Evaluation method and system for improving tobacco field soil quality by applying biochar

Through multi-source data acquisition and knowledge graph analysis, an improvement model is constructed and the impact of biochar on tobacco soil is evaluated, which solves the problem that biochar on tobacco soil in the existing technology cannot be predicted, and improves the accuracy of soil quality analysis and efficiency.

CN118536720BActive Publication Date: 2024-11-26INST OF TOBACCO ANHUI ACADEMY OF AGRI SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410978375.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-22
Publication Date
2024-11-26
Estimated Expiration
2044-07-22

AI Technical Summary

Technical Problem

The lack of data support in the prior art and the inability to predict the impact of biochar on tobacco soil, making it difficult to quickly analyze and estimate the effect of soil quality improvement in tobacco soil, and the accuracy of the analysis results is low.

Method used

Through multi-source data collection and preprocessing, a soil status data set is generated, relevant knowledge maps are constructed, evaluation indicators are determined, and improvement models are constructed based on data-driven methods, learning to predict the impact of biochar on tobacco soil quality, and obtaining actual soil quality for comparison , evaluate the improvement effect of biochar.

Benefits of technology

It has achieved rapid analysis and estimate of the improvement effect of soil quality in tobacco fields, improved the accuracy of soil quality analysis results, and provided a data basis for the precise application of biochar and carbon-based fertilizers.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118536720B_ABST
    Figure CN118536720B_ABST
Patent Text Reader

Abstract

The present invention discloses an evaluation method and system for improving tobacco field soil quality by applying biochar, including obtaining a soil status data set of a target tobacco field, extracting knowledge entities and knowledge associations using examples of the impact of biochar on soil quality and plant growth, and constructing a relevant knowledge graph; analyzing the impact direction of applying biochar on the target tobacco field soil quality according to the relevant knowledge graph combined with the soil status data set to determine the evaluation index; constructing a normal quality improvement model based on the evaluation index to predict the ideal soil quality of the target tobacco field after applying biochar; using the evaluation index to evaluate the soil status data after applying biochar to obtain the actual soil quality, and comparing it with the ideal soil quality to obtain the improvement effect of biochar on tobacco field soil. The present invention realizes rapid analysis and prediction of the improvement effect of tobacco field soil quality through a data-driven method, and provides a data basis for the precise application of biochar and charcoal-based fertilizers.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of soil quality assessment, and more specifically, to a method and system for evaluating the effect of applying biochar on improving tobacco field soil quality. Background Art

[0002] Soil is the basis of tobacco production. Good soil quality and a suitable ecological environment are the basis for producing high-quality green tobacco. However, most tobacco fields are currently forced to rotate crops for more than a few years. In addition, unreasonable farming activities such as long-term excessive application of chemical fertilizers in tobacco fields have led to a downward trend in soil quality in tobacco fields, which has restricted the production of high-quality tobacco leaves and the sustainable development of the tobacco industry to a certain extent. Improving soil quality has become the key to improving tobacco production and quality. Soil quality refers to the ability of soil to maintain or improve plant productivity, water and air quality and support human health and life in an ecosystem. The quality of tobacco field soil can be learned through soil quality evaluation. How to accurately evaluate the soil quality of tobacco fields is related to the further improvement of tobacco leaf quality and sustainable production.

[0003] Biochar has stable quality and the ability to adsorb and fix nutrients such as N, P, and K, which helps retain nutrients in the soil. In addition, biochar can increase the organic matter content of the soil, improve the physical and chemical properties of the soil, affect the growth and metabolism of soil microorganisms, and thus change the structure and richness of the microbial community in the soil. With the above excellent characteristics, biochar has demonstrated its ability to improve soil fertility, restore degraded soils, change soil element cycles, and improve nutrient utilization in actual production. After long-term application of biochar, an in-depth evaluation should be conducted to clarify the effect of its application on the improvement of the original soil quality and the direction of its impact. Therefore, how to develop an accurate and efficient evaluation method to evaluate the soil after the application of biochar is an urgent problem to be solved. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a method and system for evaluating the effect of applying biochar on improving tobacco field soil quality, so as to solve the problem that the existing technology lacks data support and cannot predict and evaluate the impact of biochar on tobacco field soil, realizes rapid analysis and estimation of the effect of improving tobacco field soil quality, and improves the accuracy of soil quality analysis results.

[0005] The first aspect of the present invention provides a method for evaluating the effect of biochar application on improving tobacco field soil quality, comprising:

[0006] The soil physical and chemical properties, soil microbial communities and tobacco plant growth conditions of the target tobacco field are obtained through multi-source data collection, and the multi-source data of the target tobacco field are preprocessed to generate a soil status data set of the target tobacco field;

[0007] Obtaining examples of the impact of biochar on soil quality and plant growth, extracting knowledge entities and knowledge associations using the impact examples, and constructing relevant knowledge graphs based on the knowledge entities and knowledge associations;

[0008] Analyze the impact direction of biochar application on soil quality of target tobacco fields according to the relevant knowledge graph combined with the soil status data set, and determine evaluation indicators according to the impact direction;

[0009] Based on the evaluation index, a normal quality improvement model is constructed by a data-driven method, and the model is trained by the impact example to learn and predict the ideal soil quality of the target tobacco field after applying biochar;

[0010] The multi-source data of the target tobacco field after applying biochar is obtained as soil condition data, the soil condition data is evaluated using the evaluation index to obtain the actual soil quality, the actual soil quality is compared with the ideal soil quality, and the improvement effect of biochar on tobacco field soil is obtained.

[0011] In this scheme, the multi-source data of the target tobacco field are preprocessed to generate the soil status data set of the target tobacco field, specifically:

[0012] Determine the geographical scope of the target tobacco field, divide the target tobacco field into grids according to the preset grid size, use multi-source data collection to obtain the soil physical and chemical properties, soil microbial communities and plant growth conditions of soil samples in the grid area, clean and standardize the multi-source data, and use sliding window processing to obtain data features to set grid area labels;

[0013] Set the neighborhood search range, randomly select a grid area, build a search window using the grid area label, calculate the label similarity of other grid areas within the neighborhood search range of the selected grid area, divide the grids that meet the similarity requirements into the same category, and reselect the grid area after the search is completed;

[0014] Traverse all grid areas within the target tobacco field, generate sub-area division results, obtain the similarity between the undivided grid area and each sub-area, divide the undivided network into sub-areas with the highest similarity, aggregate multi-source data in the sub-areas, and generate soil status data sets for each sub-area.

[0015] In this solution, knowledge entities and knowledge associations are extracted using impact instances, and relevant knowledge graphs are constructed based on the knowledge entities and knowledge associations, specifically:

[0016] Retrieve and obtain examples of the impact of biochar on soil quality and plant growth through big data methods, perform structured processing on the impact examples, perform knowledge extraction on the structured impact examples, and use soil parameters and plant parameters as knowledge entity vectors;

[0017] Performing knowledge fusion on the acquired knowledge entity vectors, extracting relationships according to the associations between different knowledge entity vectors, introducing the extracted relationship vectors into a relationship hyperplane, using the application of biochar as a target relationship in the relationship hyperplane, and using the target relationship for screening;

[0018] Determine the parameter deviation after biochar application based on the screened knowledge entity vector, mark it if the parameter deviation is greater than a preset threshold, and generate an impact direction based on the parameter change characteristics of the marked knowledge entity vector;

[0019] Triples are constructed through knowledge entity vectors, target relationships, and impact directions, and relevant knowledge graphs are constructed based on the triples.

[0020] In this solution, the impact of biochar application on the soil quality of the target tobacco field is analyzed based on the relevant knowledge graph combined with the soil status dataset, specifically:

[0021] Locating the soil status data set in the relevant knowledge graph, obtaining the triples involved in the soil parameter entity nodes corresponding to the soil status data set, and obtaining the embedded representations of the entities and relationships in the triples through TransR;

[0022] Get the triplet with each soil parameter entity node as the head node, determine the neighborhood range of each soil parameter entity node through the tail node, get the linear combination of each soil parameter entity node and the neighboring node in the neighborhood range, and introduce the multi-head attention mechanism into the linear combination to calculate the attention weight;

[0023] Set the relationship coefficients corresponding to the head node and the tail node, use the relationship coefficients to weight the attention weights from both the head node and the tail node, and add them to generate the relationship weights;

[0024] The attention weights and relationship weights of neighbor nodes are coupled to generate the weights of triples with each soil parameter entity node as the head node, and the weights of the triples are normalized using a normalization function, and message transmission is performed according to the weight guidance;

[0025] After calculating all linear combinations, the embedding representation of each knowledge entity node is updated by aggregating the weighted neighbor nodes in the neighborhood range, and the final representation is obtained for the soil parameter set and the biochar impact direction set in the knowledge graph;

[0026] The inner product of the final representation of the soil parameters and the final representation of the biochar impact direction is obtained, and the evaluation score of each soil parameter on the biochar impact direction is obtained. The impact direction of the target tobacco field after biochar application is selected according to the evaluation score, and the evaluation index is set by the soil parameters associated with the impact direction.

[0027] In this solution, a normal quality improvement model is constructed based on the evaluation index through a data-driven method, and the model is trained through the impact instance, specifically:

[0028] A normal quality improvement model is constructed based on a stacked autoencoder network, a soil quality improvement threshold is preset, and impact instances that meet the soil quality improvement threshold standard are selected for suspension. Feature parameters are extracted from the impact instances suspended using the determined evaluation indicators, an improvement benchmark data set is constructed, and the training set and validation set are divided.

[0029] The normal quality improvement model is trained by the training set, corresponding training parameters are obtained by encoding and reconstructing the training samples, the training parameters are used as hidden layer features and input into the next layer, and training is performed layer by layer in an unsupervised manner;

[0030] The reconstruction error of the training samples is used for back propagation to optimize the normal quality improvement model. The accuracy of the optimized normal quality improvement model is verified through the validation set, and the normal quality improvement model is output after meeting the accuracy requirements.

[0031] In this scenario, we learn to predict the ideal soil quality of a target tobacco field after applying biochar, specifically:

[0032] The soil status data set of the target tobacco field is used as the model input of the normal quality improvement model, and the estimated values ​​of all parameters corresponding to the evaluation indicators after the application of biochar are predicted and output, and the data mean and data variance of the multidimensional estimated values ​​are calculated to obtain the probability distribution of the multidimensional estimated values;

[0033] The probability distribution is used to characterize the ideal soil quality of the target tobacco field after applying biochar.

[0034] In this scheme, the soil condition data is evaluated using the evaluation index to obtain the actual soil quality, and the actual soil quality is compared with the ideal soil quality to obtain the improvement effect of biochar on tobacco field soil, specifically:

[0035] The evaluation index is used to screen data and extract parameters from soil condition data, and the parameters of the soil condition data subset corresponding to the evaluation index are estimated to obtain the probability distribution representing the actual soil quality.

[0036] The probability distribution of ideal soil quality is compared with the probability distribution of actual soil quality to obtain the probability distribution deviation, and different levels of deviation intervals are preset. The improvement effect level of biochar on tobacco field soil is obtained according to the deviation interval into which the probability distribution deviation falls.

[0037] The second aspect of the present invention provides an evaluation system for improving tobacco field soil quality by applying biochar, comprising: a memory, a processor, a soil multi-source data acquisition module, a related knowledge graph module, a soil quality improvement prediction module and a soil quality improvement evaluation module, wherein the memory and the processor store and execute an evaluation method program for improving tobacco field soil quality by applying biochar;

[0038] The soil multi-source data acquisition module obtains the soil physical and chemical properties, soil microbial community and tobacco plant growth status of the tobacco field soil, and constructs the soil status data set of the tobacco field after preprocessing;

[0039] The related knowledge graph module obtains examples of the effects of biochar on soil quality and plant growth to construct a related knowledge graph, analyzes the direction of the effects of biochar application on the soil quality of the target tobacco field, and determines the evaluation index according to the effects;

[0040] The soil quality improvement prediction module predicts the ideal soil quality of the tobacco field after applying biochar based on the evaluation index through a data-driven method;

[0041] The soil quality improvement assessment module uses evaluation indicators to evaluate soil condition data after applying biochar, obtains actual soil quality, and compares the actual soil quality with the ideal soil quality to evaluate the improvement effect of biochar on tobacco field soil.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] The present invention solves the problem that the prior art lacks data support and cannot predict and evaluate the impact of biochar on tobacco field soil. It realizes rapid analysis and prediction of the tobacco field soil quality improvement effect through a data-driven method, and improves the accuracy of soil quality analysis results, providing a data basis for the precise application of biochar and charcoal-based fertilizers. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments or exemplary embodiments of the present invention, the drawings required for use in the embodiments or exemplary descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained according to the drawings without paying creative work.

[0045] Figure 1 A flow chart showing a method for evaluating the effect of biochar application on improving soil quality in tobacco fields;

[0046] Figure 2 A flow chart is shown for analyzing the direction of the impact of biochar application on soil quality in target tobacco fields;

[0047] Figure 3 A flow chart for constructing a normal quality improvement model is shown;

[0048] Figure 4 A schematic diagram of an evaluation system for improving tobacco field soil quality by applying biochar is shown. DETAILED DESCRIPTION

[0049] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0050] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0051] Figure 1 A flow chart showing a method for evaluating the effect of applying biochar on improving soil quality in tobacco fields.

[0052] like Figure 1 As shown, the present invention provides a method for evaluating the effect of applying biochar on improving tobacco field soil quality, comprising:

[0053] S102, obtaining soil physical and chemical properties, soil microbial communities and tobacco plant growth conditions of the target tobacco field through multi-source data collection, pre-processing the multi-source data of the target tobacco field, and generating a soil status data set of the target tobacco field;

[0054] S104, obtaining examples of the impact of biochar on soil quality and plant growth, extracting knowledge entities and knowledge associations using the impact examples, and constructing a relevant knowledge graph based on the knowledge entities and knowledge associations;

[0055] S106, analyzing the impact direction of applying biochar on the soil quality of the target tobacco field according to the relevant knowledge graph combined with the soil status data set, and determining an evaluation index according to the impact direction;

[0056] S108, constructing a normal quality improvement model through a data-driven method based on the evaluation index, training the model through the impact instance, and learning to predict the ideal soil quality of the target tobacco field after applying biochar;

[0057] S110, obtaining multi-source data of the target tobacco field after applying biochar as soil condition data, using the evaluation index to evaluate the soil condition data to obtain actual soil quality, comparing the actual soil quality with the ideal soil quality, and obtaining the improvement effect of biochar on tobacco field soil.

[0058] It should be noted that the geographical scope of the target tobacco field is determined, the target tobacco field scope is gridded according to the preset grid size, multi-source data collection is used to obtain the soil physical and chemical properties, soil microbial communities and plant growth conditions of soil samples in the grid area, the multi-source data is cleaned and standardized, and the sliding window processing is used to obtain data features to set grid area labels; the neighborhood search range is set, the grid area is randomly selected, and the search window is constructed using the grid area label. The label similarity of other grid areas is calculated within the neighborhood search range of the selected grid area, and the grids that meet the similarity requirements are divided into the same category. After the search is completed, the grid area is reselected; all grid areas within the target tobacco field are traversed to generate sub-area division results, the similarity between the undivided grid area and each sub-area is obtained, the undivided network is divided into the sub-area with the highest similarity, and the multi-source data is converged in the sub-area to generate a soil status data set for each sub-area.

[0059] Through big data methods, we retrieve examples of the impact of biochar on soil quality and plant growth, analyze the impact of biochar or charcoal-based fertilizers on soil physical and chemical properties, soil nutrient content, soil microbial community, plant growth and yield, and perform structured processing on the impact examples. We then perform knowledge extraction on the structured impact examples, such as the Worde2Vec word vector model, which uses a neural network to train the word context to obtain a vector representation of the word, and projects the semantic information of the word into a low-dimensional vector space. The parameters of soil characteristics in the soil parameter and plant parameter characterization examples are used as knowledge entity vectors; the acquired knowledge entity vectors are fused, and the relationship is extracted according to the association between different knowledge entity vectors, and the extracted relationship vector is introduced into the relationship hyperplane, and the application of biochar is used as the target relationship in the relationship hyperplane, and the target relationship is used for screening; the parameter deviation after the application of biochar is judged according to the screened knowledge entity vector, and if the parameter deviation is greater than the preset threshold, it is marked, proving that the parameter fluctuates greatly after the application of biochar, and the influence direction is generated according to the parameter change characteristics of the marked knowledge entity vector; for example, the application of biochar alone is not conducive to the accumulation of fungal and Gram-negative bacterial community biomass in the soil, and the special structure of biochar will react with soil enzymes, thereby promoting or inhibiting the activity of a certain enzyme, and the influence direction on the soil microbial community and soil chemical properties is determined according to the parameter deviation of community biomass and enzyme activity. Triples are constructed through knowledge entity vectors, target relationships and influence directions, and relevant knowledge graphs are constructed according to the triples, and soil parameters that can reflect the influence direction are used as additional features of the influence direction knowledge nodes in the relevant knowledge graph.

[0060] Figure 2 A flow chart for analyzing the direction of impact of biochar application on soil quality in target tobacco fields is shown.

[0061] According to an embodiment of the present invention, the influence of biochar application on the soil quality of the target tobacco field is analyzed based on the relevant knowledge graph combined with the soil status dataset, specifically:

[0062] S202, locating the soil status data set in the relevant knowledge graph, obtaining triples involved in the soil parameter entity nodes corresponding to the soil status data set, and obtaining embedded representations of entities and relationships in the triples through TransR;

[0063] S204, obtaining a triplet with each soil parameter entity node as a head node, determining the neighborhood range of each soil parameter entity node through the tail node, obtaining a linear combination of each soil parameter entity node and neighboring nodes within the neighborhood range, and introducing a multi-head attention mechanism into the linear combination to calculate the attention weight;

[0064] S206, setting the relationship coefficients corresponding to the head node and the tail node, weighting the attention weights from both the head node and the tail node using the relationship coefficients, and generating the relationship weights after adding them up;

[0065] S208, coupling the attention weights and relationship weights of neighbor nodes to generate weights of triples with each soil parameter entity node as the head node, and normalizing the weights of the triples using a normalization function, and performing message transmission according to the weight guidance;

[0066] S210, after calculating all linear combinations, the embedding representation of each knowledge entity node is updated by aggregating the weighted neighbor nodes in the neighborhood range, and the final representation is obtained for the soil parameter set and the biochar impact direction set in the knowledge graph;

[0067] S212, obtaining the inner product of the final representation of the soil parameters and the final representation of the biochar impact direction, obtaining the evaluation score of each soil parameter on the biochar impact direction, selecting the impact direction of the target tobacco field after biochar application according to the evaluation score, and setting the evaluation index through the soil parameters associated with the impact direction.

[0068] It should be noted that in the representation layer, the entities and relationships in the relevant knowledge graph are represented as vectors through the TransR model, and the relevant knowledge graph structure is retained. In the propagation layer, a set of triples with each soil parameter entity node as the head node is obtained, and the tail node in the triple is used as the neighbor node. The linear combination of each soil parameter entity node and the neighbor node is obtained to represent the first-order connectivity structure of each soil parameter entity node. The multi-head attention mechanism is introduced in the linear combination to calculate the attention weight of each neighbor node. , ,in is the query vector, is the key vector, is a value vector, is the number of neighbor nodes, is the scaling factor. In addition, the influence of the relationship on the weight is considered from both the head node and the tail node, and the head node attention weight is set The relationship coefficient a and the attention weight of neighbor nodes The relationship coefficient is 1-a. and Perform weighted summation to obtain the relationship weight. Couple the attention weight and relationship weight of neighbor nodes. The higher the coupling weight, the more effective information the triple propagates. Through neighbor aggregation and repeated propagation, high-order information of related graphs is extracted. In the prediction layer, the layer aggregation mechanism connects the output representation of each layer into a vector to obtain the final representation of soil parameters. The final expression of the biochar impact direction , the evaluation score of the influence of each soil parameter on biochar is obtained through the inner product of the two .

[0069] Figure 3 A flow chart for constructing a normal quality improvement model is shown.

[0070] According to an embodiment of the present invention, a normal quality improvement model is constructed by a data-driven method based on the evaluation index, and the model is trained by the impact instance, specifically:

[0071] S302, constructing a normal quality improvement model based on a stacked autoencoder network, presetting a soil quality improvement threshold, selecting an impact instance that meets the soil quality improvement threshold standard for suspension, extracting feature parameters from the impact instance suspended using the determined evaluation index, constructing an improvement benchmark data set, and dividing it into a training set and a validation set;

[0072] S304, training the normal quality improvement model through the training set, obtaining corresponding training parameters through encoding and reconstructing the training samples, using the training parameters as hidden layer features and inputting them into the next layer, and performing layer-by-layer training in a layer-by-layer unsupervised manner;

[0073] S306, using the reconstruction error of the training sample to perform back propagation, optimize the normal quality improvement model, verify the accuracy of the optimized normal quality improvement model through the verification set, and output the normal quality improvement model after meeting the accuracy requirements.

[0074] It should be noted that data-driven modeling is based on the examples of the effects of excellent improvement on soil quality and plant growth after treatment, and a model that can remember and learn the normal improvement effect of biochar on soil quality is established. The learning goal of the constructed normal quality improvement model is to reconstruct the soil status data set of the target tobacco field. The input of the normal quality improvement model is the measured value of all parameters corresponding to the evaluation index, and the output is the estimated value of all parameters corresponding to the evaluation index at the normal improvement level after a preset time. The soil status data set of the target tobacco field is used as the model input of the normal quality improvement model, and the estimated values ​​of all parameters corresponding to the evaluation index after the application of biochar are predicted and output. The estimated value sequence is decomposed using the sliding window method, and the current sliding window data is processed to improve the normality of the data. The data mean and data variance of the multidimensional estimated value are calculated to obtain the probability distribution of the multidimensional estimated value; the ideal soil quality of the target tobacco field after the application of biochar is characterized by the probability distribution.

[0075] The evaluation index is used to screen data and extract parameters in the soil condition data, and the soil condition data subset corresponding to the evaluation index is estimated to obtain the probability distribution that characterizes the actual soil quality; the probability distribution of the ideal soil quality is compared with the probability distribution of the actual soil quality to obtain the probability distribution deviation, and the impact examples of different soil improvement effects are statistically analyzed. Several improvement levels are set in combination with expert experience, and the iconic soil condition data of the improvement level are set. The cluster center is set using the iconic soil condition data to cluster the impact examples, and the impact example subsets of different improvement levels are obtained; the residual between the measured value and the estimated value is calculated for each impact example subset using the normal quality improvement model, and the deviation interval of different levels is preset according to the residual distribution. The improvement effect level of biochar on tobacco field soil is obtained according to the deviation interval into which the probability distribution deviation falls. Preferably, the relative entropy between the probability distribution of ideal soil quality and the probability distribution of actual soil quality is used as an indicator for evaluating the improvement effect of biochar on tobacco field soil quality, which can provide more accurate and comprehensive information on soil quality changes. When the value of the relative entropy is relatively small, it means that the improvement effect of biochar on the target tobacco field soil quality is better.

[0076] The improvement effect level of biochar on tobacco field soil is analyzed. When the improvement effect level is less than the preset level threshold, an early warning message is generated. The optimization direction of the target tobacco field soil quality is obtained according to the deviation between the probability distribution of the ideal soil quality and the probability distribution of the actual soil quality. Based on the optimization direction, a carbon-based fertilizer improvement example is obtained. The recommended type of carbon-based fertilizer is obtained by calculating the similarity of soil condition data. The deviation value of the actual soil quality and the ideal soil quality in the optimization direction of the target tobacco field after applying biochar is analyzed. According to the target tobacco field environmental data and the growth condition of tobacco plants and the similarity deviation of the carbon-based fertilizer improvement example, an uncertainty coefficient is generated. The amount of carbon-based fertilizer is determined by the deviation value of the optimization direction, and the uncertainty coefficient is used for weighted generation of the recommended amount of carbon-based fertilizer, so as to ensure that the target tobacco field can achieve the ideal soil quality after adding carbon-based fertilizer.

[0077] Figure 4 A block diagram of an evaluation system for improving tobacco field soil quality by applying biochar is shown.

[0078] This embodiment provides an evaluation system 4 for improving tobacco field soil quality by applying biochar, characterized in that it includes a memory 41, a processor 42, a soil multi-source data acquisition module 43, a related knowledge graph module 44, a soil quality improvement prediction module 45 and a soil quality improvement evaluation module 46, wherein the memory 41 and the processor 46 store and execute an evaluation method program for improving tobacco field soil quality by applying biochar;

[0079] The soil multi-source data acquisition module 43 obtains the soil physical and chemical properties, soil microbial community and tobacco plant growth status of the tobacco field soil, and constructs the soil status data set of the tobacco field after preprocessing;

[0080] The related knowledge graph module 44 obtains examples of the effects of biochar on soil quality and plant growth to construct a related knowledge graph, analyzes the direction of the effects of biochar application on the soil quality of the target tobacco field, and determines the evaluation index according to the effects;

[0081] The soil quality improvement prediction module 45 predicts the ideal soil quality of the tobacco field after applying biochar based on the evaluation index through a data-driven method;

[0082] The soil quality improvement evaluation module 46 uses evaluation indicators to evaluate the soil condition data after applying biochar, obtains actual soil quality, and compares the actual soil quality with the ideal soil quality to evaluate the improvement effect of biochar on tobacco field soil.

[0083] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0084] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0085] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0086] A person of ordinary skill in the art can understand that: all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0087] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0088] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for evaluating the effect of biochar application on improving tobacco field soil quality, characterized in that: The following steps are involved: The soil physical and chemical properties, soil microbial communities and tobacco plant growth conditions of the target tobacco field are obtained through multi-source data collection, and the multi-source data of the target tobacco field are preprocessed to generate a soil status data set of the target tobacco field; Obtaining examples of the impact of biochar on soil quality and plant growth, extracting knowledge entities and knowledge associations using the impact examples, and constructing relevant knowledge graphs based on the knowledge entities and knowledge associations; Analyze the impact direction of biochar application on soil quality of target tobacco fields according to the relevant knowledge graph combined with the soil status data set, and determine evaluation indicators according to the impact direction; Based on the evaluation index, a normal quality improvement model is constructed by a data-driven method, and the model is trained by the impact example to learn and predict the ideal soil quality of the target tobacco field after applying biochar; Acquire multi-source data of the target tobacco field after applying biochar as soil condition data, use the evaluation index to evaluate the soil condition data to obtain actual soil quality, compare the actual soil quality with the ideal soil quality, and obtain the improvement effect of biochar on tobacco field soil; The impact instances are used to extract knowledge entities and knowledge associations, and relevant knowledge graphs are constructed based on the knowledge entities and knowledge associations, specifically: Retrieve and obtain examples of the impact of biochar on soil quality and plant growth through big data methods, perform structured processing on the impact examples, perform knowledge extraction on the structured impact examples, and use soil parameters and plant parameters as knowledge entity vectors; Performing knowledge fusion on the acquired knowledge entity vectors, extracting relationships according to the associations between different knowledge entity vectors, introducing the extracted relationship vectors into a relationship hyperplane, using the application of biochar as a target relationship in the relationship hyperplane, and using the target relationship for screening; Determine the parameter deviation after biochar application based on the screened knowledge entity vector, mark it if the parameter deviation is greater than a preset threshold, and generate an impact direction based on the parameter change characteristics of the marked knowledge entity vector; Constructing triples through knowledge entity vectors, target relationships, and impact directions, and constructing relevant knowledge graphs based on the triples; According to the relevant knowledge graph combined with the soil status data set, the impact direction of applying biochar on the soil quality of the target tobacco field is analyzed, specifically: Locating the soil status data set in the relevant knowledge graph, obtaining the triples involved in the soil parameter entity nodes corresponding to the soil status data set, and obtaining the embedded representations of the entities and relationships in the triples through TransR; Get the triplet with each soil parameter entity node as the head node, determine the neighborhood range of each soil parameter entity node through the tail node, get the linear combination of each soil parameter entity node and the neighboring node in the neighborhood range, and introduce the multi-head attention mechanism into the linear combination to calculate the attention weight; Set the relationship coefficients corresponding to the head node and the tail node, use the relationship coefficients to weight the attention weights from both the head node and the tail node, and add them to generate the relationship weights; The attention weights and relationship weights of neighbor nodes are coupled to generate the weights of triples with each soil parameter entity node as the head node, and the weights of the triples are normalized using a normalization function, and message transmission is performed according to the weight guidance; After calculating all linear combinations, the embedding representation of each knowledge entity node is updated by aggregating the weighted neighbor nodes in the neighborhood range, and the final representation is obtained for the soil parameter set and the biochar impact direction set in the knowledge graph; Obtaining the inner product of the final representation of the soil parameter and the final representation of the biochar impact direction, obtaining the evaluation score of each soil parameter on the biochar impact direction, selecting the impact direction of the target tobacco field after applying biochar according to the evaluation score, and setting the evaluation index through the soil parameters associated with the impact direction; Based on the evaluation index, a normal quality improvement model is constructed by a data-driven method, and the model is trained through the impact examples. The data-driven modeling is based on the impact examples of biochar with excellent improvement effects on soil quality and plant growth after treatment, and a model capable of memorizing and learning the normal improvement effect of biochar on soil quality is established, specifically: A normal quality improvement model is constructed based on a stacked autoencoder network, a soil quality improvement threshold is preset, and impact instances that meet the soil quality improvement threshold standard are selected for suspension. Feature parameters are extracted from the suspended impact instances using the determined evaluation indicators, an improvement benchmark data set is constructed, and the training set and validation set are divided. The normal quality improvement model is trained by the training set, corresponding training parameters are obtained by encoding and reconstructing the training samples, the training parameters are used as hidden layer features and input into the next layer, and training is performed layer by layer in an unsupervised manner; The reconstruction error of the training samples is used for back propagation to optimize the normal quality improvement model. The accuracy of the optimized normal quality improvement model is verified through the validation set. The normal quality improvement model is output after meeting the accuracy requirements. The input of the normal quality improvement model is the measured values ​​of all parameters corresponding to the evaluation indicators, and the output is the estimated values ​​of all parameters corresponding to the evaluation indicators at the normal improvement level after a preset time.

2. The method for evaluating the effect of biochar application on improving tobacco field soil quality according to claim 1, characterized in that: The multi-source data of the target tobacco field are preprocessed to generate the soil status data set of the target tobacco field, specifically: Determine the geographical scope of the target tobacco field, divide the target tobacco field into grids according to the preset grid size, use multi-source data collection to obtain the soil physical and chemical properties, soil microbial communities and plant growth conditions of soil samples in the grid area, clean and standardize the multi-source data, and use sliding window processing to obtain data features to set grid area labels; Set the neighborhood search range, randomly select a grid area, build a search window using the grid area label, calculate the label similarity of other grid areas within the neighborhood search range of the selected grid area, divide the grids that meet the similarity requirements into the same category, and reselect the grid area after the search is completed; Traverse all grid areas within the target tobacco field, generate sub-area division results, obtain the similarity between the undivided grid area and each sub-area, divide the undivided grid area into sub-areas with the highest similarity, aggregate multi-source data in the sub-areas, and generate soil status data sets for each sub-area.

3. The method for evaluating the effect of biochar application on improving tobacco field soil quality according to claim 1, characterized in that: Learn to predict the ideal soil quality of a target tobacco field after biochar application, specifically: The soil status data set of the target tobacco field is used as the model input of the normal quality improvement model, and the estimated values ​​of all parameters corresponding to the evaluation indicators after the application of biochar are predicted and output, and the data mean and data variance of the multidimensional estimated values ​​are calculated to obtain the probability distribution of the multidimensional estimated values; The probability distribution is used to characterize the ideal soil quality of the target tobacco field after applying biochar.

4. The method for evaluating the effect of biochar application on improving tobacco field soil quality according to claim 1, characterized in that: The soil condition data is evaluated using the evaluation index to obtain the actual soil quality, and the actual soil quality is compared with the ideal soil quality to obtain the improvement effect of biochar on tobacco field soil, specifically: The evaluation index is used to screen data and extract parameters from soil condition data, and the parameters of the soil condition data subset corresponding to the evaluation index are estimated to obtain the probability distribution representing the actual soil quality. The probability distribution of ideal soil quality is compared with the probability distribution of actual soil quality to obtain the probability distribution deviation, and different levels of deviation intervals are preset. The improvement effect level of biochar on tobacco field soil is obtained according to the deviation interval into which the probability distribution deviation falls.

5. An evaluation system for improving tobacco field soil quality by applying biochar, characterized in that: Used to execute the evaluation method for improving tobacco field soil quality by applying biochar as described in any one of claims 1 to 4, the system comprises: a memory, a processor, a soil multi-source data acquisition module, a related knowledge graph module, a soil quality improvement prediction module and a soil quality improvement evaluation module, the memory and the processor store and execute a program for evaluating the method for improving tobacco field soil quality by applying biochar; The soil multi-source data acquisition module obtains the soil physical and chemical properties, soil microbial community and tobacco plant growth status of the tobacco field soil, and constructs the soil status data set of the tobacco field after preprocessing; The related knowledge graph module obtains examples of the effects of biochar on soil quality and plant growth to construct a related knowledge graph, analyzes the direction of the effects of biochar application on the soil quality of the target tobacco field, and determines the evaluation index according to the effects; The soil quality improvement prediction module predicts the ideal soil quality of the tobacco field after applying biochar based on the evaluation index through a data-driven method; The soil quality improvement assessment module uses evaluation indicators to evaluate soil condition data after applying biochar, obtains actual soil quality, and compares the actual soil quality with the ideal soil quality to evaluate the improvement effect of biochar on tobacco field soil.