensing data, and calculating a clustering centroid and a difficult inversion degree of the remote sensing data difficult to invert; constructing a plurality of loss functions according to the difficult-to-invert degree, the clustering centroid and the plurality of soil electrical parameters, and selecting the soil electrical parameter corresponding to the minimum loss function as the optimal soil electrical parameter corresponding to the difficult-to-invert remote sensing data; based on the normal inversion remote sensing data and the corresponding soil electrical parameters, the remote sensing data difficult to invert and the corresponding optimal soil electrical parameters, constructing a soil inversion model by using a neural network model; and inverting electrical parameters of soil at different depths by using the soil inversion model and performing grounding impedance simulation. The problem that in the prior art, a soil inversion model is not accurate is effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of soil inversion, and more particularly to a soil inversion and ground impedance simulation algorithm. Background Art

[0002] In recent years, the design and maintenance of grounding systems in power systems are crucial. As part of the grounding system, the conductivity and structural characteristics of soil directly affect the size of grounding impedance, which in turn affects the stability of the power system. It involves soil inversion to obtain the electrical parameters of the soil. Grounding impedance simulation is based on electrical parameters to evaluate and optimize the design and performance of the grounding system. Soil inversion is to infer the physical and chemical properties of the soil from remote sensing data collected from the air. Grounding impedance simulation is to predict and evaluate the performance and grounding impedance of power equipment or power systems under different conditions through soil data, so as to select the optimal solution to ensure the safety and reliability of the power system.

[0003] At present, the patent document with publication number CN118348552A discloses a long-time image synthesis method for soil remote sensing inversion. This method constructs a synthetic image soil property prediction model through a machine learning algorithm and selects the optimal synthetic image. The patent document with publication number CN117688478A discloses a method for inverting the salinity of cultivated land soil based on multi-cycle classification. This method inverts the salinization of salinized cultivated land soil by constructing a multi-cycle classification and GBDT regression model. The patent document with publication number CN111025024A discloses a method for measuring the grounding impedance of a substation grounding grid through actual measurement and simulation. This method uses finite element software to establish a finite element model of the grounding grid magnetic field to simulate and calculate the grounding impedance of the grounding grid.

[0004] However, the current method has the problem that one remote sensing data may correspond to multiple soil electrical parameters, resulting in inaccurate soil inversion model. Summary of the invention

[0005] In order to solve the problem of inaccurate soil inversion model in the prior art, the present invention provides the following solution.

[0006] In a first aspect, the present invention provides a soil inversion and ground impedance simulation algorithm, comprising: collecting hyperspectral remote sensing data and observation data at multiple moments; calculating soil electrical parameters using the observation data; determining difficult-to-invert remote sensing data and normal inversion remote sensing data based on the soil electrical parameters and the hyperspectral remote sensing data, wherein when the hyperspectral remote sensing data corresponding to multiple different soil electrical parameters are the same, the hyperspectral remote sensing data is difficult-to-invert remote sensing data, and when the hyperspectral remote sensing data corresponding to different soil electrical parameters are different, the hyperspectral remote sensing data is normal inversion remote sensing data; calculating the hyperspectral remote sensing data using a first clustering algorithm; The cluster centroid of the difficult-to-invert remote sensing data is obtained, and the difficulty of inversion of the difficult-to-invert remote sensing data is calculated using a second clustering algorithm; multiple loss functions are constructed according to the difficulty of inversion, the cluster centroid and the multiple soil electrical parameters, and the soil electrical parameters corresponding to the minimum loss function are selected as the optimal soil electrical parameters corresponding to the difficult-to-invert remote sensing data; based on the normal inversion remote sensing data and the corresponding soil electrical parameters, the difficult-to-invert remote sensing data and the corresponding optimal soil electrical parameters, a soil inversion model is constructed using a neural network model; the soil electrical parameters at different depths are inverted using the soil inversion model and a ground impedance simulation is performed.

[0007] Beneficial effect: The present invention introduces the difficulty of inversion degree, and for multiple soil electrical parameters corresponding to the difficult-to-invert remote sensing data, multiple loss functions are constructed according to the difficulty of inversion degree, cluster centroid and the multiple soil electrical parameters. More reasonable data can be selected from the multiple soil electrical parameters, thereby constructing a more accurate soil inversion model, and then being able to simulate a more accurate ground impedance.

[0008] Preferably, the acquisition of hyperspectral remote sensing data and observation data at multiple moments includes: the observation data includes temperature, humidity, soil density and pH value of soil at different depths; Beneficial effects: Soil electrical parameters are affected by many factors. Collecting the temperature, humidity, soil density and pH value of soil at different depths can calculate more accurate soil electrical data.

[0009] Preferably, the soil electrical parameters include: soil conductivity and dielectric constant.

[0010] Preferably, the use of a first clustering algorithm to calculate the cluster centroid of the difficult-to-invert remote sensing data includes: clustering a plurality of different soil electrical parameters corresponding to the difficult-to-invert remote sensing data, setting the number of clusters to 1, and calculating the cluster centroid.

[0011] Preferably, the soil inversion and ground impedance simulation algorithm according to claim 2 is characterized in that the use of the second clustering algorithm to calculate the difficulty of inversion of the difficult-to-invert remote sensing data includes: The calculation formula of the difficulty inversion degree H is: ,in, is the aggregation degree corresponding to the difficult-to-invert remote sensing data, is the uncertainty of the fit, and is the weight coefficient, for The threshold value of the fit uncertainty The calculation formula is: , , where N represents the number of soil electrical parameters corresponding to the difficult-to-invert remote sensing data, represents the fitting degree of the ith electrical parameter among the N soil electrical parameters corresponding to the difficult-to-invert remote sensing data, represents the soil conductivity in the i-th soil electrical parameter, represents the mean value of soil conductivity of N soil electrical parameters, represents the dielectric constant of the i-th soil electrical parameter, represents the mean dielectric constant of N soil electrical parameters, It represents the minimum value of the fitting degree of each soil electrical parameter corresponding to the difficult-to-invert remote sensing data.

[0012] Beneficial effect: The calculation formula of the difficulty of inversion includes the aggregation degree and fit uncertainty , aggregation It reflects the influence of the distribution of multiple different soil electrical parameters on the difficulty of inversion and the uncertainty of fitting. It reflects the influence of the fitting degree of multiple different soil electrical data and cluster centroid on the difficulty of inversion. The difficulty of inversion H is calculated by combining the two aspects, which can more comprehensively reflect the difficulty of inversion of difficult-to-invert remote sensing data.

[0013] Preferably, the method further includes the step of preprocessing the hyperspectral remote sensing data and observation data collected at multiple moments, specifically: performing radiation correction and geometric correction on the hyperspectral remote sensing data; and completing and correcting the observation data.

[0014] Preferably, the is the aggregation degree corresponding to the difficult-to-invert data, including: Calculate the degree of clustering using the K-means clustering algorithm , ,in, Indicates the number of soil electrical parameters corresponding to the difficult-to-invert remote sensing data, is the number of clusters, It is Soil electrical parameters of clusters, Indicates The soil conductivity of the cluster center, yes The i-th soil electrical parameter The soil conductivity, Indicates The dielectric constant of the cluster center is Is a collection The i-th soil electrical parameter The dielectric constant of .

[0015] Beneficial effect: Aggregation The calculation formula includes two dimensions: soil conductivity and dielectric constant, which can make the aggregation The calculation results are more accurate and comprehensive.

[0016] Preferably, the multiple loss functions are constructed according to the difficulty of inversion, the cluster centroid and the multiple soil electrical parameters, including: loss function ,in, represents the loss function of the ith soil electrical parameter among the multiple different soil electrical parameters corresponding to the difficult-to-invert remote sensing data, represents the dielectric constant of the cluster centroid, is the conductivity of the cluster centroid, is the dielectric constant of the i-th soil electrical parameter, is the conductivity of the i-th soil electrical parameter.

[0017] The beneficial effect of the present invention is that: this scheme introduces the difficulty of inversion, and for multiple soil electrical parameters corresponding to the difficult-to-invert remote sensing data, multiple loss functions are constructed according to the difficulty of inversion, cluster centroids and the multiple soil electrical parameters. More reasonable data can be selected from the multiple soil electrical parameters, thereby constructing a more accurate soil inversion model. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 It is a flow chart of a soil inversion and ground impedance simulation algorithm provided by an embodiment of the present invention; DETAILED DESCRIPTION The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0019] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0020] Figure 1 A flow chart of a soil inversion and ground impedance simulation algorithm according to an embodiment of the present invention includes the following steps: S101, collecting hyperspectral remote sensing data and observation data at multiple times; Hyperspectral remote sensing data refers to remote sensing data obtained through hyperspectral imaging technology. The sensor captures electromagnetic waves from the surface and divides the spectral information into dozens or even hundreds of narrow bands to perform detailed spectral analysis of the target area.

[0021] In some embodiments, the step of preprocessing the hyperspectral remote sensing data and observation data collected at multiple times is also included, specifically: performing radiation correction and geometric correction on the hyperspectral remote sensing data; completing and correcting the observation data. When collecting hyperspectral remote sensing data, the sensor may be affected by solar radiation, atmospheric transmission, platform attitude, and earth curvature, resulting in data errors. Therefore, radiation correction and geometric correction are required to improve the accuracy of hyperspectral remote sensing data so that it can more accurately reflect the true surface conditions. Observation data are often prone to outliers and missing values, so it is necessary to complete and correct the observation data. The completion methods include interpolation, regression analysis, sequence prediction, and machine learning. The correction methods include noise removal, statistics, and machine learning. It should be noted that the specific radiation correction, geometric correction, data completion, and correction methods belong to the known prior art and will not be repeated here.

[0022] S102, calculating soil electrical parameters using the observed data; In some embodiments, the observation data include temperature, humidity, soil density and pH value of soil at different depths.

[0023] In some embodiments, the soil electrical parameters include soil conductivity and dielectric constant.

[0024] The observed data can be used to calculate soil electrical parameters at different depths. For example, the soil volume water content can be calculated based on soil moisture and soil density. , using the empirical formula Calculating the Dielectric Constant of Soil , using the empirical formula , calculate the soil conductivity , is the basic conductivity, T is the soil temperature, and the salt concentration can be calculated from the soil pH value. , , is the empirical coefficient.

[0025] S103, determining the difficult-to-invert remote sensing data and the normal-to-invert remote sensing data according to the soil electrical parameters and the hyperspectral remote sensing data, wherein when the hyperspectral remote sensing data corresponding to a plurality of different soil electrical parameters are the same, the hyperspectral remote sensing data is the difficult-to-invert remote sensing data, and when the hyperspectral remote sensing data corresponding to different soil electrical parameters are different, the hyperspectral remote sensing data is the normal-to-invert remote sensing data; S104, using a first clustering algorithm to calculate the cluster centroid of the difficult-to-invert remote sensing data, and using a second clustering algorithm to calculate the degree of difficulty of inversion of the difficult-to-invert remote sensing data; In some embodiments, the use of a first clustering algorithm to calculate the cluster centroid of the difficult-to-invert remote sensing data includes: clustering a plurality of different soil electrical parameters corresponding to the difficult-to-invert remote sensing data, setting the number of clusters to 1, and calculating the cluster centroid. In general, the K-means clustering algorithm can be used to set K to 1, that is, the plurality of different soil electrical parameters are divided into one cluster, and the cluster centroid of the cluster is calculated, and the cluster centroid refers to the arithmetic mean of all soil electrical parameters in one cluster. It should be noted that the specific K-means clustering algorithm for calculating the cluster centroid is a well-known technology and will not be described in detail here.

[0026] In some embodiments, the calculation of the degree of difficulty of inversion of the difficult-to-invert remote sensing data using the second clustering algorithm includes: the calculation formula of the degree of difficulty of inversion H is: ,in, is the aggregation degree corresponding to the difficult-to-invert remote sensing data, is the uncertainty of the fit, and is the weight coefficient, for The threshold value of the fit uncertainty The calculation formula is: , , where N represents the number of soil electrical parameters corresponding to the difficult-to-invert remote sensing data, represents the fitting degree of the ith electrical parameter among the N soil electrical parameters corresponding to the difficult-to-invert remote sensing data, represents the soil conductivity in the i-th soil electrical parameter, represents the mean value of soil conductivity of N soil electrical parameters, represents the dielectric constant of the i-th soil electrical parameter, represents the mean dielectric constant of N soil electrical parameters, It represents the minimum value of the fitting degree of each soil electrical parameter corresponding to the difficult-to-invert remote sensing data. The aggregation degree corresponding to the difficult-to-invert data includes: calculating the aggregation degree using a K-means clustering algorithm , in, Indicates the number of soil electrical parameters corresponding to the difficult-to-invert remote sensing data, is the number of clusters, It is Soil electrical parameters of clusters, Indicates The soil conductivity of the cluster center, yes The i-th soil electrical parameter The soil conductivity, Indicates The dielectric constant of the cluster center is Is a collection The i-th soil electrical parameter The dielectric constant of .

[0027] For example, a remote sensing data that is difficult to invert corresponds to 5 different sets of soil electrical parameters ( )、( )、( )、( )and( ), from which we can get N=5, based on which we can calculate the uncertainty of the fit , , is the mean value of dielectric constant among the five soil electrical parameters, is the mean value of the conductivity among the five soil electrical parameters. Set K=2 in the K-means clustering algorithm and divide the five groups of different electrical parameters into two clusters. For example, the division result is ( )、( )and( ) is the first cluster, ( )and( ) is the second cluster, calculate the distance between all points in each cluster and other points, and select the point with the smallest total distance as the cluster center of the cluster, then we can get 2 cluster centers ( )and( ), so the aggregation The degree of aggregation and fit uncertainty Substitute into the formula The degree of difficulty in inversion H can be calculated, wherein, in some embodiments, when there is only one remote sensing data that is difficult to invert, It can be preset according to the experience value. When there are multiple remote sensing data that are difficult to invert, It can represent the average of multiple aggregation degrees corresponding to the multiple difficult-to-invert remote sensing data.

[0028] The calculation formula of difficulty inversion includes the aggregation degree and fit uncertainty , aggregation It reflects the influence of the distribution of multiple different soil electrical parameters on the difficulty of inversion and the uncertainty of fitting. It reflects the influence of the fitting degree of multiple different soil electrical data and cluster centroid on the difficulty of inversion. The difficulty of inversion H is calculated by combining the two aspects, which can more comprehensively reflect the difficulty of inversion of difficult-to-invert remote sensing data.

[0029] S105, constructing multiple loss functions according to the difficulty inversion degree, the cluster centroid and the multiple soil electrical parameters, and selecting the soil electrical parameter corresponding to the minimum loss function as the optimal soil electrical parameter corresponding to the difficult-to-invert remote sensing data; In some embodiments, the loss function may be constructed according to the following formula: ,in, It represents the loss function of the ith soil electrical parameter among multiple different soil electrical parameters corresponding to the difficult-to-invert remote sensing data, The dielectric constant of the cluster centroid, is the conductivity of the cluster centroid, is the dielectric constant of the ith soil electrical parameter among multiple different soil electrical parameters corresponding to the difficult-to-invert remote sensing data, It is the conductivity of the ith soil electrical parameter among the multiple different soil electrical parameters corresponding to the difficult-to-invert remote sensing data. For example, in the above example, if there are 5 different soil electrical parameters corresponding to the difficult-to-invert remote sensing data, 5 loss function values ​​can be calculated, and the soil electrical parameter corresponding to the purple-red minimum loss function value is selected as the optimal soil electrical parameter corresponding to the difficult-to-invert remote sensing data.

[0030] S106, constructing a soil inversion model using a neural network model based on the normal inversion remote sensing data and the corresponding soil electrical parameters, the difficult inversion remote sensing data and the corresponding optimal soil electrical parameters; After obtaining the difficult-to-invert remote sensing data and the corresponding optimal soil electrical parameters, the normal inversion remote sensing data and the corresponding soil electrical parameters, the difficult-to-invert remote sensing data and the corresponding optimal soil electrical parameters can be made into a training set to train the neural network model, and finally obtain the soil inversion model. It should be noted that there are multiple training methods for training the neural network model, and they belong to the known technology, which will not be repeated here.

[0031] S107, using the soil inversion model to invert soil electrical parameters at different depths and perform ground impedance simulation.

[0032] The observation data collected by the present invention include the temperature, humidity, soil density and pH value of soil at different depths, so the electrical parameters of soil at different depths are also calculated, and the soil inversion model finally trained can also invert the electrical parameters of soil at different depths according to the hyperspectral remote sensing data. A layered soil model can be constructed according to the electrical parameters of soil at different depths, and the layered soil model is input into an existing ground impedance simulation tool such as CDEGS simulation software, and the type of grounding device is set, and the current size and frequency are defined to obtain the final ground impedance.

[0033] The present invention introduces the difficulty of inversion. For multiple soil electrical parameters corresponding to the difficult-to-invert remote sensing data, multiple loss functions are constructed according to the difficulty of inversion, cluster centroids and the multiple soil electrical parameters. More reasonable data can be selected from the multiple soil electrical parameters, thereby constructing a more accurate soil inversion model, and further simulating a more accurate ground impedance.

[0034] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0035] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A soil inversion and ground impedance simulation algorithm, characterized in that: include: Collect hyperspectral remote sensing data and observation data at multiple times; Calculating soil electrical parameters using the observed data; Determine the difficult-to-invert remote sensing data and the normal-to-invert remote sensing data according to the soil electrical parameters and the hyperspectral remote sensing data, wherein when the hyperspectral remote sensing data corresponding to a plurality of different soil electrical parameters are the same, the hyperspectral remote sensing data is the difficult-to-invert remote sensing data, and when the hyperspectral remote sensing data corresponding to different soil electrical parameters are different, the hyperspectral remote sensing data is the normal-to-invert remote sensing data; Calculating the cluster centroid of the difficult-to-invert remote sensing data using a first clustering algorithm, and calculating the degree of difficulty of inversion of the difficult-to-invert remote sensing data using a second clustering algorithm; Constructing multiple loss functions according to the difficulty inversion degree, the cluster centroid and the multiple soil electrical parameters, and selecting the soil electrical parameter corresponding to the minimum loss function as the optimal soil electrical parameter corresponding to the difficult-to-invert remote sensing data; Based on the normal inversion remote sensing data and the corresponding soil electrical parameters, the difficult inversion remote sensing data and the corresponding optimal soil electrical parameters, a soil inversion model is constructed using a neural network model; The soil inversion model is used to invert soil electrical parameters at different depths and perform ground impedance simulation.

2. The soil inversion and ground impedance simulation algorithm according to claim 1, characterized in that: The collecting of hyperspectral remote sensing data and observation data at multiple times includes: The observation data include temperature, humidity, soil density and pH value of soil at different depths.

3. The soil inversion and ground impedance simulation algorithm according to claim 1, characterized in that: The soil electrical parameters include soil conductivity and dielectric constant.

4. The soil inversion and ground impedance simulation algorithm according to claim 1, characterized in that: The method of calculating the cluster centroid of the difficult-to-invert remote sensing data by using the first clustering algorithm includes: clustering a plurality of different soil electrical parameters corresponding to the difficult-to-invert remote sensing data, setting the number of clusters to 1, and calculating the cluster centroid.

5. The soil inversion and ground impedance simulation algorithm according to claim 3, characterized in that: The calculating the degree of difficulty of inversion of the difficult-to-invert remote sensing data by using the second clustering algorithm comprises: The calculation formula of the difficulty inversion degree H is: ,in, is the aggregation degree corresponding to the difficult-to-invert remote sensing data, is the uncertainty of the fit, and is the weight coefficient, for The threshold value of the fit uncertainty The calculation formula is: , , where N represents the number of soil electrical parameters corresponding to the difficult-to-invert remote sensing data, represents the fitting degree of the ith electrical parameter among the N soil electrical parameters corresponding to the difficult-to-invert remote sensing data, represents the soil conductivity in the i-th soil electrical parameter, represents the mean value of soil conductivity of N soil electrical parameters, represents the dielectric constant of the i-th soil electrical parameter, represents the mean dielectric constant of N soil electrical parameters, It represents the minimum value of the fitting degree of each soil electrical parameter corresponding to the difficult-to-invert remote sensing data.

6. The soil inversion and ground impedance simulation algorithm according to claim 1, characterized in that: The method also includes the steps of preprocessing the hyperspectral remote sensing data and observation data collected at multiple times, specifically: Performing radiation correction and geometric correction on the hyperspectral remote sensing data; Complete and correct the observed data.

7. The soil inversion and ground impedance simulation algorithm according to claim 3, characterized in that: Said is the aggregation degree corresponding to the difficult-to-invert data, including: Calculate the degree of clustering using the K-means clustering algorithm , ,in, Indicates the number of soil electrical parameters corresponding to the difficult-to-invert remote sensing data, is the number of clusters, It is Soil electrical parameters of clusters, Indicates The soil conductivity of the cluster center, yes The i-th soil electrical parameter The soil conductivity, Indicates The dielectric constant of the cluster center is Is a collection The i-th soil electrical parameter The dielectric constant of .

8. The soil inversion and ground impedance simulation algorithm according to claim 1, characterized in that: The multiple loss functions are constructed according to the difficulty of inversion, the cluster centroid and the multiple soil electrical parameters, including: loss function ,in, represents the loss function of the ith soil electrical parameter among the multiple different soil electrical parameters corresponding to the difficult-to-invert remote sensing data, represents the dielectric constant of the cluster centroid, is the conductivity of the cluster centroid, is the dielectric constant of the i-th soil electrical parameter, is the conductivity of the i-th soil electrical parameter.

Images