Orchard cover crop cultivation management method and system based on multi-source environmental parameters

Abstract

This invention discloses a method and system for orchard cover crop management based on multi-source environmental parameters, belonging to the field of smart agriculture technology. The method includes: acquiring full-process management data for orchard cover crop cultivation; extracting key management measures for cover crop cultivation; acquiring environmental impact factors for each key management measure; constructing data-driven models corresponding to each key management measure; and establishing a mapping relationship between environmental impact factors and cover crop management. It also involves collecting multi-source environmental parameters from the target orchard, selecting cover crop species and providing early warnings for cover crop management based on the multi-source environmental parameters and the data-driven models; and obtaining the optimal cover crop management plan by associating different key management measures with the cover crop management early warnings. This invention uses environmental parameters to select cover crop species and scientifically manage planting in orchards, promoting the maintenance of a complete food chain and industrial chain in the orchard and fostering the healthy growth of fruit trees.

Description

Technical Field

[0001] This invention relates to the field of smart agriculture technology, and more specifically, to a method and system for orchard grass cultivation management based on multi-source environmental parameters. Background Technology

[0002] Currently, many orchards suffer from excessive weeds, hindering tree growth. While soil cleaning and herbicide application can effectively eliminate weeds, they are detrimental to fruit tree growth and the sustainable development of the orchard. Soil cover cultivation, as an advanced orchard soil management model, plays a crucial role in coordinating the sustainable development of fruit trees and the environment. It effectively improves soil structure, enhances soil fertility, and regulates the orchard microclimate. Orchard soil cover is currently being widely promoted and applied. This method primarily involves planting grass between rows of fruit trees, between individual trees, or throughout the orchard. The grass used is mostly annual or biennial, or perennial herbaceous plants. Orchard soil cover has heat insulation and moisture retention properties, reducing soil temperature rise caused by daytime sun exposure and mitigating nighttime ground cooling, thus narrowing the diurnal temperature range and making soil temperature changes more gradual throughout the day. It also effectively increases soil organic matter content, enhances soil fertility, improves fruit quality, and increases fruit yield.

[0003] Soil cover cultivation not only conserves water and soil, improves the orchard environment, and promotes vigorous growth of fruit trees, but also significantly increases fruit yield and improves fruit quality, resulting in more substantial economic benefits. However, at present, the management of soil cover cultivation is mostly determined by human intervention, such as the selection of grass species, control of grass growth, and water and fertilizer management. This human intervention is inefficient and often negatively impacts soil health and fruit tree cultivation. Because orchard soil cover management is complex, influenced by numerous environmental factors with interrelationships, and therefore, how to manage the entire process of soil cover cultivation based on multiple environmental parameters is a problem that needs to be addressed. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a method and system for orchard cover cultivation and management based on multi-source environmental parameters. The aim is to scientifically manage orchard cover according to environmental parameters, thereby promoting the maintenance of a complete food chain and industrial chain in the orchard and fostering the healthy growth of fruit trees.

[0005] The first aspect of this invention provides a method for orchard cover crop management based on multi-source environmental parameters, comprising the following steps:

[0006] Acquire full-process management data of orchard grass cultivation, extract key management measures for grass cultivation, divide the full-process management data based on the key management measures, and obtain the environmental impact factors of each key management measure based on correlation analysis;

[0007] Environmental parameters are extracted based on the environmental influencing factors. Environmental parameters that meet the preset requirements are selected by using the growth status of the grass. Data-driven models corresponding to each key management measure are constructed using the environmental parameters to establish a mapping relationship between environmental influencing factors and grass cultivation management.

[0008] Collect multi-source environmental parameters of the target orchard, and based on the multi-source environmental parameters and data-driven model, conduct grass seed selection and grass management early warning for the target orchard;

[0009] The optimal grass management plan is obtained by associating different key management measures with grass management early warnings. The grass species selection information, grass management early warning information and grass management plan are visualized using visualization methods.

[0010] In this solution, the entire process management data of orchard cover cropping is acquired, key management measures for cover cropping are extracted, and the entire process management data is divided based on these key management measures, specifically as follows:

[0011] A dataset was constructed by using a big data retrieval engine to obtain relevant knowledge and historical examples of orchard grass cultivation management. The dataset was preprocessed and sorted according to the time sequence corresponding to the growth stage of orchard grass to obtain full-process management data of orchard grass cultivation.

[0012] Clustering is performed on the full-process management data to obtain clustering results. The data volume of each cluster in the clustering results is obtained. Clusters that do not meet the preset data volume requirements are removed based on the data volume, thereby simplifying the full-process management data.

[0013] In the remaining clusters, keywords are extracted from the management data samples in each cluster. Based on the keyword vector, the corresponding grass growth stage and corresponding management measures are determined. After redundancy removal, the key management measures for each grass growth stage are obtained.

[0014] Based on the aforementioned key management measures, the data is located within the simplified full-process management data, and neighborhood data of the key management measures is extracted according to a preset time range.

[0015] In this plan, the environmental impact factors of each key management measure are obtained based on correlation analysis, specifically:

[0016] Data segments corresponding to key management measures for orchard cover cultivation are obtained. Correlation analysis is used to perform primary attribution of key management measures in the data segments. The Pearson correlation coefficient is used to improve the maximum correlation minimum redundancy method. Environmental influencing factors that meet the preset correlation conditions are obtained through the improved maximum correlation minimum redundancy method.

[0017] The environmental impact factors selected for primary attribution are obtained. Based on the environmental impact factors, environmental parameters and key management measures are extracted from the data segments to construct data samples. The SHAP interpretation model is used to perform deep attribution analysis on the data samples to generate Shapley values ​​for each environmental impact factor and construct a local interpretation matrix.

[0018] For each environmental impact factor, the average Shapley value of all data samples for that environmental impact factor is calculated based on the local interpretation matrix as a global interpretation. The global interpretation is used as the importance value of the environmental impact factor. The importance value is used for ranking. Based on the ranking results, a preset number of environmental impact factors are selected and matched with corresponding key management measures.

[0019] In this scheme, environmental parameters are extracted based on the aforementioned environmental influencing factors. The growth status of the grass is used to screen for environmental parameters that meet preset requirements. These environmental parameters are then used to construct data-driven models corresponding to each key management measure. Specifically:

[0020] Key management measures are divided according to the selection of grass species and the management of grass plants. Environmental parameters are screened based on the environmental impact factors matched by each key management measure after the division. Grass growth stages and grass growth status are obtained from the instances corresponding to the full-process management data. Growth status standards for different grass growth stages are generated based on grass growth data under suitable conditions.

[0021] By comparing the grass growth status in the example with the corresponding growth status standard, environmental parameters that meet the preset requirements are obtained, and a set of environmental parameters corresponding to grass species selection and grass plant management is generated.

[0022] Data-driven models for grass seed selection and grass plant management were constructed based on support vector machines and autoencoder networks, respectively. The corresponding data-driven models were trained using environmental parameter sets combined with orchard grass seed selection data and orchard grass management data. The mapping relationship between environmental influencing factors and grass cultivation management was constructed from two aspects.

[0023] In this scheme, multi-source environmental parameters of the target orchard are collected. Based on these multi-source environmental parameters and a data-driven model, the target orchard is provided with grass seed selection and grass management early warning. Specifically, the grass seed selection involves:

[0024] Based on the selection of grass species, select the corresponding environmental influencing factors and collect multi-source environmental parameters of the target orchard. Use the multi-source environmental parameters as input to the support vector machine to generate corresponding feature vectors. Use the data-driven model formed by the trained support vector machine to predict and score the feature vectors.

[0025] The gscatter function is used to obtain a scatter plot of data corresponding to the feature vectors. The contour function is used to generate decision boundaries for different grass species labels in the scatter plot. The evaluation scores of different decision boundaries are calculated. Based on the evaluation scores, the grass species selection results for the target orchard are generated.

[0026] In this solution, multi-source environmental parameters of the target orchard are collected. Based on these multi-source environmental parameters and a data-driven model, the target orchard is provided with grass seed selection and grass management early warning. Specifically, the grass management early warning includes:

[0027] The growth stage of the target orchard's cover grass is obtained. Based on the growth stage, corresponding environmental influencing factors are collected from the environmental influencing factors corresponding to the cover grass plant management. The corresponding multi-source environmental parameters are then imported into the autoencoder network.

[0028] The data-driven model constructed by the trained autoencoder network is used to reconstruct the multi-source environmental parameters, obtain the estimated values ​​of the multi-source environmental parameters after the implementation of key management measures, and calculate the multidimensional residuals between the estimated values ​​and the measured values ​​of the multi-source environmental parameters.

[0029] The residual thresholds for the data-driven models corresponding to each key management measure are preset. The multidimensional residuals are compared with the corresponding residual thresholds, and the number of multidimensional residuals exceeding the residual thresholds is counted. When the number reaches the preset standard, a grass management early warning is generated.

[0030] In this plan, the optimal grass management plan is associated with the early warning system for different key management measures, specifically as follows:

[0031] Based on the aforementioned grass management early warning, corresponding key management measures are obtained. Using the key management measures and multi-source environmental parameters, a similarity search is performed on historical orchard grass cultivation management examples. Based on the similarity ranking, the historical orchard grass cultivation management example with the highest similarity is obtained.

[0032] Management parameters are extracted from historical orchard cover cropping management examples. Environmental deviations between the target orchard and the examples are calculated based on multi-source environmental parameters. Adjustment coefficients are generated based on the environmental deviations to adjust the management parameters. The adjusted management parameters are then used to configure the key management measures to generate the optimal cover cropping management plan.

[0033] The second aspect of the present invention provides an orchard cover cultivation management system based on multi-source environmental parameters. The system includes an environmental impact factor analysis unit, a multi-source environmental parameter acquisition unit, a data-driven model unit, and an orchard cover cultivation management output unit.

[0034] The environmental impact factor analysis unit uses the acquired knowledge of orchard cover cultivation and historical examples of orchard cover cultivation management to construct full-process management data, extract key management measures for cover cultivation, and obtain the environmental impact factors of key management measures based on correlation analysis.

[0035] The multi-source environmental parameter acquisition unit extracts multi-source environmental parameters based on environmental influencing factors at different growth stages;

[0036] The data-driven model unit constructs data-driven models for grass seed selection and grass plant management based on support vector machines and autoencoder networks, respectively. It trains the corresponding data-driven models using environmental parameter sets combined with orchard grass seed selection data and orchard grass management data, constructing a mapping relationship between environmental influencing factors and grass cultivation management from two aspects. Based on the multi-source environmental parameters of the target orchard and the data-driven model, it provides early warning for grass seed selection and grass management in the target orchard.

[0037] The orchard cover crop management output unit obtains the optimal cover crop management plan based on the cover crop management early warning associated with different key management measures, outputs cover crop selection information, cover crop management early warning information and cover crop management plan, and displays them visually using visualization methods.

[0038] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0039] This invention uses attribution analysis to screen environmental influencing factors in orchard cover crop management practices. It constructs corresponding data-driven models for cover crop selection and cover crop management, enabling early warning for cover crop selection and management. Based on environmental parameters, it scientifically manages orchard cover crop to ensure that cover crop has a positive impact on fruit trees, promotes the maintenance of a complete food chain and industrial chain in the orchard, and fosters the healthy growth of fruit trees. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments or examples of the present invention, the drawings used in the embodiments or examples will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained according to these drawings without creative effort.

[0041] Figure 1 A flowchart illustrating the orchard cover crop management method based on multi-source environmental parameters is shown.

[0042] Figure 2 A flowchart illustrating the acquisition of environmental impact factors for each key management measure is shown;

[0043] Figure 3The flowchart illustrates the construction of data-driven models corresponding to each key management measure;

[0044] Figure 4 A block diagram of an orchard cover crop management system based on multi-source environmental parameters is shown. Detailed Implementation

[0045] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0046] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0047] like Figure 1 As shown, the first embodiment of the present invention provides a method for orchard cover crop management based on multi-source environmental parameters, including:

[0048] S102, acquire the full-process management data of orchard grass cultivation, extract the key management measures of grass cultivation, divide the full-process management data based on the key management measures, and obtain the environmental impact factors of each key management measure based on correlation analysis.

[0049] S104, Based on the environmental influencing factors, extract environmental parameters, use the growth status of the grass to screen environmental parameters that meet the preset requirements, use the environmental parameters to construct a data-driven model corresponding to each key management measure, and establish a mapping relationship between environmental influencing factors and grass cultivation management.

[0050] S106, Collect multi-source environmental parameters of the target orchard, and based on the multi-source environmental parameters and data-driven model, perform grass seed selection and grass management early warning for the target orchard;

[0051] S108. Based on the early warning of grass management, the optimal grass management plan is obtained by associating different key management measures. The grass species selection information, grass management early warning information and grass management plan are visualized using visualization methods.

[0052] It should be noted that a dataset was constructed using a big data retrieval engine to obtain relevant knowledge and historical examples of orchard cover cultivation management. This dataset underwent preprocessing such as data cleaning, dimensionality reduction, and normalization. Data was then sorted according to the time sequence corresponding to the growth stages of the orchard cover to obtain comprehensive management data for the entire orchard cover cultivation process. Clustering was performed on this comprehensive management data to obtain clustering results. The data volume of each cluster was then determined, and clusters that did not meet the preset data volume requirements were removed, simplifying the comprehensive management data. If the data volume of a cluster is too small, it indicates that the corresponding management data lacks universality and does not fall under the category of matters requiring special attention in cover cultivation management. In the remaining clusters, keywords were extracted from the management data samples in each cluster. Based on the keyword vectors, the corresponding cover growth stage and corresponding management measures were determined, such as orchard drainage during the seedling stage and the removal of weeds and harmful weeds to prevent malicious competition with the cover. By removing redundancy from similar expressions of management measures, key management measures for each grass growth stage are obtained, such as sowing management, seedling management, water and fertilizer management, and mowing management. Based on these key management measures, they are located in the simplified full-process management data, and neighborhood data of the key management measures are extracted according to a preset time range.

[0053] Figure 2 A flowchart is shown to obtain the environmental impact factors of each key management measure.

[0054] According to an embodiment of the present invention, the environmental impact factors of each key management measure are obtained based on correlation analysis, specifically as follows:

[0055] S202, obtain the data segment corresponding to the key management measures of orchard grass cultivation, use correlation analysis in the data segment to perform primary attribution of key management measures, use Pearson correlation coefficient to improve the maximum correlation minimum redundancy method, and obtain environmental influencing factors that meet the preset correlation conditions through the improved maximum correlation minimum redundancy method.

[0056] S204, Obtain the environmental impact factors selected for primary attribution, extract environmental parameters and key management measures from the data segment based on the environmental impact factors to construct a data sample, use the SHAP interpretation model to perform deep attribution analysis on the data sample, generate the Shapley value for each environmental impact factor to construct a local interpretation matrix;

[0057] S206, for each environmental impact factor, the average value of the Shapley value of all data samples for that environmental impact factor is calculated based on the local interpretation matrix as the global interpretation. The global interpretation is used as the importance value of the environmental impact factor. The importance value is used for sorting. Based on the sorting results, a preset number of environmental impact factors are selected and matched with corresponding key management measures.

[0058] It should be noted that the SHAP interpretation model, based on cooperative game theory and local interpretation, calculates the contribution of each feature to the prediction result according to the marginal contribution of each feature in the instance, thus achieving an interpretation for each instance. Using the SHAP interpretation model, for each data sample, the local interpretation calculates the corresponding value, i.e., the Shapley value, for each environmental impact factor, representing its contribution to the prediction result. For any environmental impact factor, its weighted summation over all possible combinations of feature values ​​yields the Shapley value for that environmental impact factor. During training, the Shapley values ​​of each environmental impact factor and the Shapley mean of all data samples are obtained. When the Shapley value of an environmental impact factor is greater than the Shapley mean, a positive impact is generated; otherwise, a negative impact is generated. For a given feature variable, the Shapley values ​​corresponding to that environmental impact factor in all data samples are calculated, and their average is used as the importance value of that environmental impact factor, thus obtaining a global interpretation. Preferably, in the selection of grass species for cover, the environmental impact factors include soil physicochemical properties, climatic conditions, orchard weed characteristics, and orchard-related pest and disease characteristics, etc.

[0059] Figure 3 The flowchart shows the process of constructing a data-driven model for each key management measure.

[0060] According to an embodiment of the present invention, a data-driven model corresponding to each key management measure is constructed to establish a mapping relationship between environmental influencing factors and grass cover cultivation management, specifically as follows:

[0061] S302, the key management measures are divided according to the selection of grass species and the management of grass plants. Based on the environmental impact factors matched by each key management measure after division, environmental parameters are screened, and the grass growth stage and grass growth status are obtained in the instances corresponding to the full-process management data. Based on the grass growth data under suitable conditions, growth status standards for different grass growth stages are generated.

[0062] S304, compare the grass growth status in the example with the corresponding growth status standard, obtain environmental parameters that meet the preset requirements, and generate a set of environmental parameters corresponding to grass species selection and grass plant management.

[0063] S306 constructs data-driven models for grass seed selection and grass plant management based on support vector machines and autoencoder networks, respectively. The corresponding data-driven models are trained using environmental parameter sets combined with orchard grass seed selection data and orchard grass management data. The mapping relationship between environmental influencing factors and grass cultivation management is constructed from two aspects.

[0064] It should be noted that multi-source environmental parameters are collected from the target orchard, including but not limited to soil conditions, climate conditions, orchard weed conditions, and parameters of the cover grass plants. Cover grass species selection and management early warning are implemented for the target orchard. Specifically, cover grass species selection involves collecting multi-source environmental parameters from the target orchard based on the corresponding environmental influencing factors. Support Vector Machine (SVM) is a supervised learning model used for classification and regression analysis, which helps find the optimal hyperplane in classification problems to help distinguish data points of different categories. The multi-source environmental parameters are used as input to a support vector machine (SVM) to generate corresponding feature vectors. The data-driven model generated by the trained SVM is used to predict and score the feature vectors. During training, a classification space is constructed using the environmental parameter set combined with orchard grass seed selection data. In the classification space, orchard grass seed selection is aggregated and located based on the corresponding grass seed labels. The gscatter function is used to obtain a scatter plot of the data corresponding to the feature vectors. The contour function is used to generate decision boundaries for different grass seed labels in the scatter plot. The evaluation scores of different decision boundaries are calculated based on the distribution of feature vectors falling within the decision areas corresponding to different grass seed labels. Based on the evaluation scores, the grass seed selection results for the target orchard are generated.

[0065] In addition, the following steps are taken to provide early warning for orchard cover management during the growth process: Based on the data recorded by the cover plants, the growth stage of the target orchard cover is determined. Based on this growth stage, corresponding environmental influencing factors are collected from the environmental influencing factors related to cover plant management, and corresponding multi-source environmental parameters are collected. These multi-source environmental parameters are then imported into an autoencoder network. The input is the measured value of the multi-source environmental parameters during the growth process of the target orchard cover, and the output is the estimated value of the multi-source environmental parameters under cultivation management. The trained autoencoder network is used to construct a data-driven model to reconstruct the multi-source environmental parameters, obtaining estimated values ​​after implementing key management measures. The multi-dimensional residuals between the estimated and measured values ​​of the multi-source environmental parameters are calculated. Residual thresholds for each key management measure's corresponding data-driven model are preset. The multi-dimensional residuals are compared with the corresponding residual thresholds. Based on the multi-dimensional residual data, it is easy to determine whether management is needed. The number of multi-dimensional residuals exceeding the residual threshold is counted. When the number reaches a preset standard, a cover management early warning is generated. Preferably, during the emergence period of the cover crop, appropriate chemical or organic fertilizers should be applied to promote rapid seedling growth. During droughts, the orchard should be irrigated promptly. The measured water and fertilizer characteristics of the target orchard soil are then imported into an autoencoder network to obtain estimated water and fertilizer characteristics after water and fertilizer management. If the estimated water and fertilizer characteristics deviate significantly from the measured values, it indicates that water and fertilizer management is needed for the cover crop in the target orchard.

[0066] It should be noted that, based on the aforementioned grass management early warning, corresponding key management measures are obtained. When multiple grass management early warnings exist, the corresponding key management measures are integrated, duplicate management steps are removed, and their order priority is set. Similarity searches are performed on historical orchard grass cultivation management examples using the aforementioned key management measures and multi-source environmental parameters. The historical orchard grass cultivation management example with the highest similarity is obtained based on the similarity ranking. Management measure parameters are extracted from the selected historical orchard grass cultivation management examples, and the environmental deviation between the target orchard and the examples is calculated based on the multi-source environmental parameters. An adjustment coefficient is generated based on the environmental deviation to adjust the management measure parameters. The adjusted management measure parameters are then used to configure the key management measures, generating the optimal grass management plan. Furthermore, the grass seed selection information, grass management early warning information, and grass management plan are output and visualized according to a preset visualization device and transmission method, ensuring that orchard managers can conduct timely orchard grass management to improve soil organic matter content, increase soil fertility, improve fruit quality, and increase fruit yield.

[0067] Figure 4 A block diagram of an orchard cover crop management system based on multi-source environmental parameters is shown.

[0068] The second aspect of the present invention provides an orchard cover cultivation management system 4 based on multi-source environmental parameters. The system includes an environmental impact factor analysis unit 401, a multi-source environmental parameter acquisition unit 402, a data-driven model unit 403, and an orchard cover cultivation management output unit 404.

[0069] The environmental impact factor analysis unit uses the acquired knowledge of orchard cover cultivation and historical examples of orchard cover cultivation management to construct full-process management data, extract key management measures for cover cultivation, and obtain the environmental impact factors of key management measures based on correlation analysis.

[0070] The multi-source environmental parameter acquisition unit extracts multi-source environmental parameters based on environmental influencing factors at different growth stages;

[0071] The data-driven model unit constructs data-driven models for grass seed selection and grass plant management based on support vector machines and autoencoder networks, respectively. It trains the corresponding data-driven models using environmental parameter sets combined with orchard grass seed selection data and orchard grass management data, constructing a mapping relationship between environmental influencing factors and grass cultivation management from two aspects. Based on the multi-source environmental parameters of the target orchard and the data-driven model, it provides early warning for grass seed selection and grass management in the target orchard.

[0072] The orchard cover crop management output unit obtains the optimal cover crop management plan based on the cover crop management early warning associated with different key management measures, outputs cover crop selection information, cover crop management early warning information and cover crop management plan, and displays them visually using visualization methods.

[0073] The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components may be combined, or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling, direct coupling, or communication connection between the various components shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms. Additionally, in the various embodiments of the present invention, all functional units may be integrated into one processing unit, or each unit may be a separate unit, or two or more units may be integrated into one unit; the integrated unit may be implemented in hardware or in a combination of hardware and software functional units.

[0074] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0075] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for orchard cover crop cultivation and management based on multi-source environmental parameters, characterized in that, Includes the following steps: Acquire full-process management data of orchard grass cultivation, extract key management measures for grass cultivation, divide the full-process management data based on the key management measures, and obtain the environmental impact factors of each key management measure based on correlation analysis; Environmental parameters are extracted based on the environmental influencing factors. Environmental parameters that meet the preset requirements are selected by using the growth status of the grass. Data-driven models corresponding to each key management measure are constructed using the environmental parameters to establish a mapping relationship between environmental influencing factors and grass cultivation management. Collect multi-source environmental parameters of the target orchard, and based on the multi-source environmental parameters and data-driven model, conduct grass seed selection and grass management early warning for the target orchard; The optimal grass management plan is obtained by associating different key management measures with grass management early warnings. The grass species selection information, grass management early warning information and grass management plan are visualized using visualization methods. Environmental parameters are extracted based on the aforementioned environmental influencing factors. Environmental parameters that meet preset requirements are then selected based on the growth status of the grass. These environmental parameters are then used to construct data-driven models corresponding to each key management measure. Specifically: Key management measures are divided according to the selection of grass species and the management of grass plants. Environmental parameters are screened based on the environmental impact factors matched by each key management measure after the division. Grass growth stages and grass growth status are obtained from the instances corresponding to the full-process management data. Growth status standards for different grass growth stages are generated based on grass growth data under suitable conditions. By comparing the grass growth status in the example with the corresponding growth status standard, environmental parameters that meet the preset requirements are obtained, and a set of environmental parameters corresponding to grass species selection and grass plant management is generated. Data-driven models for grass seed selection and grass plant management were constructed based on support vector machines and autoencoder networks, respectively. The corresponding data-driven models were trained using environmental parameter sets combined with orchard grass seed selection data and orchard grass management data. The mapping relationship between environmental influencing factors and grass cultivation management was constructed from two aspects. Multi-source environmental parameters of the target orchard are collected. Based on these multi-source environmental parameters and a data-driven model, the target orchard is provided with grass seed selection and grass management early warning. Specifically, grass seed selection involves: Based on the selection of grass species, select the corresponding environmental influencing factors and collect multi-source environmental parameters of the target orchard. Use the multi-source environmental parameters as input to the support vector machine to generate corresponding feature vectors. Use the data-driven model formed by the trained support vector machine to predict and score the feature vectors. The gscatter function is used to obtain the data scatter plot corresponding to the feature vector. The contour function is used to generate decision boundaries for different grass species labels in the scatter plot. The evaluation scores of different decision boundaries are calculated. Based on the evaluation scores, the grass species selection results for the target orchard are generated. Multi-source environmental parameters of the target orchard are collected. Based on these parameters and a data-driven model, grass species selection and grass management early warning are provided for the target orchard. Specifically, the grass management early warning includes: The growth stage of the target orchard's cover grass is obtained. Based on the growth stage, corresponding environmental influencing factors are collected from the environmental influencing factors corresponding to the cover grass plant management. The corresponding multi-source environmental parameters are then imported into the autoencoder network. The data-driven model constructed by the trained autoencoder network is used to reconstruct the multi-source environmental parameters, obtain the estimated values ​​of the multi-source environmental parameters after the implementation of key management measures, and calculate the multidimensional residuals between the estimated values ​​and the measured values ​​of the multi-source environmental parameters. The residual thresholds for the data-driven models corresponding to each key management measure are preset. The multidimensional residuals are compared with the corresponding residual thresholds. The number of multidimensional residuals exceeding the residual thresholds is counted. When the number reaches the preset standard, a grass management early warning is generated. Based on the early warning system for grass management in different key management measures, the optimal grass management plan is associated with the following: Based on the aforementioned grass management early warning, corresponding key management measures are obtained. Using the key management measures and multi-source environmental parameters, a similarity search is performed on historical orchard grass cultivation management examples. Based on the similarity ranking, the historical orchard grass cultivation management example with the highest similarity is obtained. Management parameters are extracted from historical orchard cover cropping management examples. Environmental deviations between the target orchard and the examples are calculated based on multi-source environmental parameters. Adjustment coefficients are generated based on the environmental deviations to adjust the management parameters. The adjusted management parameters are then used to configure the key management measures to generate the optimal cover cropping management plan.

2. The orchard cover crop cultivation management method based on multi-source environmental parameters according to claim 1, characterized in that, Acquire full-process management data for orchard sod cultivation, extract key management measures for sod cultivation, and divide the full-process management data based on these key management measures, specifically as follows: A dataset was constructed by using a big data retrieval engine to obtain relevant knowledge and historical examples of orchard grass cultivation management. The dataset was preprocessed and sorted according to the time sequence corresponding to the growth stage of orchard grass to obtain full-process management data of orchard grass cultivation. Clustering is performed on the full-process management data to obtain clustering results. The data volume of each cluster in the clustering results is obtained. Clusters that do not meet the preset data volume requirements are removed based on the data volume, thereby simplifying the full-process management data. In the remaining clusters, keywords are extracted from the management data samples in each cluster. Based on the keyword vector, the corresponding grass growth stage and corresponding management measures are determined. After redundancy removal, the key management measures for each grass growth stage are obtained. Based on the aforementioned key management measures, the data is located within the simplified full-process management data, and neighborhood data of the key management measures is extracted according to a preset time range.

3. The orchard cover crop cultivation management method based on multi-source environmental parameters according to claim 1, characterized in that, The environmental impact factors of each key management measure were obtained based on correlation analysis, specifically: Data segments corresponding to key management measures for orchard cover cultivation are obtained. Correlation analysis is used to perform primary attribution of key management measures in the data segments. The Pearson correlation coefficient is used to improve the maximum correlation minimum redundancy method. Environmental influencing factors that meet the preset correlation conditions are obtained through the improved maximum correlation minimum redundancy method. The environmental impact factors selected for primary attribution are obtained. Based on the environmental impact factors, environmental parameters and key management measures are extracted from the data segments to construct data samples. The SHAP interpretation model is used to perform deep attribution analysis on the data samples to generate Shapley values ​​for each environmental impact factor and construct a local interpretation matrix. For each environmental impact factor, the average Shapley value of all data samples for that environmental impact factor is calculated based on the local interpretation matrix as a global interpretation. The global interpretation is used as the importance value of the environmental impact factor. The importance value is used for ranking. Based on the ranking results, a preset number of environmental impact factors are selected and matched with corresponding key management measures.

4. A management system for orchard cover crop cultivation based on multi-source environmental parameters, characterized in that, To implement any one of the orchard cover cropping management methods based on multi-source environmental parameters as described in claims 1-3, the system includes an environmental impact factor analysis unit, a multi-source environmental parameter acquisition unit, a data-driven model unit, and an orchard cover cropping management output unit. The environmental impact factor analysis unit uses the acquired knowledge of orchard cover cultivation and historical examples of orchard cover cultivation management to construct full-process management data, extract key management measures for cover cultivation, and obtain the environmental impact factors of key management measures based on correlation analysis. The multi-source environmental parameter acquisition unit extracts multi-source environmental parameters based on environmental influencing factors at different growth stages; The data-driven model unit constructs data-driven models for grass seed selection and grass plant management based on support vector machines and autoencoder networks, respectively. It trains the corresponding data-driven models using environmental parameter sets combined with orchard grass seed selection data and orchard grass management data, constructing a mapping relationship between environmental influencing factors and grass cultivation management from two aspects. Based on the multi-source environmental parameters of the target orchard and the data-driven model, it provides early warning for grass seed selection and grass management in the target orchard. The orchard cover crop management output unit obtains the optimal cover crop management plan based on the cover crop management early warning associated with different key management measures, outputs cover crop selection information, cover crop management early warning information and cover crop management plan, and displays them visually using visualization methods.

Images