A soil salinity monitoring model construction method, device, system and storage medium

Abstract

The embodiment of the present invention discloses a method, device, system and storage medium for constructing a soil salinity monitoring model. The method includes: S1: obtaining a soil environment dataset {x} and a soil salinity measured dataset {y} in a preset area, and using the soil environment dataset {x} and the soil salinity measured dataset {y} as an initial dataset {x, y}, wherein the soil environment dataset {x} includes meteorological data, soil attribute data, groundwater level data, vegetation cover data and remote sensing data; S2: performing data processing on the initial dataset {x, y} according to a preset method to obtain training samples; S3: using the training samples to train a machine learning model to obtain a soil salinity monitoring model. By calculating a random soil salinity dataset {y} containing a sufficient amount of data, 产生} and soil environment random data set {x 产生} as training samples, and then use the training samples to perform neural network fitting and various machine data learning fitting to obtain the soil salinity monitoring model.

Description

Technical Field

[0001] The present invention relates to the field of computer software technology, and in particular to a soil salinity monitoring model construction method, equipment, system and storage medium. Background Art

[0002] Soil salinization is a global factor hindering agricultural yields. Along with desertification, soil degradation, soil erosion, and soil pollution, it constitutes one of the five major soil problems that have long plagued humanity. Soil salinization typically occurs in areas with arid climates, high evaporation rates, high groundwater levels, and high concentrations of soluble salts. Soil salinization damages arable land resources and causes significant losses in agricultural production. It also poses a threat to the biosphere and ecological environment, and impacts sustainable economic development.

[0003] Traditional destructive sampling methods using soil drills require that samples be collected first and then subjected to indoor experimental analysis. This is not only time-consuming and labor-intensive, but also costly and slow to update dynamically. Remote sensing, on the other hand, provides an effective means for dynamic monitoring of large-scale soil salinization. Soil salinity changes are affected by many factors, such as meteorology (temperature, precipitation), soil properties (water content, organic matter, and texture), groundwater levels, and aboveground vegetation (crops). The influence of a single influencing factor on soil salinity may be nonlinear, and the combined effects of multiple factors on soil salinity changes produce multiple effects. These multiple effects make the relationship between soil salinity and remote sensing spectral characteristics more complex, which poses a great challenge to the dynamic monitoring of soil salinity by remote sensing.

[0004] Currently, due to the limited availability of soil sample datasets, effective parameter estimation is impossible. This is especially true when there are too many attributes, which often leads to overfitting. Furthermore, when the sample dataset is less than 50, it is difficult to create sufficient validation and test sets. Consequently, it is impossible to use machine learning to build a soil salinity monitoring model that meets the required validity. Summary of the Invention

[0005] In response to the technical defects in the existing technology, the purpose of the embodiments of the present invention is to provide a soil salinity monitoring model construction method, equipment, system and storage medium. The soil salinity monitoring model based on inverse neural network fitting can effectively establish the relationship between salinity and remote sensing spectra, produce regional-scale high-precision soil salinity satellite remote sensing monitoring products, realize satellite remote sensing dynamic monitoring of soil salinity, provide scientific basis and technical support for establishing regional-scale soil salinity dynamic forecasting models, and provide data support and decision-making basis for the comprehensive management and efficient utilization of salinized farmland and the sustainable and healthy development of local agriculture.

[0006] To achieve the above objectives, in a first aspect, an embodiment of the present invention provides a method for constructing a soil salinity monitoring model, comprising:

[0007] S1: Obtain a soil environment dataset {x} and a soil salinity measurement dataset {y} within a preset area, and use the soil environment dataset {x} and the soil salinity measurement dataset {y} as an initial dataset {x, y}, wherein the soil environment dataset {x} includes meteorological data, soil property data, groundwater level data, vegetation cover data, and remote sensing data;

[0008] S2: Process the initial data set {x, y} according to a preset method to obtain training samples;

[0009] S3: Use the training samples to train the machine learning model to obtain a soil salinity monitoring model.

[0010] Furthermore, in step S1, a preset number of data are taken from the initial data set {x, y} as verification samples;

[0011] And after step S3, the method further includes:

[0012] S4: Inputting the verification sample into the soil salinity monitoring model to verify the validity of the soil salinity monitoring model.

[0013] Furthermore, the processing of the initial data set {x, y} according to a preset method to obtain training samples includes:

[0014] S21: Take the nth recursive function in the preset recursive function list as the instrumental variable, use the measured soil salinity data to perform reverse fitting on the soil environment data, and obtain the overfitting relationship Σ -1 , where n is a positive integer;

[0015] S22: Use the segmented interpolation method or Gaussian mixture model to obtain the random distribution of the measured soil salinity data set {y}, and generate a random soil salinity data set {y} containing a preset number of data. 产生};

[0016] S23: According to the overfitting relationship Σ -1 and soil salinity random dataset {y 产生}Calculate a random soil environment dataset containing a preset number of data {x 产生}, the soil environment random data set {x 产生} and soil salinity random dataset {y 产生} as the training sample.

[0017] Furthermore, after step S4, the following steps are further included:

[0018] S5: If the validity of the dry soil salinity monitoring model is less than a preset threshold, n=n+1 is taken and step S21 is executed again.

[0019] In a second aspect, an embodiment of the present invention further provides a soil salinity monitoring model construction device, comprising:

[0020] A data acquisition module is configured to acquire a soil environment dataset {x} and a soil salinity measurement dataset {y} within a preset area, and use the soil environment dataset {x} and the soil salinity measurement dataset {y} as an initial dataset {x, y}, wherein the soil environment dataset {x} includes meteorological data, soil property data, groundwater level data, vegetation cover data, and remote sensing data;

[0021] A training sample acquisition module is used to process the initial data set {x, y} according to a preset method to obtain training samples;

[0022] The model building module is used to train the machine learning model using the training samples to obtain a soil salinity monitoring model.

[0023] Furthermore, the data acquisition module is further configured to take a preset number of data from the initial data set {x, y} as verification samples;

[0024] The device further comprises:

[0025] The model verification module is used to input the verification sample into the soil salinity monitoring model to verify the validity of the soil salinity monitoring model.

[0026] Furthermore, the training sample acquisition module is specifically configured to perform the following steps:

[0027] S61: Take the nth recursive function in the preset recursive function list as the instrumental variable, use the measured soil salinity data to perform reverse fitting on the soil environment data, and obtain the overfitting relationship Σ -1 , where n is a positive integer;

[0028] S62: Use the segmented interpolation method or Gaussian mixture model to obtain the random distribution of the measured soil salinity data set {y}, and generate a random soil salinity data set {y} containing a preset number of data. 产生};

[0029] S63: According to the overfitting relationship Σ -1 and soil salinity random dataset {y 产生}Calculate a random soil environment dataset containing a preset number of data {x 产生}, the soil environment random data set {x 产生} and soil salinity random dataset {y 产生} as the training sample.

[0030] In a third aspect, an embodiment of the present invention further provides a soil salinity monitoring model construction system, comprising a processor, an input device, an output device, and a memory, wherein the processor, input device, output device, and memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the method described in the first aspect.

[0031] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes the method described in the first aspect.

[0032] Implement the method provided in the embodiment of the present invention, and obtain a random soil salinity data set {y 产生} and soil environment random data set {x 产生 The researchers then used the training samples to perform neural network fitting and various machine learning methods to develop a soil salinity monitoring model. Because the training sample size was sufficiently large, the resulting soil salinity monitoring model was more effective, with more accurate predictions, effectively establishing a relationship between salinity and remote sensing spectra. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The accompanying drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present application, and together with the specification, are used to explain the principles of the present application. In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for describing the embodiments. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without inventive work.

[0034] Figure 1 Schematic diagram of a method for constructing a soil salinity monitoring model according to an embodiment of the present invention;

[0035] Figure 2 Schematic diagram of the structure of a soil salinity monitoring model construction device provided by an embodiment of the present invention;

[0036] Figure 3 Schematic diagram of the structure of the soil salinity monitoring model construction system provided by an embodiment of the present invention.

[0037] The purpose of this application, its features, and advantages will be further described in conjunction with the embodiments and with reference to the accompanying drawings. The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and the accompanying text are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of this application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0038] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0039] It should be noted that, in this document, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, components, features, and elements with the same name in different embodiments of the present application may have the same meaning or different meanings, and their specific meanings need to be determined by their explanation in the specific embodiment or further combined with the context of the specific embodiment.

[0040] It should be understood that although the terms "first," "second," "third," etc. may be used herein to describe various information, such information should not be limited to these terms. These terms are used solely to distinguish information of the same type from one another. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the term "if," as used herein, may be interpreted as "upon," "when," or "in response to a determination." Furthermore, as used herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context indicates otherwise. It should be further understood that the terms "comprising" and "including" indicate the presence of the recited features, steps, operations, elements, components, items, types, and / or groups, but do not preclude the presence, occurrence, or addition of one or more other features, steps, operations, elements, components, items, types, and / or groups. The terms "or," "and / or," "including at least one of the following," etc., as used herein, may be interpreted as inclusive, meaning any one or any combination. For example, “comprising at least one of the following: A, B, C” means “any of the following: A; B; C; A and B; A and C; B and C; A and B and C”; and for another example, “A, B or C” or “A, B and / or C” means “any of the following: A; B; C; A and B; A and C; B and C; A and B and C”. An exception to this definition will occur only when a combination of elements, functions, steps or operations are inherently mutually exclusive in some manner.

[0041] It should be understood that, although the various steps in the flowchart in the embodiment of the present application are shown in sequence according to the indication of the arrows, these steps are not necessarily performed in sequence in the order indicated by the arrows. Unless clearly stated herein, the execution of these steps is not strictly limited in order, and they can be performed in other orders. Moreover, at least a portion of the steps in the figure may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and their execution order is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0042] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0043] It should be noted that in this article, step codes such as S1 and S2 are used for the purpose of expressing the corresponding content more clearly and concisely, and do not constitute a substantial restriction on the order. When implementing the step, those skilled in the art may execute S2 first and then S1, etc., but these should all be within the scope of protection of this application.

[0044] like Figure 1 As shown, an embodiment of the present invention provides a method for constructing a soil salinity monitoring model, comprising:

[0045] S1: Obtain a soil environment dataset {x} and a soil salinity measurement dataset {y} within a preset area, and use the soil environment dataset {x} and the soil salinity measurement dataset {y} as the initial dataset {x, y}, wherein the soil environment dataset {x} includes meteorological data, soil property data, groundwater level data, vegetation cover data, and remote sensing data.

[0046] Specifically, the measured soil salinity dataset {y} includes soil salinity data at a depth of 0 to 30 cm, soil salinity data at a depth of 30 to 60 cm, and soil salinity data at a depth of 60 to 100 cm. The remote sensing data is satellite remote sensing spectral feature data.

[0047] Furthermore, in step S1, a preset number of data are taken from the initial data set {x, y} as verification samples, and the validity of the soil salinity monitoring model constructed in this embodiment is verified using the verification samples.

[0048] S2: Process the initial data set {x, y} according to a preset method to obtain training samples.

[0049] In actual sampling, the limited sample datasets available make it difficult to generate valid parameter estimates. In particular, the soil environment dataset {x} contains too many attributes, so typical fitting methods often overfit. Furthermore, when the sample dataset is smaller than 50, it is difficult to create sufficient validation and test sets. Therefore, instrumental variables are needed for data augmentation to generate a sufficiently large sample set (>1,000,000), which can then be used for various learning and data fitting tasks.

[0050] First, let’s introduce the instrumental variables:

[0051] Instrumental variables require the cov operation. cov stands for covariance, and its formula is:

[0052] cov(X,Y)=E((XE[X])(YE[Y]))=E[XY]-E[X]E[Y]

[0053] Intuitively, covariance represents the common expectation of the population errors of two variables. If the two variables have the same trend, then the covariance between them is positive; if the two variables are statistically independent, then the covariance between them is zero. However, the reverse is not true. That is, if the covariance of X and Y is zero, then they are not necessarily statistically independent. The unit of measurement for covariance is the covariance of X multiplied by the covariance of Y. Two random variables with a covariance of zero are said to be uncorrelated; otherwise, they are correlated.

[0054] During model estimation, a variable used as an instrument to replace the random explanatory variable associated with the error term in the model is called an instrumental variable λ. Assume the relationship to be solved is y = a x + b + ε, where the independent variable is x, the dependent variable is y, and ε is the random error term.

[0055] As an instrumental variable, λ must meet the following four conditions:

[0056] (1) Highly correlated with the independent variable x, i.e., cov(x,λ)>0;

[0057] (2) uncorrelated with the random error term ε, i.e., cov(ε,λ) = 0;

[0058] (3) With constant b, cov(b,λ)=0;

[0059] (4) When multiple instrumental variables need to be introduced into the same model, these instrumental variables are uncorrelated.

[0060] Among them, the first item limits the correlation between the instrumental variable λ and x, and the following three items limit the randomness of the instrumental variable λ.

[0061] So we get the coefficients of the above equation:

[0062] a=cov(y,λ) / cov(x,λ)

[0063] b=E(y)-aE(x)

[0064] For any x, y, according to the previous universal approximation theorem, we can use neural networks or machine learning to obtain these two relations Σ: and Σ -1 : And if the value of m or n is large enough, then there must be one of {x} or {y}. Since {x} or {y} is random, according to the conditions of the instrumental variable, check whether it meets the following conditions:

[0065] cov(y,x i >=α i cov(x i ,x i >≠0(correlation)

[0066] cov(x j ,x i >=0(i≠j)(randomness, see polynomial definition)

[0067] Choose λ = x i

[0068] Check that the vector {x i} (i is an integer, i≠0) are all random, related to {y}, and {x j}(j≠i) are not related, it can be used to find the coefficients of the polynomial {α i Otherwise, continue to select the next function in the recursive function list Δ below to calculate the new instrumental variable.

[0069] and

[0070] cov(x,y i >=β i cov(y i ,y i >≠0(correlation)

[0071] cov(y j ,y i >=0(i≠j)(randomness, see polynomial definition)

[0072] Choose λ = y i

[0073] Check and confirm that the vector {y i} (i is an integer, i≠0) are all random, related to {x}, and {y j}(j≠i) are not related, it can be used to find the coefficients of the polynomial {β i Otherwise, continue to select the next function in the recursive function list Δ below to calculate the new instrumental variable.

[0074] So we have:

[0075] cov(y,x i )=α i cov(x i ,x i ),then:

[0076] α i =cov(y,x i ) / cov(x i ,x i )

[0077] or

[0078] cov(x,y i )=βi cov(y i ,y i ),then:

[0079] β i =cov(x,y i ) / cov(y i ,y i )

[0080] Select a variable {x i} or {y i}, as an instrumental variable for a random explanatory variable {x} or {y} in the model, together with other variables in the model, constructs the corresponding parameter {α i}, and {β i This estimation method is called the instrumental variable method.

[0081] The key to the instrumental variable method is to select an effective instrumental variable. Due to the difficulty in selecting instrumental variables, the instrumental variable method itself has two shortcomings:

[0082] First, because the instrumental variable is not unique, the estimator of the instrumental variable is somewhat arbitrary;

[0083] Second, since the error term is actually unobservable, it is difficult to find a variable that is strictly unrelated to the error term but highly correlated with the random explanatory variable it replaces. To solve these problems, it is necessary to continuously adjust the recursive function until an ideal instrumental variable is calculated.

[0084] Generally speaking, the selection conditions of instrumental variables are:

[0085] 1. The instrumental variable must be highly correlated with x or y, that is, cov(x, λ) ≠ 0, or cov(y, λ) ≠ 0.

[0086] 2. The instrumental variable must be uncorrelated with the error ε, that is, cov(ε,λ)=0.

[0087] Polynomial expansion method for instrumental variables:

[0088] Without loss of generality, assume that the polynomial coefficients are {α i} and {β i}, so using forward and reverse overfitting, we get two relationships Σ: and Σ -1 : And the value of m or n is large enough, then {x i} or {y i}There must be one, for every x i or y i, according to the above instrumental variable selection condition Φ:

[0089] cov(y,x i >=α i cov(x i ,x i >≠0(i≠0)(correlation)

[0090] cov(x,x i >=0(i≠1)(randomness, see polynomial definition)

[0091] or

[0092] cov(x,y i >=β i cov(y i ,y i >≠0(i≠0)(correlation)

[0093] cov(y,y i >=0(i≠1)(randomness, see polynomial definition)

[0094] Or the instrumental variable selection condition Ψ:

[0095] 1. y obtained by sampling using the Y→X overfitting formula 采样 Calculate x. Check cov(y, x i ) is correlated with cov(x,x i >(i≠1) is not relevant.

[0096] 2. The x obtained by sampling using the X→Y overfitting formula 采样 Calculate y. cov(x, y i > must be correlated with cov(y,y i >(i≠1) is not relevant.

[0097] When the condition Φ or Ψ is not satisfied, a new instrumental variable is selected according to the order of the recursive function. A recursive function is a computational process in which each step requires the result of the previous step or steps. It is called recursive. Functions defined by recursive procedures are called recursive functions, such as continuous addition, continuous multiplication, and factorial. All recursive functions are computable, that is, they can be done. Recursive functions are used here because they are self-referential, such as

[0098] F3(x,y)=(y→F2(x,y)→y)

[0099] It represents the fitting of y to the previous result, and then the fitting of the previous result to y. According to information theory, this biased self-referentiality leads to ergodicity.

[0100] The recursive function list Δ is specifically:

[0101] 1. When the Y→X overfitting formula cannot meet the instrumental variable requirement Ψ, select one of the following recursive functions to calculate the instrumental variable:

[0102] F1(x,y)=y→(y→x)→y

[0103] F2(x,y)=(x→F1(x,y)→x)

[0104] F3(x,y)=(y→F2(x,y)→y)

[0105] F4(x,y)=(x→F3(x,y)→x)

[0106] F5(x,y)=(y→F4(x,y)→y)

[0107] F6(x,y)=(x→F5(x,y)→x) ...

[0109] 2. When the X→Y overfitting formula cannot meet the instrumental variable requirement Ψ, select one of the following recursive functions to calculate the instrumental variable:

[0110] F1(x,y)=x→(x→y)→x

[0111] F2(x,y)=(y→F1(x,y)→y)

[0112] F3(x,y)=(x→F2(x,y)→x)

[0113] F4(x,y)=(y→F3(x,y)→y)

[0114] F5(x,y)=(x→F4(x,y)→x)

[0115] F6(x,y)=(y→F5(x,y)→y) ...

[0117] 3. When {x} fails to meet the instrumental variable requirement Φ, one of the instrumental variables is selected according to this recursive function order:

[0118] F1(x,y)=y→(x)→y

[0119] F2(x,y)=(x→F1(x,y)→x)

[0120] F3(x,y)=(y→F2(x,y)→y)

[0121] F4(x,y)=(x→F3(x,y)→x)

[0122] F5(x,y)=(y→F4(x,y)→y)

[0123] F6(x,y)=(x→F5(x,y)→x) ...

[0125] 4. When {y} fails to meet the instrumental variable requirement Φ, select one of the instrumental variables according to this recursive function order:

[0126] F1(x,y)=x→(y)→x

[0127] F2(x,y)=(y→F1(x,y)→y)

[0128] F3(x,y)=(x→F2(x,y)→x)

[0129] F4(x,y)=(y→F3(x,y)→y)

[0130] F5(x,y)=(x→F4(x,y)→x)

[0131] F6(x,y)=(y→F5(x,y)→y) ...

[0133] 5. When the Y→X overfitting formula cannot meet the instrumental variable requirement Ψ, select one of the following recursive functions to calculate the instrumental variable:

[0134] F1(x,y)=(y→x)→y

[0135] F2(x,y)=(F1(x,y)→x)

[0136] F3(x,y)=(F2(x,y)→y)

[0137] F4(x,y)=(F3(x,y)→x)

[0138] F5(x,y)=(F4(x,y)→y)

[0139] F6(x,y)=(F5(x,y)→x) ...

[0141] 6. When the X→Y overfitting formula cannot meet the instrumental variable requirement Ψ, select one of the following recursive functions to calculate the instrumental variable:

[0142] F1(x,y)=(x→y)→x

[0143] F2(x,y)=(F1(x,y)→y)

[0144] F3(x,y)=(F2(x,y)→x)

[0145] F4(x,y)=(F3(x,y)→y)

[0146] F5(x,y)=(F4(x,y)→x)

[0147] F6(x,y)=(F5(x,y)→y) ...

[0149] 7. When {x} fails to meet the instrumental variable requirement Φ, choose one of the instrumental variables in this recursive function order:

[0150] F1(x,y)=(x)→y

[0151] F2(x,y)=(F1(x,y)→x)

[0152] F3(x,y)=(F2(x,y)→y)

[0153] F4(x,y)=(F3(x,y)→x)

[0154] F5(x,y)=(F4(x,y)→y)

[0155] F6(x,y)=(F5(x,y)→x) ...

[0157] 8. When {y} fails to meet the instrumental variable requirement Φ, choose one of the instrumental variables in this recursive function order:

[0158] F1(x,y)=(y)→x

[0159] F2(x,y)=(F1(x,y)→y)

[0160] F3(x,y)=(F2(x,y)→x)

[0161] F4(x,y)=(F3(x,y)→y)

[0162] F5(x,y)=(F4(x,y)→x)

[0163] F6(x,y)=(F5(x,y)→y) ...

[0165] The recursive function means: use overfitting or a large number of samples as the intermediate value, and continue to alternately fit x or y to the intermediate value, or the intermediate value to x or y.

[0166] For example, the function (y→x) means to overfit from y to x and choose x as the middle value.

[0167] For the vector {x i} and {y i}(i≠0), check whether the instrumental variable selection condition Φ or Ψ is met. If not, select the next one in the recursive function list Δ above, which can be used to find each coefficient α of the polynomial i and β i Since the recursive function list Δ is self-referential, it brings about various states of experience, and we can always find a sequence of instrumental variables to complete the calculation.

[0168] Using the neural network fitting method or various machine learning fitting methods can produce a similar effect as the above polynomial coefficient, which is equivalent to finding the polynomial coefficient {α i} and {β i}.

[0169] So neural network overfitting (X→Y) is equivalent to:

[0170] cov(y,x i )=α i cov(x i ,x i ), so α i =cov(y,x i ) / cov(x i ,x i )

[0171] Neural network overfitting (Y→X) is equivalent to:

[0172] cov(x,y i )=β i cov(x i ,x i ), so β i =cov(x,y i ) / cov(x i ,x i )

[0173] In summary, specifically, the data processing of the initial data set {x, y} according to a preset method to obtain training samples includes:

[0174] S21: Take the nth recursive function in the preset recursive function list as the instrumental variable, use the measured soil salinity data to perform reverse fitting on the soil environment data, and obtain the overfitting relationship Σ -1 , where n is a positive integer;

[0175] S22: Use the segmented interpolation method or Gaussian mixture model to obtain the random distribution of the measured soil salinity data set {y}, and generate a random soil salinity data set {y} containing a preset number of data. 产生};

[0176] S23: According to the overfitting relationship Σ -1 and soil salinity random dataset {y 产生}Calculate a random soil environment dataset containing a preset number of data {x 产生}, the soil environment random data set {x 产生} and soil salinity random dataset {y 产生} as the training sample.

[0177] S3: Use the training samples to train the machine learning model to obtain a soil salinity monitoring model.

[0178] In this embodiment, after step S3, the following steps are further included:

[0179] S4: Inputting the verification sample into the soil salinity monitoring model to verify the validity of the soil salinity monitoring model.

[0180] S5: If the validity of the dry soil salinity monitoring model is less than a preset threshold, n=n+1 is taken and step S21 is executed again.

[0181] The same inventive concept can be used, such as Figure 2 As shown, an embodiment of the present invention also provides a soil salinity monitoring model construction device. The device may include:

[0182] Data acquisition module 201 is used to obtain a soil environment dataset {x} and a soil salinity measured dataset {y} within a preset area, and use the soil environment dataset {x} and the soil salinity measured dataset {y} as an initial dataset {x, y}, wherein the soil environment dataset {x} includes meteorological data, soil property data, groundwater level data, vegetation cover data, and remote sensing data;

[0183] The training sample acquisition module 202 is used to process the initial data set {x, y} according to a preset method to obtain training samples;

[0184] The model building module 203 is used to train the machine learning model using the training samples to obtain a soil salinity monitoring model.

[0185] Furthermore, the data acquisition module 201 is further configured to obtain a preset amount of data from the initial data set {x, y} as verification samples.

[0186] The device further comprises:

[0187] The model verification module 204 is configured to input the verification sample into the soil salinity monitoring model to verify the validity of the soil salinity monitoring model.

[0188] Specifically, the training sample acquisition module 202 is specifically configured to perform the following steps:

[0189] S61: Take the nth recursive function in the preset recursive function list as the instrumental variable, use the measured soil salinity data to perform reverse fitting on the soil environment data, and obtain the overfitting relationship Σ -1 , where n is a positive integer;

[0190] S62: Use the segmented interpolation method or Gaussian mixture model to obtain the random distribution of the measured soil salinity data set {y}, and generate a random soil salinity data set {y} containing a preset number of data. 产生};

[0191] S63: According to the overfitting relationship Σ -1 and soil salinity random dataset {y 产生}Calculate a random soil environment dataset containing a preset number of data {x 产生}, the soil environment random data set {x 产生} and soil salinity random dataset {y 产生} as the training sample.

[0192] Based on the same inventive concept, an embodiment of the present invention also provides a soil salinity monitoring model construction system. Figure 3 As shown, the system may include: one or more processors 101, one or more input devices 102, one or more output devices 103, and a memory 104. The processors 101, input devices 102, output devices 103, and memory 104 are interconnected via a bus 105. The memory 104 is used to store a computer program, which includes program instructions. The processor 101 is configured to call the program instructions to execute the method of the embodiment of the soil salinity monitoring model construction method.

[0193] It should be understood that in the embodiment of the present invention, the processor 101 may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0194] The input device 102 may include a keyboard, etc., and the output device 103 may include a display (LCD, etc.), a speaker, etc.

[0195] The memory 104 may include a read-only memory and a random access memory, and provides instructions and data to the processor 101. A portion of the memory 104 may also include a non-volatile random access memory. For example, the memory 104 may also store device type information.

[0196] In a specific implementation, the processor 101, input device 102, and output device 103 described in the embodiment of the present invention can execute the implementation method described in the embodiment of the soil salinity monitoring model construction method provided by the embodiment of the present invention, which will not be repeated here.

[0197] It should be noted that, for the specific workflow of the soil salinity monitoring model construction system, please refer to the aforementioned method embodiment part, which will not be repeated here.

[0198] Furthermore, an embodiment of the present invention also provides a readable storage medium storing a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the above-mentioned soil salinity monitoring model construction method is implemented.

[0199] The computer-readable storage medium may be the internal storage unit of the background server described in the aforementioned embodiment, such as the hard disk or memory of the system. The computer-readable storage medium may also be an external storage device of the system, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the system. Furthermore, the computer-readable storage medium may also include both the internal storage unit of the system and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the system. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0200] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the composition and steps of each example according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0201] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0202] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0203] The serial numbers of the above embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0204] In this application, the same or similar terminology, technical solutions and / or application scenario descriptions are generally only described in detail the first time they appear. When they appear again later, they are generally not repeated for the sake of brevity. When understanding the technical solutions and other contents of this application, for the same or similar terminology, technical solutions and / or application scenario descriptions that are not described in detail later, you can refer to the previous relevant detailed descriptions.

[0205] In this application, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0206] The various technical features of the technical solution of this application can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0207] The above description is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the scope of protection of the present application.

Claims

1. A method for constructing a soil salinity monitoring model, characterized in that: include: S1: Obtain a soil environment dataset {x} and a soil salinity measurement dataset {y} within a preset area, and use the soil environment dataset {x} and the soil salinity measurement dataset {y} as an initial dataset {x, y}, wherein the soil environment dataset {x} includes meteorological data, soil property data, groundwater level data, vegetation cover data, and remote sensing data; S2: Process the initial data set {x, y} according to a preset method to obtain training samples; S3: Using the training samples to train a machine learning model to obtain a soil salinity monitoring model; In step S1, a preset number of data are taken from the initial data set {x, y} as verification samples; And after step S3, the method further includes: S4: inputting the verification sample into the soil salinity monitoring model to verify the validity of the soil salinity monitoring model; The step of processing the initial data set {x, y} according to a preset method to obtain training samples includes: S21: Take the nth recursive function in the preset recursive function list as the instrumental variable, use the measured soil salinity data to perform reverse fitting on the soil environment data, and obtain the overfitting relationship Σ -1 , where n is a positive integer; S22: Use the segmented interpolation method or Gaussian mixture model to obtain the random distribution of the measured soil salinity data set {y}, and generate a random soil salinity data set {y} containing a preset number of data. 产生 }; S23: According to the overfitting relationship Σ -1 and soil salinity random dataset {y 产生 }Calculate a random soil environment dataset containing a preset number of data {x 产生 }, the soil environment random data set {x 产生 } and soil salinity random dataset {y 产生 } as the training sample.

2. A soil salinity monitoring model construction method according to claim 1, characterized in that: After step S4, the following steps are also included: S5: If the effectiveness of the soil salinity monitoring model is less than a preset threshold, n=n+1 is taken and step S21 is executed again.

3. A soil salinity monitoring model construction device, characterized in that: include: A data acquisition module is configured to acquire a soil environment dataset {x} and a soil salinity measurement dataset {y} within a preset area, and use the soil environment dataset {x} and the soil salinity measurement dataset {y} as an initial dataset {x, y}, wherein the soil environment dataset {x} includes meteorological data, soil property data, groundwater level data, vegetation cover data, and remote sensing data; A training sample acquisition module is used to process the initial data set {x, y} according to a preset method to obtain training samples; A model building module is used to train a machine learning model using the training samples to obtain a soil salinity monitoring model; The data acquisition module is further configured to take a preset number of data from the initial data set {x, y} as verification samples; The device further comprises: A model verification module, configured to input the verification sample into the soil salinity monitoring model to verify the validity of the soil salinity monitoring model; The training sample acquisition module is specifically configured to perform the following steps: S61: Take the nth recursive function in the preset recursive function list as the instrumental variable, use the measured soil salinity data to perform reverse fitting on the soil environment data, and obtain the overfitting relationship Σ -1 , where n is a positive integer; S62: Use the segmented interpolation method or Gaussian mixture model to obtain the random distribution of the measured soil salinity data set {y}, and generate a random soil salinity data set {y} containing a preset number of data. 产生 }; S63: According to the overfitting relationship Σ -1 and soil salinity random dataset {y 产生 }Calculate a random soil environment dataset containing a preset number of data {x 产生 }, the soil environment random data set {x 产生 } and soil salinity random dataset {y 产生 } as the training sample.

4. A soil salinity monitoring model construction system, characterized in that: The method comprises a processor, an input device, an output device and a memory, wherein the processor, the input device, the output device and the memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the method according to any one of claims 1 to 2.

5. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions. When the program instructions are executed by a processor, the processor is caused to perform the method according to any one of claims 1 to 2.

Images