A method for rapid estimation of lithium content in clay

Abstract

The present application relates to the field of hyperspectral remote sensing estimation of soil lithium content, and particularly relates to a method for rapidly estimating lithium content in clay. The method comprises the following steps: obtaining spectral reflectance data and corresponding lithium element content of a clay sample; obtaining a plurality of spectral resolution spectral data according to the spectral reflectance data, combining the spectral data and the lithium element content to obtain an optimal spectral resolution; obtaining a plurality of fractional derivative spectra based on the optimal spectral resolution, obtaining a characteristic spectral data set through the correlation between the fractional derivative spectra and the lithium element content; constructing a rapid lithium content estimation model, training and verifying the rapid lithium content estimation model using the characteristic spectral data set, and completing the rapid estimation of the lithium content in the clay through the trained rapid lithium content estimation model. The present application obtains a characteristic spectral data set under a plurality of fractional derivative spectra at an optimal resolution, and further uses a rapid estimation model to solve the problem of rapidly, efficiently and accurately estimating the lithium content in clay.

Description

Technical Field

[0001] This invention relates to the field of hyperspectral remote sensing for estimating soil lithium content, and specifically to a rapid method for estimating lithium content in clay. Background Technology

[0002] Lithium is an "emerging" environmental pollutant, and its presence in soil, in particular, has garnered widespread attention due to its potential to harm human health through the food chain. Traditional survey methods primarily rely on field sampling and large-scale indoor chemical analysis. While these methods offer high precision, they are costly, inefficient, and for large-scale monitoring, they are mainly based on spatial interpolation, making them highly dependent on the accuracy and spatial uniformity of the data.

[0003] Although existing studies have used hyperspectral remote sensing to estimate the heavy metal content in soil, these studies mainly utilize raw reflectance spectral data or perform various transformations such as reciprocal, logarithmic, and envelope removal to construct models. However, there has been no systematic analysis of the impact of spectral resolution on the accuracy of the estimation model. Furthermore, the spectral transformations are all performed on integer order, without analyzing the impact of non-zero integer derivatives on the accuracy of the estimation model. In terms of feature band selection, feature band extraction is only performed on a single type of spectrum, without integrating multiple forms of spectra for feature extraction. In particular, there are very few studies on estimating lithium content in clay.

[0004] Therefore, there is an urgent need for a method that is highly applicable and can quickly, efficiently, and accurately estimate the lithium content in clay. Summary of the Invention

[0005] To address the shortcomings of existing methods and the needs of practical applications, and in order to comprehensively extract features from multiple forms of spectra, improve the accuracy of estimation models for non-zero integer derivative pairs, and solve the problem of rapidly, efficiently, and accurately estimating the lithium content in clay, this invention provides a rapid estimation method for lithium content in clay, comprising the following steps: acquiring spectral reflectance data and corresponding lithium content of clay samples; obtaining spectral data at multiple spectral resolutions based on the spectral reflectance data, and obtaining the optimal spectral resolution by combining the spectral data and the lithium content; obtaining multiple sets of fractional derivative spectra based on the optimal spectral resolution, and obtaining a feature spectral dataset by analyzing the correlation between the fractional derivative spectra and the lithium content; constructing a rapid lithium content estimation model, training and validating the rapid lithium content estimation model using the feature spectral dataset, and using the trained rapid lithium content estimation model to quickly estimate the lithium content in clay. This invention determines the optimal spectral resolution of spectral data through window sampling, then divides the data into feature spectral datasets under multiple fractional derivative spectra, extracts the feature spectral features, and then uses the rapid lithium content estimation model to quickly estimate the lithium content in clay, thereby improving the applicability of the invention and solving the problem of rapidly, efficiently, and accurately estimating the lithium content in clay.

[0006] Optionally, obtaining spectral data with multiple spectral resolutions based on the spectral reflectance data includes the following steps:

[0007] The size of the sliding window is set; based on the sliding window, the spectral reflectance data is divided to obtain spectral data at multiple spectral resolutions. By setting different sliding window sizes, spectral reflectance data at different spectral resolutions can be obtained, which helps to improve the estimation accuracy of the present invention.

[0008] Optionally, obtaining the optimal spectral resolution by combining the spectral data and the lithium content includes the following steps:

[0009] Based on the spectral data and the lithium content, a spectral data training set and a spectral data validation set are established. A machine classification model is constructed, trained using the spectral data training set, and the validation results are obtained by combining the spectral data validation set and the trained machine classification model. The validation results are then comprehensively analyzed using the coefficient of determination, relative root mean square error, and relative analysis error to obtain the optimal spectral resolution. This invention utilizes multiple error analysis methods, which facilitates the rapid and accurate acquisition of the optimal spectral resolution.

[0010] Optionally, the multiple sets of fractional derivative spectra obtained based on the optimal spectral resolution include the following steps:

[0011] A fractional derivative spectrum of orders 0 to 2 is obtained based on the fractional derivative interval. This invention further subdivides the integer derivative spectrum, reducing the impact of non-zero integer derivatives on the accuracy of the estimation model.

[0012] Optionally, the fractional derivative spectrum of orders 0 to 2 is obtained according to the fractional interval, satisfying the following formula:

[0013]

[0014] in, Represents the function The a-th order fractional derivative, This indicates the order of the fraction. Indicates the band as Spectral reflectance at that location , Indicates the band position. Represents the Gamma function. This represents the spectral reflectance at wavelength t. This invention obtains multiple sets of fractional derivative spectra through model formulas, which further facilitates feature extraction and thus improves the accuracy of the invention.

[0015] Optionally, obtaining the characteristic spectral dataset through the correlation between the fractional derivative spectrum and the lithium content includes the following steps:

[0016] The correlation between the fractional derivative spectrum and the lithium content is calculated using a first correlation model. Based on this correlation, the fractional derivative spectra are filtered to obtain a significantly correlated spectral dataset. Significant feature spectral bands are extracted from this dataset to construct a feature spectral dataset. This invention uses a first correlation model to filter useful spectral datasets and extract effective feature spectra, further improving the estimation speed of this invention.

[0017] Logically, the first correlation model satisfies the following formula:

[0018]

[0019] in, Indicates the first correlation. Indicates the number of samples. express The reflectance of a sample, representing Indicates the first The lithium content value of each sample, This represents the band mean of all samples. This represents the average lithium content across all samples. This invention utilizes a model to quantify the first correlation, ensuring objectivity and accuracy, and further guaranteeing data accuracy.

[0020] Optionally, the fractional derivative spectra are filtered based on the correlation to obtain a significantly correlated spectral dataset, satisfying the following formula:

[0021]

[0022] in, This indicates the first correlation. The present invention filters out portions with a first correlation of less than 0.6, reducing the computational data and improving the efficiency of the invention.

[0023] Optionally, extracting salient feature spectral bands from the significantly correlated spectral dataset includes the following steps:

[0024] Band features are extracted from the significantly correlated spectral dataset; a band feature importance evaluation model is constructed, and the importance of the band features is obtained using the band feature importance evaluation model; based on the importance, salient feature spectral bands are obtained. This invention obtains salient feature spectral bands through an evaluation model, integrating feature extraction from multiple forms of spectroscopy, further improving the accuracy of the invention.

[0025] Optionally, the band feature importance assessment model satisfies the following formula:

[0026]

[0027] in, Indicates the first Band characteristics in Importance of the dataset This indicates that sample data was not used to train a specific decision branch. Indicates the first Branches The number of data samples Indicates the number of branches. Indicates an indicator function, Indicates the first The true results of a sample Indicates the first The first sample The prediction results for each branch, Represents the i-th random permutation Branch pairs The data in the first The prediction results for each sample. Attached Figure Description

[0028] Figure 1 A flowchart illustrating a rapid method for estimating lithium content in clay, provided in an embodiment of the present invention;

[0029] Figure 2 A framework diagram of a rapid estimation system for lithium content in clay provided in an embodiment of the present invention;

[0030] Figure 3 This is a schematic diagram of a device for rapidly estimating the lithium content in clay, provided as an embodiment of the present invention. Detailed Implementation

[0031] Specific embodiments of the present invention will now be described in detail. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the invention.

[0032] Throughout this specification, references to "an embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with that embodiment or example is included in at least one embodiment of the invention. Therefore, the phrases "in an embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily refer to the same embodiment or example. Furthermore, specific features, structures, or characteristics can be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. Moreover, those skilled in the art will understand that the illustrations provided herein are for illustrative purposes and are not necessarily drawn to scale.

[0033] Please see Figure 1 To comprehensively extract features from various spectral forms, improve the accuracy of estimation models using non-zero integer derivative pairs, and solve the problem of rapidly, efficiently, and accurately estimating lithium content in clay, this invention provides a rapid method for estimating lithium content in clay, such as... Figure 1 As shown, in one embodiment, the method includes the following steps:

[0034] S1. Obtain the spectral reflectance data and corresponding lithium content of the clay sample.

[0035] In this embodiment, soil samples collected in the field from the study area were tested in the laboratory to obtain spectral reflectance data and corresponding lithium content of the samples.

[0036] Specifically, the spectral reflectance data of the sample was obtained by measuring it at a vertical distance of 20 cm using a FieldSpec4 portable spectrometer; the lithium content of the sample was obtained by detecting it using an iCAP7400 inductively coupled plasma atomic emission spectrometer at a temperature of 21 degrees Celsius and a relative humidity of 40%.

[0037] S2. Based on the spectral reflectance data, obtain spectral data with multiple spectral resolutions, and combine the spectral data with the lithium content to obtain the optimal spectral resolution.

[0038] Specifically, step S2, which involves obtaining spectral data with multiple spectral resolutions based on the spectral reflectance data, includes the following steps:

[0039] S21. Set the size of the sliding window.

[0040] The spectral resolution of the spectral data obtained in step S1 is 1 nm. In order to obtain spectral data with different spectral resolutions, a window sliding method is used for spectral resampling.

[0041] S22. Based on the sliding window, divide the spectral reflectance data to obtain spectral data with multiple spectral resolutions.

[0042] Depending on the size of the sliding window, spectral data with different sliding window resolutions are obtained segment by segment. In the embodiments, the sliding window sizes are 5, 10, 20, 25, and 40, thereby obtaining spectral data with spectral resolutions of 5nm, 10nm, 20nm, 25nm, and 40nm.

[0043] Specifically, spectral data with a spectral resolution of 5 nm is obtained by using spectral data with a spectral resolution of 1 nm and a smoothing method with a window size of 5. That is, starting from the first band and ending at the fifth band, the average spectral value of these five bands is calculated as the first band with a spectral resolution of 5 nm. Then the window is moved forward by five bands, that is, the average spectral value of the sixth to tenth bands is calculated as the second band with a spectral resolution of 5 nm. This process is repeated for the remaining bands, and finally the sample spectral data with a spectral resolution of 5 nm is obtained.

[0044] Further, step S2, which combines the spectral data and the lithium content to obtain the optimal spectral resolution, includes the following steps:

[0045] Based on the spectral data and the lithium content, establish a spectral data training set and a spectral data validation set;

[0046] A machine classification model is constructed, trained using the spectral data training set, and the validation results are obtained by combining the spectral data validation set and the trained machine classification model.

[0047] The optimal spectral resolution is obtained by comprehensively analyzing the verification results using the coefficient of determination, relative root mean square error, and relative analysis error.

[0048] Specifically, based on spectral data with spectral resolutions of 1nm, 5nm, 10nm, 20nm, 25nm, and 40nm, the dataset is divided into training and testing sets according to a certain ratio to construct a machine classification model. The machine classification model includes support vector machines and neural networks, or it can be constructed based on partial least squares. Based on the performance on the validation set, the spectral resolution with the best model performance is determined by combining the coefficient of determination, relative root mean square error, and relative analysis error.

[0049] Furthermore, the coefficient of determination satisfies the following formula:

[0050]

[0051] The relative root mean square error satisfies the following formula:

[0052]

[0053] The relative analysis error satisfies the following formula:

[0054]

[0055] in, , and These are samples The actual value, predicted value, and average value.

[0056] S3. Based on the optimal spectral resolution, obtain multiple sets of fractional derivative spectra, and obtain a characteristic spectral dataset by correlating the fractional derivative spectra with the lithium content.

[0057] In this embodiment, S3, obtaining multiple sets of fractional derivative spectra based on the optimal spectral resolution, includes the following steps:

[0058] S31. Set the step interval.

[0059] Specifically, the size of the fractional interval affects the number of fractional derivative spectra. In this embodiment, the fractional interval is 0.05.

[0060] S32. Based on the said fractional interval, obtain the fractional derivative spectra of orders 0 to 2.

[0061] Forty sets of fractional derivative spectra from 0 to 2nd order, spaced at 0.05, were calculated at the optimal spectral resolution. In other embodiments, more fractional derivative spectra can be obtained depending on actual computing power and other factors.

[0062] Specifically, the fractional derivative spectrum of orders 0 to 2 is obtained based on the fractional interval, satisfying the following formula:

[0063]

[0064] in, Represents the function The a-th order fractional derivative, This indicates the order of the fraction. Indicates the band as Spectral reflectance at that location , Indicates the band position. Represents the Gamma function. This represents the spectral reflectance at wavelength t.

[0065] Furthermore, by correlating the fractional derivative spectrum with the lithium content, a characteristic spectral dataset is obtained, including the following steps:

[0066] S33. Calculate the correlation between the fractional derivative spectrum and the lithium content using the first correlation model.

[0067] Specifically, the first correlation model satisfies the following formula:

[0068]

[0069] in, Indicates the first correlation. Indicates the number of samples. express The reflectance of a sample, representing Indicates the first The lithium content value of each sample, This represents the band mean of all samples. This represents the mean lithium content of all samples. The correlation between all fractional derivative spectra and lithium content was calculated using the first correlation model.

[0070] S34. Based on the correlation, filter the fractional derivative spectra to obtain a significantly correlated spectral dataset.

[0071] Specifically, the fractional derivative spectra are filtered based on the correlation to obtain a significantly correlated spectral dataset, satisfying the following formula:

[0072]

[0073] in, The first correlation is indicated. All spectral data with a correlation greater than 0.6 between each fractional derivative spectrum and lithium content are extracted and combined to obtain a significantly correlated spectral dataset.

[0074] S35. Extract significant feature spectral bands from the significantly correlated spectral dataset, and then construct a feature spectral dataset.

[0075] In this embodiment, extracting salient spectral bands from the significantly correlated spectral dataset includes the following steps:

[0076] S351. Extract band features from the significantly correlated spectral dataset.

[0077] Band features are extracted from the significantly correlated spectral dataset using machine learning feature extraction models, including gradient boosting trees, random forests, and decision stumps.

[0078] S352. Construct a band feature importance assessment model, and use the band feature importance assessment model to obtain the importance of the band features.

[0079] In this embodiment, band features are extracted using machine learning models such as random forests, and the band feature importance evaluation model satisfies the following formula:

[0080]

[0081] in, Indicates the first Band characteristics in Importance of the dataset This indicates that sample data was not used to train a specific decision branch. Indicates the first Branches The number of data samples Indicates the number of branches. Indicates an indicator function, Indicates the first The true results of a sample Indicates the first The first sample The prediction results for each branch, Represents the i-th random permutation Branch pairs The data in the first The prediction results for each sample.

[0082] S353. Based on the aforementioned importance, obtain the significant characteristic spectral bands.

[0083] A random forest model is trained using all band features. The importance of each band feature is then calculated. The band features are then sorted according to their importance. The top k band features with the highest importance are selected. The model performance of the selected band features is evaluated using cross-validation, taking into account the coefficient of determination, relative root mean square error, and relative analysis error. Significant feature spectral bands are obtained, and then a feature spectral dataset is constructed by combining the corresponding lithium content.

[0084] S4. Construct a rapid lithium content estimation model, train and validate the rapid lithium content estimation model using the characteristic spectral dataset, and complete the rapid estimation of lithium content in clay using the trained rapid lithium content estimation model.

[0085] Based on the feature spectral dataset from step S3, the training set and validation set are divided according to a certain ratio. Then, a rapid lithium content estimation model is constructed using partial least squares or other machine learning algorithms. The rapid lithium content estimation model is trained and validated using the feature spectral dataset. The rapid lithium content estimation in the clay is then completed by combining the spectral data of the clay area to be tested with the trained rapid lithium content estimation model.

[0086] Please see Figure 2In an embodiment, to efficiently execute the rapid estimation method for lithium content in clay provided by this invention, this invention also provides a rapid estimation system for lithium content in clay, comprising: an input device, an output device, a processor, and a memory, wherein the input device, output device, processor, and memory are interconnected, and the memory contains program instructions for the steps of the rapid estimation method for lithium content in clay. The rapid estimation system for lithium content in clay provided by this invention has a compact structure and stable performance, and can stably execute the rapid estimation method for lithium content in clay provided by this invention, further improving the overall applicability and practical application capability of this invention.

[0087] In embodiments, the processor may be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), 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. Input devices can be used to acquire data information. Output devices can be used to output the results obtained by storing program instructions contained in a computer program in the memory provided by this invention. The memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory may also include non-volatile random access memory.

[0088] In yet another alternative embodiment, please refer to Figure 3 To efficiently implement the rapid estimation method for lithium content in clay provided by this invention, this embodiment also provides a rapid estimation device for lithium content in clay, such as... Figure 3 As shown, it includes:

[0089] The memory 10 stores the computer program; the processor 20 executes the computer program to implement the aforementioned method for rapidly estimating the lithium content in clay. The system includes the memory 10, processor 20, communication interface 31, and communication bus 32. The memory 10, processor 20, and communication interface 31 all communicate with each other via the communication bus 32.

[0090] In this embodiment, the memory 10 is used to store one or more program instructions. The memory 10 may store program instructions for implementing the following functions:

[0091] Acquire spectral reflectance data and corresponding lithium content of clay samples; obtain spectral data at multiple spectral resolutions based on the spectral reflectance data, and determine the optimal spectral resolution by combining the spectral data and lithium content; obtain multiple sets of fractional derivative spectra based on the optimal spectral resolution, and obtain a characteristic spectral dataset by analyzing the correlation between fractional derivative spectra and lithium content; construct a rapid lithium content estimation model, train and validate the rapid lithium content estimation model using the characteristic spectral dataset, and complete the rapid estimation of lithium content in clay using the trained rapid lithium content estimation model.

[0092] In one possible implementation, memory 10 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created during use. Furthermore, memory 10 may include read-only memory and random access memory, providing instructions and data to the processor. A portion of the memory may also include NVRAM. The memory stores the operating system and operating instructions, executable modules, or data structures, or subsets thereof, or extended sets thereof. The operating instructions may include various operation instructions for implementing various operations. The operating system may include various system programs for implementing various basic tasks and handling hardware-based tasks.

[0093] Processor 20 can be a Central Processing Unit (CPU), an application-specific integrated circuit, a digital signal processor, a field-programmable gate array, or other programmable logic device. Processor 20 can be a microprocessor or any conventional processor. Processor 20 can call programs stored in memory 10. Communication interface 31 can be an interface for a communication module, used to connect with other devices or systems.

[0094] Of course, it should be noted that, Figure 3 The structure shown does not constitute a limitation on the rapid estimation device for lithium content in clay in this embodiment. In practical applications, rapid estimation devices for lithium content in clay may include those that... Figure 3 More or fewer components as shown, or combinations of certain components.

[0095] The embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method for rapidly estimating the lithium content in clay.

[0096] The storage medium can include various media that can store program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] In summary, this invention determines the optimal spectral resolution of spectral data through window sampling, then divides the data into multiple characteristic spectral datasets under fractional derivative spectra, extracts the characteristic spectral features, and then uses a rapid lithium content estimation model to quickly estimate the lithium content in clay. This improves the applicability of the invention and solves the problem of rapidly, efficiently, and accurately estimating the lithium content in clay.

[0098] Therefore, this invention effectively overcomes the various shortcomings of the prior art and has high industrial application value.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the present invention.

Claims

1. A rapid method for estimating lithium content in clay, characterized in that, The rapid method for estimating the lithium content in clay includes the following steps: Obtain the spectral reflectance data and corresponding lithium content of clay samples; Based on the spectral reflectance data, spectral data with multiple spectral resolutions are obtained. Combining the spectral data with the lithium content, the optimal spectral resolution is obtained. Based on the optimal spectral resolution, multiple sets of fractional derivative spectra are obtained. By analyzing the correlation between the fractional derivative spectra and the lithium content, a characteristic spectral dataset is obtained. A rapid lithium content estimation model is constructed, and the model is trained and validated using the characteristic spectral dataset. The trained rapid lithium content estimation model is then used to quickly estimate the lithium content in clay. Obtaining multiple sets of fractional derivative spectra based on the optimal spectral resolution includes the following steps: Set the interval between steps; Based on the aforementioned order interval, fractional derivative spectra of orders 0 to 2 are obtained; The fractional derivative spectra of orders 0 to 2 are obtained based on the fractional intervals, satisfying the following formula: in, Represents the function The a-th order fractional derivative, This indicates the order of the fraction. Indicates the band as Spectral reflectance at that location , Indicates the band position. Represents the Gamma function. This represents the spectral reflectance at wavelength t. The process of obtaining a characteristic spectral dataset by correlating the fractional derivative spectrum with the lithium content includes the following steps: The correlation between the fractional derivative spectrum and the lithium content is calculated using the first correlation model. Based on the correlation, the fractional derivative spectra are filtered to obtain a significantly correlated spectral dataset; Significant feature spectral bands are extracted from the significantly correlated spectral dataset, and then a feature spectral dataset is constructed. The extraction of significant feature spectral bands from the significantly correlated spectral dataset includes the following steps: Extract band features from the significantly correlated spectral dataset; Construct a band feature importance assessment model, and use the band feature importance assessment model to obtain the importance of the band features; Based on the aforementioned importance, significant characteristic spectral bands are obtained.

2. The rapid estimation method for lithium content in clay according to claim 1, characterized in that, The step of obtaining spectral data with multiple spectral resolutions based on the spectral reflectance data includes the following steps: Set the size of the sliding window; Based on the sliding window, the spectral reflectance data is divided to obtain spectral data with multiple spectral resolutions.

3. The rapid estimation method for lithium content in clay according to claim 1, characterized in that, The process of combining the spectral data and the lithium content to obtain the optimal spectral resolution includes the following steps: Based on the spectral data and the lithium content, establish a spectral data training set and a spectral data validation set; A machine classification model is constructed, trained using the spectral data training set, and the validation results are obtained by combining the spectral data validation set and the trained machine classification model. The optimal spectral resolution is obtained by comprehensively analyzing the verification results using the coefficient of determination, relative root mean square error, and relative analysis error.

4. The rapid estimation method for lithium content in clay according to claim 1, characterized in that, The first correlation model satisfies the following formula: in, Indicates the first correlation. Indicates the number of samples. express The reflectance of each sample Indicates the first The lithium content value of each sample, This represents the band mean of all samples. This represents the average lithium content across all samples.

5. The rapid estimation method for lithium content in clay according to claim 1, characterized in that, The fractional derivative spectra are filtered based on the correlation to obtain a significantly correlated spectral dataset that satisfies the following formula: in, This indicates the first correlation.

6. The rapid estimation method for lithium content in clay according to claim 1, characterized in that, The band feature importance assessment model satisfies the following formula: in, Indicates the first Band characteristics in Importance of the dataset This indicates that sample data was not used to train a specific decision branch. Indicates the first Branches The number of data samples Indicates the number of branches. Indicates an indicator function, Indicates the first The true results of a sample Indicates the first The first sample The prediction results for each branch, Indicates the i-th random permutation Branch pairs The data in the first The prediction results for each sample.

Images