Soil type mapping progressive modeling method and system based on sample enhancement

Abstract

The invention belongs to the technical field of soil modeling, and provides a soil type mapping progressive modeling method and system based on sample enhancement, and the method is characterized in that the method comprises the steps: processing target soil data through a soil type mapping model, obtaining a modeling result of the target soil data; performing training through the first soil training data to generate a first classification model, and performing classification and enhancement through the first classification model to obtain second soil training data; the training subsets are enhanced through land utilization type label division and a virtual sample generation method; respectively training through the enhanced training subsets to obtain a prediction result of each enhanced training subset; and splicing the prediction results to obtain a final prediction result, and further obtaining a soil type mapping model. The method has the beneficial effects that the soil modeling precision is improved, and the soil sample acquisition cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of graphic modeling technology, and in particular to a progressive modeling method and system for soil type mapping based on sample enhancement. Background Technology

[0002] In complex, multi-class geographic information mapping, especially in hierarchical classification of soil types, traditional methods face challenges such as sparse training samples and high acquisition costs. Field sampling in remote areas is both time-consuming and costly, leading to insufficient data. Furthermore, existing models often fail to effectively utilize multi-source environmental data, resulting in insufficient generalization ability, particularly when dealing with hierarchical classification systems, which may lead to high error rates. Summary of the Invention

[0003] Aimed at at least in solving one of the technical problems existing in the prior art, the present invention provides a progressive modeling method and system for soil type mapping based on sample enhancement, which improves the accuracy of soil modeling and reduces the cost of obtaining soil samples.

[0004] One aspect of the present invention provides a progressive modeling method for soil type mapping based on sample enhancement, comprising: Obtain target soil data, process the target soil data using a soil type mapping model to obtain the modeling result of the target soil data, wherein the target soil data includes geographic coordinates, environmental variables, and land use type labels: The training steps for the soil type mapping model include: A first classification model is generated by training the first soil training data, the first classification model is used to classify the first soil training data, and the classification results are used to enhance the first soil training data to obtain the second soil training data. The second soil training data is divided into multiple training subsets by the land use type label, and the training subsets are enhanced by the virtual sample generation method to obtain the enhanced training subset; By training separately on the augmented training subset, the prediction results of each augmented training subset are obtained; The prediction results of all enhanced training subsets are concatenated to obtain the final prediction result. Based on the final prediction result and through prediction rounds of training, a soil type mapping model is obtained.

[0005] According to the aforementioned progressive modeling method for soil type mapping based on sample enhancement, a first classification model is generated by training first soil training data, the first classification model is used to classify the first soil training data, and the classification result is used as a feature variable to enhance the first soil training data to obtain second soil training data, including: The first soil training data was cleaned and standardized, including selecting environmental variables related to soil characteristics as input features based on soil genesis theory; Random forest, support vector machine or neural network are used to train the first classification model and the classification result. The first classification model is optimized for hyperparameters using Bayesian search method and the model performance is evaluated using five-fold cross-validation method. By using the classification results as feature variables to enhance the first soil training data, the second soil training data is obtained.

[0006] According to the aforementioned progressive modeling method for soil type mapping based on sample enhancement, the second soil training data is divided into multiple training subsets using the land use type labels. These training subsets are then enhanced using a virtual sample generation method to obtain enhanced training subsets, including: Based on the hierarchical relationship of soil types and combined with land use types, the second soil training data is divided and balanced according to land use types to obtain multiple training subsets; When performing classification training on each training subset using the corresponding machine learning algorithm, the virtual samples generated by the Latin hypercube sampling algorithm are combined with the sample size of the training subset and the complexity of the model to obtain the augmented training subset through the real samples and virtual samples of the training subset.

[0007] According to the aforementioned progressive modeling method for soil type mapping based on sample augmentation, the method involves training separately using augmented training subsets to obtain prediction results for each augmented training subset, including: When training separately using an enhanced training subset, the model during training is comprehensively evaluated using the kappa coefficient, accuracy, recall, and F1-score to obtain the evaluation results. If the preset performance threshold is not met according to the evaluation results, virtual samples are added again in the corresponding training subset and the training is repeated until the preset performance threshold is met. The prediction results are obtained by predicting real samples that meet the preset performance threshold.

[0008] According to the aforementioned progressive modeling method for soil type mapping based on sample augmentation, the prediction results of all augmented training subsets are concatenated to obtain the final prediction result. Through training through prediction rounds, a soil type mapping model is obtained, including: All prediction results are spatially stitched together to generate a soil type prediction map for each region. The prediction results of different models are fused using a weighted average or mode voting method to obtain the final prediction result. The samples of the enhanced training subset and the model used for training are dynamically adjusted based on the final prediction result. Closed-loop feedback control is performed until the model performance meets the requirements, thus obtaining the soil type mapping model.

[0009] Another aspect of the present invention discloses a progressive modeling module for soil type mapping based on sample enhancement, comprising: The first module is used to acquire target soil data, process the target soil data using a soil type mapping model, and obtain the modeling results of the target soil data, wherein the target soil data includes geographic coordinates, environmental variables, and land use type labels: The soil type mapping model was trained through the following modules: The second module is used to train and generate a first classification model using the first soil training data, classify the first soil training data using the first classification model, and enhance the first soil training data with the classification results to obtain the second soil training data. The third module is used to divide the second soil training data into multiple training subsets through the land use type label, and enhance the training subsets through a virtual sample generation method to obtain an enhanced training subset. The fourth module is used to train separately using augmented training subsets to obtain the prediction results for each augmented training subset; The fifth module is used to stitch together the prediction results of all augmented training subsets to obtain the final prediction result. Based on the final prediction result and through prediction rounds of training, a soil type mapping model is obtained.

[0010] According to the aforementioned progressive modeling module for soil type mapping based on sample enhancement, the second module includes: The preprocessing unit is used to clean and standardize the first soil training data, and selects environmental variables related to soil characteristics as input features based on soil genesis theory. The basic training unit is used to train the model using random forest, support vector machine or neural network to obtain the first classification model and classification results. The first classification model is optimized for hyperparameters using Bayesian search method and the model performance is evaluated using five-fold cross-validation method. The training data augmentation unit is used to augment the first soil training data by using the classification results as feature variables to obtain the second soil training data.

[0011] According to the aforementioned progressive modeling module for soil type mapping based on sample enhancement, the third module includes: Land division units are used to divide and balance the second soil training data according to land use type based on the hierarchical relationship of soil type and combined with land use type, so as to obtain multiple training subsets; The data augmentation unit is used to augment the training subset by combining virtual samples generated by the Latin hypercube sampling algorithm when classifying each training subset using the corresponding machine learning algorithm, taking into account the number of samples in the training subset and the complexity of the model. The augmented training subset is obtained by combining the real samples and virtual samples of the training subset.

[0012] According to the aforementioned progressive modeling module for soil type mapping based on sample enhancement, the fourth module includes: The evaluation unit is used to comprehensively evaluate the trained model using the kappa coefficient, accuracy, recall, and F1-score when training separately with augmented training subsets, and obtain the evaluation results. The loop training unit is used to add virtual samples to the corresponding training subset and retrain if the evaluation result does not meet the preset performance threshold, until the preset performance threshold is met. The prediction unit is used to predict real samples by meeting preset performance thresholds and obtain prediction results.

[0013] According to the aforementioned progressive modeling module for soil type mapping based on sample enhancement, the fifth module includes: The fusion unit is used to spatially stitch together all prediction results to generate a soil type prediction map for each region. The closed-loop feedback control unit is used to fuse the prediction results of different models using a weighted average or mode voting method to obtain the final prediction result. Based on the final prediction result, the sample of the enhanced training subset and the model used for training are dynamically adjusted, and closed-loop feedback control processing is performed until the model performance meets the requirements, thus obtaining the soil type mapping model.

[0014] The beneficial effects of this invention are as follows: by generating virtual samples, the reliance on real-world sampling is reduced, thus lowering the cost of sample acquisition; the hierarchical structure of first- and second-level models enhances the generalization ability of the model, thereby improving classification accuracy; the feedback mechanism of the kappa coefficient effectively addresses the imbalanced classification problem, ensuring balanced model performance across all categories; data partitioning and hierarchical classification reduce the complexity and error rate of direct fine-grained classification; and the closed-loop feedback control system enables dynamic adjustment of sample generation and model training, ensuring continuous optimization of model performance. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the progressive modeling process for soil type mapping based on sample enhancement, according to an embodiment of the present invention.

[0016] Figure 2 This is a schematic diagram of the second soil training data generation process according to an embodiment of the present invention.

[0017] Figure 3 This is a schematic diagram of the enhanced training subset generation process according to an embodiment of the present invention.

[0018] Figure 4 This is a schematic diagram of the enhanced training subset prediction process according to an embodiment of the present invention.

[0019] Figure 5 This is a schematic diagram of the closed-loop feedback control process according to an embodiment of the present invention.

[0020] Figure 6 This is a schematic diagram of the progressive modeling module for soil type mapping based on sample enhancement, according to an embodiment of the present invention.

[0021] Figure 7 This is the second module of the progressive modeling module for soil type mapping based on sample enhancement, as illustrated in this embodiment of the invention.

[0022] Figure 8 This is the third module in the schematic diagram of the progressive modeling module for soil type mapping based on sample enhancement in this embodiment of the invention.

[0023] Figure 9 This is the fourth module in the schematic diagram of the progressive modeling module for soil type mapping based on sample enhancement in this embodiment of the invention.

[0024] Figure 10 This is a schematic diagram of the fifth module of the progressive modeling module for soil type mapping based on sample enhancement in an embodiment of the present invention. Detailed Implementation

[0025] The embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings. Throughout the description, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions. In the following description, suffixes such as "module," "part," or "unit" used to denote elements are used only for the purpose of illustrative purposes and have no specific meaning in themselves. Therefore, "module," "part," or "unit" can be used interchangeably. Terms such as "first," "second," etc., are used only to distinguish technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the sequential relationship of the indicated technical features. In the following description, the consecutive reference numerals for method steps are for ease of review and understanding. Adjusting the implementation order of steps, in conjunction with the overall technical solution of the present invention and the logical relationship between the various steps, will not affect the technical effect achieved by the technical solution of the present invention. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0026] Figure 1 This is a schematic diagram of a progressive modeling process for soil type mapping based on sample enhancement, according to an embodiment of the present invention, which includes, but is not limited to, steps S100-S500: S100, acquire target soil data, process the target soil data using a soil type mapping model, and obtain the modeling results of the target soil data, which includes geographic coordinates, environmental variables and land use type labels; The training steps for the soil type mapping model include S200~S500: S200: A first classification model is generated by training the first soil training data. The first classification model is used to classify the first soil training data. The classification results are used to enhance the first soil training data to obtain the second soil training data.

[0027] In some embodiments, Figure 2 This is a schematic diagram of the second soil training data generation process, which includes, but is not limited to, steps S210~S230: S210, Cleaning and standardizing the first soil training data, including selecting environmental variables related to soil characteristics as input features based on soil genesis theory; S220 uses random forest, support vector machine or neural network for training to obtain the first classification model and classification results. The first classification model is optimized for hyperparameters using Bayesian search method and the model performance is evaluated using five-fold cross-validation method. S230, by using the classification results as feature variables to enhance the first soil training data, the second soil training data is obtained.

[0028] S300 divides the land use type labels of the second soil training data into multiple training subsets, and enhances the training subsets through a virtual sample generation method to obtain the enhanced training subset.

[0029] In some embodiments, such as Figure 3 The illustrated flowchart for generating augmented training subsets includes, but is not limited to, steps S310 to S320: S310, based on the hierarchical relationship of soil types and combined with land use types, the second soil training data is divided and balanced according to land use types to obtain multiple training subsets; S320: When performing classification training on each training subset using the corresponding machine learning algorithm, the virtual samples generated by the Latin hypercube sampling algorithm are combined with the sample size of the training subset and the complexity of the model to obtain the enhanced training subset through the real samples and virtual samples of the training subset.

[0030] S400 is trained separately using augmented training subsets to obtain the prediction results for each augmented training subset.

[0031] In some embodiments, such as Figure 4 The diagram shown illustrates the strong training subset prediction process, which includes, but is not limited to, steps S410 to S430: S410, when training separately using an enhanced training subset, comprehensively evaluates the trained model using the kappa coefficient, accuracy, recall, and F1-score to obtain the evaluation results; S420. If the preset performance threshold is not met according to the evaluation results, virtual samples are added again in the corresponding training subset and the training is retrained until the preset performance threshold is met. S430 predicts real samples by meeting preset performance thresholds and obtains prediction results.

[0032] S500 concatenates the prediction results of all enhanced training subsets to obtain the final prediction result. Based on the final prediction result and through prediction rounds of training, a soil type mapping model is obtained.

[0033] like Figure 5 The diagram showing the closed-loop feedback control process includes, but is not limited to, steps S510 to S520: S510 spatially stitches together all prediction results to generate a soil type prediction map for each region; S520 uses a weighted average or mode voting method to fuse the prediction results of different models to obtain the final prediction result. Based on the final prediction result, it dynamically adjusts the samples of the enhanced training subset and the model used for training, and performs closed-loop feedback control processing until the model performance meets the requirements, thus obtaining the soil type mapping model.

[0034] Figure 6 This is a schematic diagram of a progressive modeling module for soil type mapping based on sample enhancement, according to an embodiment of the present invention, which includes: The first module is used to acquire target soil data. This data is then processed using a soil type mapping model to obtain the modeling results. The target soil data includes geographic coordinates, environmental variables, and land use type labels. The soil type mapping model was trained through the following modules: The second module is used to train and generate a first classification model using the first soil training data, classify the first soil training data using the first classification model, and enhance the first soil training data with the classification results to obtain the second soil training data. The third module is used to divide the land use type labels of the second soil training data into multiple training subsets, and to enhance the training subsets through a virtual sample generation method to obtain the enhanced training subsets. The fourth module is used to train separately using augmented training subsets to obtain the prediction results for each augmented training subset; The fifth module is used to stitch together the prediction results of all augmented training subsets to obtain the final prediction result. Based on the final prediction result and through prediction rounds of training, a soil type mapping model is obtained.

[0035] like Figure 7 The second module of the progressive modeling module for soil type mapping based on sample augmentation, as shown in the diagram, includes: a preprocessing unit for cleaning and standardizing the first soil training data, and selecting environmental variables related to soil characteristics as input features based on soil genesis theory; a basic training unit for training the first classification model and classification results using random forest, support vector machine, or neural network, wherein the first classification model is optimized for hyperparameters using Bayesian search method and its performance is evaluated using five-fold cross-validation; and a training data augmentation unit for augmenting the first soil training data by using the classification results as feature variables to obtain the second soil training data.

[0036] like Figure 8This is the third module of the schematic diagram of the progressive modeling module for soil type mapping based on sample augmentation. It includes: a land division unit, used to divide and balance the second soil training data according to land use type based on the hierarchical relationship of soil types and combined with land use type to obtain multiple training subsets; and a data augmentation unit, used to generate virtual samples through the Latin hypercube sampling algorithm when classifying and training each training subset using the corresponding machine learning algorithm, taking into account the number of samples in the training subset and the complexity of the model, and to obtain the augmented training subset through the real samples and virtual samples of the training subset.

[0037] like Figure 9 The fourth module of the progressive modeling module for soil type mapping based on sample augmentation shown in the diagram includes: an evaluation unit, used to comprehensively evaluate the model during training using the kappa coefficient, accuracy, recall, and F1-score when training separately with augmented training subsets, and obtain the evaluation result; a loop training unit, used to add virtual samples to the corresponding training subset and retrain if the evaluation result does not meet the preset performance threshold, until the preset performance threshold is met; and a prediction unit, used to predict real samples that meet the preset performance threshold, and obtain the prediction result.

[0038] like Figure 10 The schematic diagram of the progressive modeling module for soil type mapping based on sample augmentation shown is as follows: Module 5: Fusion Unit, which is used to spatially stitch all prediction results to generate a soil type prediction map for each region; Closed-loop Feedback Control Unit, which is used to fuse the prediction results of different models using a weighted average or mode voting method to obtain the final prediction result, dynamically adjust the samples of the augmentation training subset and the model used for training according to the final prediction result, and perform closed-loop feedback control processing until the model performance meets the requirements to obtain the soil type mapping model.

[0039] This invention also provides an electronic device, which includes a processor and a memory; The memory stores the program; The processor executes the program to perform the aforementioned progressive modeling method for soil type mapping based on sample enhancement; the electronic device has the function of carrying and running the software system for progressive modeling of soil type mapping based on sample enhancement provided in the embodiments of the present invention, such as a personal computer, minicomputer, mainframe, workstation, network or distributed computing environment, standalone or integrated computer platform, or communicating with charged particle tools or other imaging devices, etc.

[0040] This invention also provides a computer-readable storage medium storing a program that is executed by a processor to implement the progressive modeling method for soil type mapping based on sample enhancement as described above.

[0041] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented in the embodiments of this invention. Alternative embodiments are contemplated, in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.

[0042] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned progressive modeling method for soil type mapping based on sample enhancement.

[0043] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the described functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, considering the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed in the embodiments of the invention, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0044] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0045] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can include, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0046] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0047] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0048] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0049] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0050] The above is a detailed description of the preferred embodiments of the present invention, but the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A progressive modeling method for soil type mapping based on sample augmentation, characterized in that, include: Obtain target soil data, process the target soil data using a soil type mapping model, and obtain the modeling result of the target soil data, wherein the target soil data includes geographic coordinates, environmental variables, and land use type labels: The training steps for the soil type mapping model include: A first classification model is generated by training the first soil training data, the first classification model is used to classify the first soil training data, and the classification results are used to enhance the first soil training data to obtain the second soil training data. The second soil training data is divided into multiple training subsets by the land use type label, and the training subsets are enhanced by the virtual sample generation method to obtain the enhanced training subset; By training separately on the augmented training subset, the prediction results of each augmented training subset are obtained; The prediction results of all enhanced training subsets are concatenated to obtain the final prediction result. Based on the final prediction result and through prediction rounds of training, a soil type mapping model is obtained.

2. The progressive modeling method for soil type mapping based on sample enhancement according to claim 1, characterized in that, The process of training a first classification model using first soil training data, classifying the first soil training data using the first classification model, and using the classification result as a feature variable to enhance the first soil training data to obtain second soil training data includes: The first soil training data was cleaned and standardized, including selecting environmental variables related to soil characteristics as input features based on soil genesis theory; Random forest, support vector machine or neural network are used to train the first classification model and the classification result. The first classification model is optimized for hyperparameters using Bayesian search method and the model performance is evaluated using five-fold cross-validation method. By using the classification results as feature variables to enhance the first soil training data, the second soil training data is obtained.

3. The progressive modeling method for soil type mapping based on sample enhancement according to claim 1, characterized in that, The step of dividing the second soil training data into multiple training subsets using the land use type label, and then enhancing the training subsets using a virtual sample generation method to obtain an enhanced training subset includes: Based on the hierarchical relationship of soil types and combined with land use types, the second soil training data is divided and balanced according to land use types to obtain multiple training subsets; When performing classification training on each training subset using the corresponding machine learning algorithm, the virtual samples generated by the Latin hypercube sampling algorithm are combined with the sample size of the training subset and the complexity of the model to obtain the augmented training subset through the real samples and virtual samples of the training subset.

4. The progressive modeling method for soil type mapping based on sample enhancement according to claim 3, characterized in that, The step of training separately using augmented training subsets to obtain the prediction results for each augmented training subset includes: When training separately using an enhanced training subset, the model during training is comprehensively evaluated using the kappa coefficient, accuracy, recall, and F1-score to obtain the evaluation results. If the preset performance threshold is not met according to the evaluation results, virtual samples are added again in the corresponding training subset and the training is repeated until the preset performance threshold is met. The prediction results are obtained by predicting real samples that meet the preset performance threshold.

5. The progressive modeling method for soil type mapping based on sample enhancement according to claim 4, characterized in that, The prediction results of all enhanced training subsets are concatenated to obtain the final prediction result. Through training through prediction rounds, a soil type mapping model is obtained, including: All prediction results are spatially stitched together to generate a soil type prediction map for each region. The prediction results of different models are fused using a weighted average or mode voting method to obtain the final prediction result. Based on the final prediction result, the samples of the enhanced training subset and the model used for training are dynamically adjusted, and closed-loop feedback control is performed until the model performance meets the requirements, thus obtaining the soil type mapping model.

6. A progressive modeling module for soil type mapping based on sample augmentation, characterized in that, include: The first module is used to acquire target soil data, process the target soil data using a soil type mapping model, and obtain the modeling results of the target soil data, wherein the target soil data includes geographic coordinates, environmental variables, and land use type labels: The soil type mapping model was trained through the following modules: The second module is used to train and generate a first classification model using the first soil training data, classify the first soil training data using the first classification model, and enhance the first soil training data with the classification results to obtain the second soil training data. The third module is used to divide the second soil training data into multiple training subsets through the land use type label, and enhance the training subsets through a virtual sample generation method to obtain an enhanced training subset. The fourth module is used to train separately using augmented training subsets to obtain the prediction results for each augmented training subset; The fifth module is used to stitch together the prediction results of all augmented training subsets to obtain the final prediction result. Based on the final prediction result and through prediction rounds of training, a soil type mapping model is obtained.

7. The progressive modeling module for soil type mapping based on sample enhancement according to claim 6, characterized in that, The second module includes: The preprocessing unit is used to clean and standardize the first soil training data, and selects environmental variables related to soil characteristics as input features based on soil genesis theory. The basic training unit is used to train the model using random forest, support vector machine or neural network to obtain the first classification model and classification results. The first classification model is optimized for hyperparameters using Bayesian search method and the model performance is evaluated using five-fold cross-validation method. The training data augmentation unit is used to augment the first soil training data by using the classification results as feature variables to obtain the second soil training data.

8. The progressive modeling module for soil type mapping based on sample enhancement according to claim 6, characterized in that, The third module includes: Land division units are used to divide and balance the second soil training data according to land use type based on the hierarchical relationship of soil type and combined with land use type, so as to obtain multiple training subsets; The data augmentation unit is used to augment the training subset by combining virtual samples generated by the Latin hypercube sampling algorithm when classifying each training subset using the corresponding machine learning algorithm, taking into account the number of samples in the training subset and the complexity of the model. The augmented training subset is obtained by combining the real samples and virtual samples of the training subset.

9. The progressive modeling module for soil type mapping based on sample enhancement according to claim 8, characterized in that, The fourth module includes: The evaluation unit is used to comprehensively evaluate the trained model using the kappa coefficient, accuracy, recall, and F1-score when training separately with augmented training subsets, and obtain the evaluation results. The loop training unit is used to add virtual samples to the corresponding training subset and retrain if the evaluation result does not meet the preset performance threshold, until the preset performance threshold is met. The prediction unit is used to predict real samples by meeting preset performance thresholds and obtain prediction results.

10. The progressive modeling module for soil type mapping based on sample enhancement according to claim 9, characterized in that, The fifth module includes: The fusion unit is used to spatially stitch together all prediction results to generate a soil type prediction map for each region. The closed-loop feedback control unit is used to fuse the prediction results of different models using a weighted average or mode voting method to obtain the final prediction result. Based on the final prediction result, the sample of the enhanced training subset and the model used for training are dynamically adjusted, and closed-loop feedback control processing is performed until the model performance meets the requirements, thus obtaining the soil type mapping model.

Images