Intelligent green manure spreading control method and system for rocky desertification restoration

Abstract

The invention relates to the technical field of rocky desertification restoration, in particular to an intelligent green manure spreading control method and system for rocky desertification restoration, and the method comprises the following steps: obtaining surface survey data of a target rocky desertification restoration area, and carrying out the spatial heterogeneous analysis to obtain a micro site type; key site parameters of the micro site type are extracted, and an ecological restoration quantification target is determined; establishing a green manure variety ecological function attribute resource library to determine a candidate green manure variety set; constructing a multi-target optimization model through the sowing optimization target, and solving the multi-target optimization model to obtain an optimal mixed sowing scheme; and generating a variable sowing prescription in combination with the micro site type and the optimal mixed sowing scheme, and executing the variable sowing prescription through an unmanned aerial vehicle to realize intelligent sowing control of green manure. According to the invention, the whole process from data acquisition, decision optimization to precise execution is intelligent, and the precision and ecological benefits of rocky desertification restoration are improved.

Description

Technical Field

[0001] This invention relates to the field of rocky desertification remediation technology, and in particular to a method and system for intelligent green manure spreading and control for rocky desertification remediation. Background Technology

[0002] For the restoration of rocky desertification, existing technologies typically employ green manure application methods based on human experience or rough geographical zoning; the approximate extent of rocky desertification areas is identified through remote sensing images, and the areas are divided into major categories according to common soil types or topographic features; the selection of green manure varieties largely relies on historical experience, using single varieties or fixed mixed application formulas for uniform application.

[0003] With technological advancements, drone seeding technology has been introduced. However, its application is limited to executing pre-set flight path tasks to achieve large-scale, high-efficiency seed dispersal. Seeding decisions are based on static maps and fixed seeding amounts, lacking dynamic responses to micro-scale site conditions. Furthermore, existing methods rely heavily on satellite remote sensing or rough ground surveys for data collection, which is insufficient for identifying micro-heterogeneities such as soil patches and rock fissures in rocky desertification areas. Moreover, restoration targets are often set as uniform vegetation coverage or soil improvement indicators.

[0004] Existing technologies have several shortcomings: First, they ignore the spatial heterogeneity of rocky desertification areas, leading to seed waste in barren rocky areas and insufficient seed input in fertile patches, resulting in low restoration efficiency and high costs. Second, the limited selection of green manure varieties fails to adapt to diverse microsite types, restricting the comprehensive functioning of ecosystems and hindering sustainable restoration. Furthermore, the decision-making process relies on human experience, with mixed-sowing schemes designed based on trial and error or qualitative judgment, failing to quantify and balance sowing objectives, leading to poor community stability. Finally, the low level of intelligence in drone sowing control, with fixed flight path planning and sowing commands, prevents dynamic adjustments based on real-time environmental data, limiting sowing accuracy and increasing the risk of missed or repeated sowing in complex terrains, thus affecting the restoration effect on rocky desertification areas. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for intelligent green manure spreading control for rocky desertification remediation.

[0006] To achieve the above objectives, in a first aspect, this invention provides an intelligent green manure spreading control method for rocky desertification remediation. The method includes the following steps: acquiring surface survey data of the target rocky desertification remediation area; performing spatial heterogeneous analysis based on the surface survey data to obtain micro-site types; extracting key site parameters of the micro-site types and determining quantitative ecological restoration targets based on the key site parameters; establishing a resource library of ecological functional attributes of green manure varieties; determining a set of candidate green manure varieties by combining the key site parameters and the quantitative ecological restoration targets; constructing a multi-objective optimization model through spreading optimization objectives; solving the multi-objective optimization model based on the set of candidate green manure varieties to obtain an optimal mixed-sowing scheme; generating a variable spreading prescription by combining the micro-site types and the optimal mixed-sowing scheme; and executing the variable spreading prescription using a drone to achieve intelligent green manure spreading control. This invention, through spatial heterogeneous analysis and multi-objective optimization, allows the spreading scheme to conform to micro-site conditions and remediation targets, improving green manure utilization and remediation effects, promoting a shift from experience-driven to data-driven rocky desertification remediation, and enhancing the scientific rigor and efficiency of remediation work.

[0007] Optionally, acquiring surface survey data of the target rocky desertification remediation area and performing spatial heterogeneous analysis based on the surface survey data to obtain micro-site types includes: mounting a multispectral camera and lidar on the UAV as a surface survey device; conducting aerial surveys of the target rocky desertification remediation area using the surface survey device to collect surface survey data, including surface image data and three-dimensional terrain data; establishing a rocky desertification surface type identification model; and performing spatial heterogeneous analysis on the surface survey data to obtain the micro-site types, including soil patches, rock fissures, and shallow depressions. This invention solves the problems of time-consuming and labor-intensive traditional surveys and incomplete data, providing refined data support for subsequent remediation plan formulation, improving the targeting of rocky desertification remediation, and increasing the accuracy of green manure application scheme adaptation.

[0008] Optionally, the extraction of key site parameters for the micro-site type includes: obtaining the average soil layer thickness of the micro-site type based on the surface survey data; conducting laboratory tests on soil samples of the micro-site type to obtain soil sample test results, thereby obtaining the rocky desertification level and soil pH range; obtaining the rainy season window period of the micro-site type through historical meteorological data of the target rocky desertification remediation area; and using the average soil layer thickness, the rocky desertification level, the soil pH range, and the rainy season window period as the key site parameters. This invention provides a scientific basis for determining ecological restoration targets and selecting green manure varieties, making green manure application more aligned with regional realities, reducing resource waste caused by indiscriminate application, and improving the feasibility of application schemes.

[0009] Optionally, determining the quantitative target for ecological restoration based on the key site parameters includes: obtaining the first-season target vegetation coverage and core ecological limiting factors for the micro-site type based on the key site parameters; determining the dominant functional types of the micro-site type based on the core ecological limiting factors, including rapid coverage, deep soil stabilization, and continuous nitrogen supply; and using the first-season target vegetation coverage and the dominant functional types as the quantitative target for ecological restoration. This invention avoids the problems of unclear direction and difficulty in measuring the effects of restoration work, ensuring that green manure application is always carried out around key ecological needs, and improving the first-season vegetation coverage effect and the achievement rate of core ecological functions.

[0010] Optionally, establishing a resource library of ecological functional attributes for green manure varieties, and determining a set of candidate green manure varieties by combining the key site parameters and the quantitative goals of ecological restoration, includes: establishing a rule library in which each green manure variety is labeled with root type, nitrogen fixation attribute, life form, ecological function, and seed physical characteristics, serving as the resource library of ecological functional attributes for green manure varieties; performing initial matching of the resource library of ecological functional attributes for green manure varieties according to the key site parameters to obtain preliminary screening of green manure varieties; obtaining the functional suitability score of the preliminary screening of green manure varieties in achieving the quantitative goals of ecological restoration; and selecting the preliminary screening of green manure varieties to determine the set of candidate green manure varieties. This invention achieves standardization and precision in green manure variety selection, overcoming the limitations of traditional variety selection relying on experience, ensuring a high degree of compatibility between candidate varieties and site conditions and restoration goals, laying the foundation for subsequent mixed sowing schemes, and improving the reliability of sowing schemes.

[0011] Optionally, obtaining the functional suitability score of the initially screened green manure varieties to achieve the quantitative ecological restoration target includes: obtaining the ecological function attribute information of the initially screened green manure varieties based on the green manure variety ecological function attribute resource library; determining the evaluation indicators and evaluation weights of the initially screened green manure varieties based on the ecological function attribute information and the quantitative ecological restoration target; and obtaining the functional suitability score by weighted calculation using the evaluation indicators and the evaluation weights. This invention achieves objective quantification of suitability assessment, avoids variety selection bias caused by subjective judgment, accurately identifies the optimal initially screened varieties, and ensures the high quality of the candidate variety set.

[0012] Optionally, the step of constructing a multi-objective optimization model through sowing optimization objectives, and solving the multi-objective optimization model based on the candidate green manure variety set to obtain the optimal mixed sowing scheme, includes: maximizing functional redundancy and minimizing interspecific competition as the sowing optimization objectives; establishing a bi-objective optimization function through the sowing optimization objectives, and using the bi-objective optimization function as the multi-objective optimization model; solving the multi-objective optimization model based on the candidate green manure variety set to obtain a set of mixed sowing schemes; and selecting the optimal mixed sowing scheme from the set of mixed sowing schemes according to the ecological restoration quantification objective. This invention enables mixed sowing schemes to possess both high ecological function and stability, reduce interspecific competition and internal friction, maximize the restoration efficiency of green manure communities, and improve the long-term effectiveness of ecological restoration in rocky desertification areas.

[0013] Optionally, the step of solving the multi-objective optimization model based on the candidate green manure variety set to obtain a set of mixed-sowing schemes includes: using the green manure variety combinations and seed number ratios of the candidate green manure variety set as decision vectors to construct the solution space of the multi-objective optimization model; obtaining the functional redundancy index and inter-species competition index of any decision vector through the multi-objective optimization model based on the green manure variety ecological function attribute resource library; and obtaining the set of mixed-sowing schemes that satisfy the constraints in the solution space based on the sowing optimization objective, using the key site parameters as constraints. This invention accurately defines the solution space of the model based on decision vectors, and combined with constraints, ensures that the solution results are both optimized and practical, selecting mixed-sowing schemes that better meet actual needs and improving the stability of the restoration effect.

[0014] Optionally, the step of generating a variable-rate sowing prescription by combining the micro-site type and the optimal mixed-sowing scheme, and executing the variable-rate sowing prescription by a drone to achieve intelligent green manure sowing control, includes: acquiring the spatial distribution information of the micro-site type; generating the variable-rate sowing prescription by combining the spatial distribution information and the optimal mixed-sowing scheme; generating a planned sowing control command sequence based on the variable-rate sowing prescription and the planned flight path of the drone; adjusting the planned sowing control command sequence based on the real-time positioning information of the drone and real-time environmental data to obtain a real-time sowing control command sequence; and performing differentiated green manure variety delivery by executing the real-time sowing control command sequence by the drone to achieve intelligent green manure sowing control. This invention achieves differentiated and dynamic adaptation of sowing, improves the accuracy of drone sowing, forms a closed loop of planning, execution, and adjustment, and ensures the quality of green manure sowing and the effect of rocky desertification restoration.

[0015] Secondly, this invention provides an intelligent green manure spreading control system for rocky desertification restoration. The system executes the intelligent green manure spreading control method for rocky desertification restoration provided by this invention. The system includes input devices, output devices, a processor, and a memory, which are interconnected. The memory stores a computer program, which includes program instructions, and the processor is configured to invoke the program instructions. This invention, through the synergy of high-performance hardware, improves the intelligence and standardization of green manure spreading control, providing reliable hardware support for rocky desertification restoration work. Attached Figure Description

[0016] Figure 1 This is a flowchart of a smart green manure spreading control method for rocky desertification remediation according to an embodiment of the present invention; Figure 2 This is a framework diagram of a smart green manure spreading control system for rocky desertification remediation according to an embodiment of the present invention. Detailed Implementation

[0017] Specific embodiments of the present invention will now be described in detail. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the invention.

[0018] Throughout this specification, references to "an embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with that embodiment or example is included in at least one embodiment of the invention. Therefore, the phrases "in an embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily refer to the same embodiment or example. Furthermore, specific features, structures, or characteristics can be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. Moreover, those skilled in the art will understand that the illustrations provided herein are for illustrative purposes and are not necessarily drawn to scale.

[0019] Please see Figure 1 One embodiment of the present invention provides a smart green manure spreading control method for rocky desertification remediation, the method comprising the following steps: S1. Obtain surface survey data of the target rocky desertification remediation area, and perform spatial heterogeneous analysis based on the surface survey data to obtain micro-site types.

[0020] In this embodiment, a multi-rotor drone is used as the flight platform. At the bottom of the drone fuselage, a multispectral camera (such as one equipped with blue, green, red, red-edge and near-infrared band sensors) and a lightweight lidar are mounted on a three-axis stabilized gimbal with shock absorption function. The two are fixed in space and their relative positions are known.

[0021] The multispectral camera and lidar are connected to the UAV's flight control system via data cables, and are powered and synchronously triggered by the flight control system. The flight control system is configured to synchronously trigger the multispectral camera to perform fixed-point shooting and the lidar to perform continuous scanning according to the preset aerial survey mission, ensuring that the surface image data and 3D terrain data collected in a single flight are consistent in time and space.

[0022] Before aerial surveying, several ground control points are set up in the target rocky desertification restoration area, and the geographic coordinates of the ground control points are obtained using Real-Time Kinematic (RTK) technology. Subsequently, the flight path of the UAV is planned. The flight path design must ensure that the forward overlap rate is not less than 70% and the lateral overlap rate is not less than 60%. The flight altitude is set according to the required resolution (e.g., 80m-120m) to ensure data integrity.

[0023] During UAV aerial surveying, the onboard equipment works synchronously: the multispectral camera triggers exposure according to preset intervals or positions, capturing multispectral images of multiple spectral bands as surface image data; the lidar scans the surface at a specific frequency, recording the three-dimensional coordinates (X, Y, Z) and echo intensity of each laser point, generating high-density three-dimensional point cloud data as three-dimensional terrain data; all raw data (surface image data and three-dimensional terrain data) are stored in real time in the UAV's onboard memory or transmitted to the ground station via data link.

[0024] After the aerial survey, radiometric, geometric, and orthorectification corrections were performed on the multispectral images to generate multispectral orthophotos with accurate geographic coordinates. Noise removal and classification (separating ground and non-ground points) were performed on the 3D point cloud data to generate a Digital Elevation Model (DEM) and a Digital Surface Model (DSM). The preprocessed multispectral orthophotos, DEM, and DSM together constitute the surface survey dataset for subsequent analysis.

[0025] In this embodiment, the surface survey dataset is registered so that each spatial cell simultaneously possesses spectral features (such as reflectance of each band and calculated vegetation index), texture features (calculated through gray-level co-occurrence matrix), and topographic structure features (such as elevation, slope, aspect, topographic relief, and surface roughness calculated based on DEM). A rocky desertification surface type identification model for spatial heterogeneous analysis is constructed based on multidimensional feature vectors, satisfying the following relationship: in, For the first Micro-site type label for each spatial unit This indicates that the operation should take the parameter corresponding to the maximum value. For micro-site type tags, A collection of micro-location type labels. Represents the posterior probability. For the first Multidimensional feature vectors of spatial units This is the model parameter vector.

[0026] It should be noted that the model can employ supervised classification methods (random forest, support vector machine) or deep learning models; in random forest, the model parameter vector represents the structure of all decision trees (split features, split thresholds) and the class distribution of leaf nodes; in support vector machine, the model parameter vector represents the normal vector and bias term of the optimal splitting hyperplane; in deep learning models, the model parameter vector represents the weights and biases of all convolutional kernels in the network.

[0027] Training the rocky desertification surface type identification model requires manually labeled sample data. Ecological experts, based on high-resolution imagery and field survey experience, label training sample areas for typical micro-site types such as soil patches, rock fissures (exposed rock areas), shallow depressions (catchment areas), and vegetation remnants. The model establishes a mapping relationship from the feature space to micro-site type labels by learning the multi-dimensional feature vector combination rules of the training sample areas.

[0028] After training, the multidimensional feature vector of the target rocky desertification restoration area is input into the rocky desertification surface type identification model. Each basic analysis unit (such as a pixel or grid) is classified, and a distribution map of micro-site types is output as the classification result. The classification result is then post-processed (such as using majority filtering to remove noise patches and vectorizing the raster classification result) to analyze and extract the spatially distributed micro-site types with clear boundaries.

[0029] S2. Extract the key site parameters of the micro-site type and determine the quantitative target for ecological restoration based on the key site parameters.

[0030] Specifically, S2 includes the following steps: S21. Extract the key site parameters of the micro-site type.

[0031] In this embodiment, based on the distribution map of micro-site types, the average soil layer thickness is obtained by using DEM and DSM data acquired by lidar and a combination of indirect estimation and sampling verification.

[0032] Specifically, firstly, within each micro-site type area, several sampling points are randomly deployed using spatial analysis tools. For each sampling point, the combined height of soil and vegetation is initially estimated using the elevation difference between the DSM and DEM (i.e., DSM elevation minus DEM elevation). Subsequently, a soil drill or profile measuring ruler is used to conduct on-site measurements at the corresponding sampling points to obtain the accurate effective soil layer thickness (vertical depth from the surface to the hard bedrock). Next, under this micro-site type, a regression model (such as a linear or nonlinear model) is established between the combined height and the measured soil layer thickness. Using the regression model, the combined height of the micro-site type area is converted to estimate its soil layer thickness. Finally, the arithmetic mean of all estimated values ​​in the area is calculated to obtain the average soil layer thickness of this micro-site type.

[0033] In this embodiment, soil samples from micro-site types are subjected to laboratory testing to obtain soil sample test results, thereby obtaining the rocky desertification level and soil pH range; the process includes three steps: soil sampling, laboratory testing, and determination of key indicators. First, soil sampling is carried out for each micro-site type: using an S-shaped or quincunx pattern, 3-5 representative sampling points are set up in each micro-site type area. A stainless steel soil auger is used to collect mixed soil samples from the 0cm-20cm surface layer. Stones and roots are removed to obtain soil samples, which are then placed in clean sample bags. The coordinates of the sampling points and the micro-site type label are recorded.

[0034] Secondly, laboratory tests were conducted on the soil samples: the air-dried, ground, and sieved soil samples were divided into two portions. One portion was used to determine key diagnostic indicators such as soil organic matter content and rock exposure rate, and the organic matter content was determined using the potassium dichromate oxidation-external heating method to comprehensively determine the rocky desertification level of the micro-site type. The other portion was used to determine the soil pH value: using the potentiometric method, the soil sample was mixed with deionized water at a ratio of 1:2.5 (by weight), thoroughly stirred, and allowed to stand. The supernatant value was measured using a calibrated pH meter, and each sample was repeated three times and the average value was taken.

[0035] Finally, key indicators were determined: Based on industry standards, the organic matter content obtained from laboratory tests, combined with the previously estimated average soil layer thickness and rock exposure rate, were substituted into the rocky desertification degree classification index table to determine the rocky desertification level (e.g., potential, mild, moderate, severe) for each sampling point; the level corresponding to the majority of sampling points within the micro-site type was taken as the rocky desertification level for that type; in addition, the minimum and maximum pH values ​​of all sampling points in the same micro-site type were statistically analyzed to determine the soil pH range for that micro-site type.

[0036] In this embodiment, historical meteorological data of the target rocky desertification restoration area is obtained through meteorological stations, and the data is quality checked and processed. At the same time, the start and end criteria of the rainy season window are defined: the first day when the five-day moving average precipitation is greater than a certain threshold (e.g., 5 mm) is defined as the start of the rainy season; the first day after the start of the rainy season when the five-day moving average precipitation is less than the threshold is defined as the end of the rainy season.

[0037] Based on the start and end criteria of the rainy season window, historical meteorological data is analyzed to determine the specific start and end dates of the rainy season each year. The distribution patterns of the start and end dates of the rainy season in all years (such as the mean, median, 25th and 75th percentiles) are statistically analyzed to obtain a general rainy season window.

[0038] Different micro-site types have varying water retention capacities (e.g., shallow depressions and rock fissures have poor water retention), requiring fine-tuning: The general rainy season window should be adaptively modified based on the topographic characteristics of each micro-site type. For example, for shallow depressions (catchment areas), due to their superior water conditions, their rainy season window can be appropriately extended forward and backward from the general rainy season window; for steep slopes and rock fissures, due to their extremely poor water retention capacity, their rainy season window can be shortened to the period of concentrated heavy rainfall. Finally, a rainy season window for guiding green manure application should be determined for each micro-site type, represented as a continuous range of dates.

[0039] S22. Determine the quantitative targets for ecological restoration based on the key site parameters.

[0040] In this embodiment, the target vegetation cover rate for the first quarter is determined based on key site parameters, which are strongly correlated with the rocky desertification level and the average soil thickness. First, a lookup table for "rocky desertification level - baseline target vegetation cover rate" is pre-set based on ecological engineering experience. For example, "potential" rocky desertification corresponds to 60%-70%, "mild" to 50%-60%, "moderate" to 30%-50%, and "severe" to 20%-30%. Then, the baseline value is dynamically fine-tuned based on the average soil thickness: if the soil thickness is less than 10 cm, indicating an extremely fragile site foundation, the lower limit of the first-quarter target vegetation cover rate range for that level is taken and then lowered by 5%; if the thickness is greater than 30 cm, indicating relatively better conditions, the upper limit is taken and can be increased by 5%. Finally, a specific first-quarter target vegetation cover rate is output for each micro-site type.

[0041] In this embodiment, core ecological limiting factors are identified as the single key factors that constitute the primary limitation on plant growth. The identification is achieved through sequential logical judgment: First, the length and stability of the rainy season window are analyzed. If the window is shorter than 60 days or the precipitation is extremely unstable, then "water stress" is determined to be the core ecological limiting factor. Second, if water conditions are not the primary limitation, the average soil layer thickness is analyzed. If it is less than 15 cm, then "shallow soil" (poor water and fertilizer retention capacity) is determined to be the core ecological limiting factor. Third, if none of the above conditions are met, it means that the main limitation in this area comes from extremely barren soil, especially related to "moderate" or higher levels of rocky desertification, and "nutrient deficiency (especially nitrogen)" is determined to be the core ecological limiting factor.

[0042] Furthermore, based on core ecological limiting factors, dominant function types are determined through ecological function mapping. When the core ecological limiting factor is "water stress," the urgent task of restoration is to reduce resource loss and stabilize the site environment. Therefore, the dominant function type is determined to be "rapid cover." This type requires that the selected green manure combination must have the characteristics of rapid emergence, rapid early growth, and lush foliage, which can quickly form surface shading, thereby reducing soil moisture evaporation, inhibiting soil erosion, and creating a habitat for soil microorganisms. When the core ecological limiting factor is "shallow soil" (focusing on preventing deep soil loss or landslides) or a steep slope requiring base reinforcement, the dominant function type is determined to be "deep soil stabilization." The core requirement for this type is that the green manure variety has a well-developed taproot system or deep root system, which can penetrate into soil fissures or lower layers, mechanically anchoring the soil through a strong root network, enhancing the soil's shear resistance, and thus effectively preventing landslides and deep erosion. When the core ecological limiting factor is "nutrient deficiency (especially nitrogen)," the core task of restoration is to initiate and maintain the soil nitrogen cycle to provide basic fertility for ecosystem recovery. Therefore, the dominant functional type is determined to be "continuous nitrogen supply." This type requires that the combination of green manure varieties must have strong biological nitrogen fixation capacity and good nutrient return efficiency. Leguminous plants that can efficiently coexist with rhizobia, have a large nitrogen fixation capacity, and have high nitrogen content in their aboveground biomass are given priority in order to rapidly and continuously increase the soil nitrogen pool.

[0043] S3. Establish a resource database of ecological function attributes of green manure varieties, and determine a set of candidate green manure varieties by combining the key site parameters and the quantitative targets of ecological restoration.

[0044] In this embodiment, the green manure variety ecological function attribute resource library is a structured rule library, in which each record uniquely corresponds to a green manure variety and its multidimensional attributes are systematically labeled.

[0045] Specifically, the labeling information for each green manure variety should include at least the following five dimensions: 1) Root type: Classified labeling, such as "taproot system" (distinct taproot, deep development, beneficial for soil stabilization), "fibrous root system" (shallow and dense root system, beneficial for surface soil and water conservation), or "rhizome type" (with lateral spreading ability, beneficial for rapid mulching). 2) Nitrogen fixation attribute: Clearly indicate its nitrogen fixation ability and method, such as "leguminous symbiotic nitrogen fixation (high efficiency)", "leguminous symbiotic nitrogen fixation (medium)", "non-leguminous nitrogen fixation", or "no nitrogen fixation ability". 3) Life form: Describe its growth cycle and morphology, such as "annual herb", "biennial herb", "perennial herb", or "shrub", which relates to its growth rate and growth cycle length. 4) Ecological Functions: The core ecological functions are labeled, based on a summary of their physiological characteristics, such as "fast-growing cover," "soil improvement (providing organic matter)," "biological nitrogen fixation," "anchoring deep soil layers," "tolerance to poor soil," "drought resistance," or "acid and alkali resistance." A single green manure variety may possess multiple functional labels. 5) Seed Physical Characteristics: This includes quantitative parameters such as "thousand-seed weight (grams)," "seed diameter (millimeters)," "percentage of hard seeds (the proportion of dormant seeds requiring treatment)," and "recommended sowing depth (cm)."

[0046] It should be noted that the establishment of the green manure variety ecological function attribute resource database was completed by collecting publicly published floras, agricultural green manure cultivation manuals, scientific research literature and field test data, and inputting the information after standardizing and coding it, which constitutes the knowledge base for subsequent intelligent screening and decision-making.

[0047] In this embodiment, the initial matching is a filtering process based on hard constraints, which aims to screen varieties adapted to specific microsite types from the resource pool. For each microsite type, its key site parameters are compared with the known ecological tolerance range of each variety in the resource pool.

[0048] Specifically: First, based on the soil pH range, varieties whose suitable pH ranges marked in the resource bank overlap with the pH range of the micro-site type are selected. For example, if the soil pH range of a certain micro-site type is 5.0-6.5, green manure varieties with a tolerance range of 5.0-8.0 can be retained, while green manure varieties that only adapt to neutral soil (6.5-7.5) will be eliminated. Second, combining the rocky desertification level and average soil thickness, varieties with corresponding tolerance labels included in the ecological function tags are selected. For example, for micro-site types with "severe" rocky desertification or soil thickness less than 10 cm, only varieties labeled "tolerant to poor soil" are retained; if the soil layer is shallow, "drought tolerance" may also be required. Finally, based on the length of the rainy season window, varieties whose life forms and growth cycles match are selected. For example, for areas with a window period of less than 3 months, "annual" and "fast-growing" varieties are prioritized to ensure that they can complete the critical growth stages within the limited rainy season. Through the above multi-step sequential screening, a batch of varieties that are feasible under basic survival conditions were finally retained for this micro-site type, forming the initial screening of green manure varieties.

[0049] In this embodiment, the functional fit scoring is a quantitative comprehensive evaluation process aimed at assessing the potential contribution of each initially screened green manure variety to achieving the specific ecological restoration quantitative goals of this micro-site type, and determining the candidate green manure variety set, including the following steps: First, based on the ecological function attribute information of the green manure variety ecological function attribute resource bank, evaluation indicators and weights are determined in conjunction with the quantitative goals of ecological restoration. The core of the evaluation system revolves around how to achieve the "dominant function type" and reach the "target vegetation coverage rate in the first season": for example, if the dominant function type is "rapid coverage", the evaluation indicators are set as "early growth rate", "maximum leaf area index potential" and "seed germination rate", and these indicators are given high evaluation weights (e.g., a total of 70%); at the same time, the "expected contribution to the target vegetation coverage rate in the first season" (estimated through its single-seeding theoretical coverage model) is used as a comprehensive indicator and given the remaining evaluation weight (30%). If the dominant function type is "continuous nitrogen supply", the evaluation indicators are adjusted to "nitrogen fixation potential per unit area", "aboveground nitrogen content" and "nutrient return efficiency", and given high evaluation weights.

[0050] Secondly, a weighted calculation is performed to obtain the functional suitability score: Based on the standardized data in the "Green Manure Variety Ecological Function Attribute Resource Library", each evaluation indicator of each initially screened variety is quantitatively assigned a value (such as using a scale of 1-10 points); the functional suitability score is obtained by weighted summation according to the evaluation weight.

[0051] Finally, selection is based on functional suitability scores: all initially screened green manure varieties are sorted from highest to lowest according to their functional suitability scores. An absolute score threshold (e.g., 7.0 points) is set for further selection, and the top N (e.g., top 10) green manure varieties are selected as the candidate green manure variety set.

[0052] S4. Construct a multi-objective optimization model by optimizing the sowing objective, and solve the multi-objective optimization model based on the candidate green manure variety set to obtain the optimal mixed sowing scheme.

[0053] In this embodiment, the seeding optimization objective focuses on constructing an artificial green manure community with high resilience and low internal friction.

[0054] Maximizing functional redundancy refers to the rational allocation of multiple green manure varieties in a mixed-sowing scheme designed for a specific micro-site type, so that they have complementary capabilities in achieving core advantageous functions (such as rapid coverage, deep soil stabilization, or continuous nitrogen supply). The aim is to optimize the variety combination and ratio so that the mixed-sowing community has the strongest comprehensive potential in the target ecological function. Furthermore, when encountering local environmental stresses (such as short-term drought or pests and diseases), even if individual varieties perform poorly, other varieties with similar or complementary functions can maintain the stability of the overall community function and avoid a significant decline in the restoration effect.

[0055] Minimizing interspecific competition means that when optimizing mixed sowing schemes, it is necessary to minimize the excessive competition among different green manure varieties for limited resources such as light, water, nutrients, and space within the community. By quantitatively assessing and minimizing the niche overlap between different green manure varieties, harmonious coexistence can be promoted, and the decline in total community biomass and the exclusion of some green manure varieties due to vicious competition can be avoided, thereby ensuring the efficient use of sowing resources and the sustainable development of the community.

[0056] In this embodiment, niche overlap is a comprehensive indicator that quantifies the similarity of the ecological niches of any two green manure varieties from three dimensions: time (the proportion of overlapping phenological periods), space (differences in root depth), and nutrition (whether nitrogen fixation characteristics conflict), combined with corresponding weighting coefficients. Higher overlap indicates more intense potential competition. Niche overlap satisfies the following relationship: in, Niche overlap. This indicates two different green manure varieties. These are the weighting coefficients for the time dimension. The percentage overlap in the phenological periods of the two green manure varieties. For the spatial dimension, the weighting coefficients are... Green manure varieties Typical root depth, Green manure varieties Typical root depth, The maximum typical root depth of candidate green manure varieties. These are the weighting coefficients for the nutritional dimension. This is the competitive adjustment coefficient for nitrogen fixation properties.

[0057] It should be noted that the rule for assigning the competition adjustment coefficient is as follows: if and If both are leguminous nitrogen-fixing plants, a higher value (e.g., 0.8-1.0) should be used because they may compete for the same soil resources; if one is a leguminous plant and the other is not, a lower value (e.g., 0.2-0.4) should be used because they complement each other in nitrogen source utilization (leguminous plants utilize atmospheric nitrogen, while non-leguminous plants utilize soil nitrogen); if neither is a leguminous plant, a middle value (e.g., 0.5-0.7) should be used.

[0058] Furthermore, based on the seeding optimization objective, a formalized bi-objective optimization function is established as a multi-objective optimization model. The aim is to find mixed seeding schemes for each micro-site type.

[0059] The first objective function aims to maximize functional redundancy. It consists of two parts: the first part is a weighted sum of the functional value vectors of each variety in the mixed-sowing scheme in achieving the micro-site advantage function type, directly reflecting the core functional strength of the scheme; the second part is a penalty term used to constrain the potential of the mixed-sowing scheme to achieve the target vegetation cover rate for the first season: when the expected cover rate of the mixed-sowing scheme is lower than the set target, a penalty proportional to the square of the gap will be applied, forcing the optimization process to prioritize meeting the minimum cover rate requirement. The first objective function satisfies the following relationship: in, To maximize, As a functional redundancy indicator, Let be the decision vector. This represents the total number of mixed broadcast schemes. This serves as an index for the hybrid broadcast scheme. For the first A mixed-sowing scheme, For the first Functional value vector of the hybrid broadcasting scheme For the weight vector, The penalty coefficient is... The target vegetation coverage rate for the first quarter is... For the first The theoretical vegetation coverage of the mixed sowing scheme.

[0060] The second objective function aims to minimize interspecific competition. The overall competition intensity is quantified by calculating the niche overlap between all possible species pairs in the mixed planting scheme and then weighting and summing the results. The second objective function satisfies the following relationship: in, Indicates minimization. As an indicator of interspecific competition, Let be the decision vector. This indicates two different green manure varieties. For the first The seed count ratio of each green manure variety For the first The seed count ratio of each green manure variety Niche overlap. For the first The appropriate volume per unit area for each green manure variety For the first The appropriate capacity per unit area for each type of green manure.

[0061] In this embodiment, solving the multi-objective optimization model includes the following steps: First, for the current microsite type, select one or more varieties from its candidate green manure variety set to form a potential combination, and assign each variety in the combination a seed number ratio to be optimized (the value ranges from 0 to 1), and the sum of the ratios of all varieties is 1; the green manure variety combination and the seed number ratio constitute a decision vector, and all possible decision vectors form the solution space.

[0062] Secondly, the objective function value of each decision vector in the solution space is obtained. Based on the ecological function attribute resource library of green manure varieties, for any decision vector, the functional attributes and attributes such as phenology, root depth, and nitrogen fixation type of each variety are extracted, so as to calculate the specific functional redundancy index and interspecific competition index of the mixed sowing scheme.

[0063] Then, the solution space is screened using key site parameters as constraints. For example, the average soil thickness is converted into a limit on the maximum root depth of the varieties, eliminating mixed sowing schemes that include varieties with excessively deep roots; the soil pH range is used as a threshold to eliminate mixed sowing schemes that include varieties that are completely unsuitable for this pH range.

[0064] Finally, a multi-objective optimization algorithm is used to search within the constrained solution space. Multi-objective evolutionary algorithms such as the second-generation non-dominated sorting genetic algorithm (NSGA-II) are used to simultaneously optimize the seeding optimization objective. After the algorithm runs, it will output a Pareto optimal solution set, which is the set of mixed seeding schemes.

[0065] In this embodiment, selecting the optimal mixed-sowing scheme from the set of schemes is a decision-making process based on quantitative ecological restoration goals. The core criterion of this decision is to prioritize ensuring the achievement of the first-quarter target vegetation coverage and to enhance the achievement of advantageous functional types.

[0066] Specifically, the two objective function values ​​of each scheme in the mixed-species scheme set are normalized to make them comparable; based on the urgency and focus of the micro-site type restoration task, two methods can be used for screening: Firstly, set priority thresholds: For example, require that the target vegetation coverage of the mixed-sowing scheme in the first season must be no less than a certain percentage of the target coverage (such as 95%), and eliminate all schemes that do not meet this mandatory coverage requirement. Among the remaining schemes, if the dominant functional type is "rapid coverage," indicating that initial functional stability is crucial, then the mixed-sowing scheme with the highest functional redundancy index should be selected first; if the focus is more on building a long-term stable community, then a mixed-sowing scheme with a relatively lower interspecific competition index may be selected.

[0067] Secondly, a weighted compromise is used: based on the target preference, weights are assigned to the functional redundancy index and the interspecific competition index, and the highest mixed-species scheme is selected after calculating the comprehensive utility value.

[0068] The final selected mixed sowing scheme, with its determined combination of green manure varieties and seed ratio, is the optimal mixed sowing scheme for guiding sowing.

[0069] S5. Combine the micro-site type and the optimal mixed seeding scheme to generate a variable seeding prescription, and execute the variable seeding prescription through a drone to realize intelligent seeding control of green manure.

[0070] In this embodiment, the spatial distribution information of micro-site types is obtained. The spatial distribution information is a vector map, in which each micro-site type region is assigned a specific type code (such as "Type_A: shallow depression"). At the same time, the optimal mixed sowing scheme corresponding to each micro-site type is obtained, including the specific green manure variety composition and its precise seed number ratio. The generation of variable sowing prescription is to associate and map the spatial distribution information with the optimal mixed sowing scheme.

[0071] Specifically, firstly, the spatial connectivity function of the Geographic Information System (GIS) is used to associate and match spatial distribution information with the optimal mixed sowing scheme. Next, sowing density is calculated: the total number of seeds sown in the area is determined (e.g., 2000 seeds / m²), and the target sowing density (seeds / m²) for each green manure variety is obtained by combining the seed ratio of each variety; for example, if variety A accounts for 60% in a certain type of scheme, then the target density for variety A in this micro-site type is 1200 seeds / m². Finally, the variable sowing prescription is output, adding the target sowing density for each variety, thus forming a complete executable operation map.

[0072] In this embodiment, based on the variable sowing prescription and UAV operating parameters (such as sowing width and flight speed), a parallel flight path network covering the entire area is automatically generated in the flight planning software. One or more sowing control commands are then generated based on this network, forming a planned sowing control command sequence, which is essentially a list of commands ordered by time and flight distance. Each command includes at least the following key parameters: 1) Trigger position: the coordinates of discrete points along the UAV flight path; 2) Action command: such as "switch seed bins" or "set seeding motor speed"; 3) Action parameters: specific execution parameters calculated from the target sowing density for the corresponding target variety, such as the motor pulse frequency or gate opening.

[0073] It should be noted that for micro-site types that require mixed seeding, the instruction sequence will contain a set of continuous instructions that trigger the start-up, shutdown and parameter settings of different seed bins in sequence according to the formula ratio, so as to achieve dynamic mixed seeding in the air. The planned seeding control instruction sequence is pre-generated and loaded into the UAV flight control system before the operation as a baseline execution plan.

[0074] In this embodiment, to ensure seeding accuracy, the planned seeding control command sequence is dynamically adjusted in a closed loop based on real-time flight conditions. This is accomplished collaboratively by the UAV flight control system and the mission computer.

[0075] First, positioning calibration is performed: the drone acquires its own real-time positioning information. This real-time position is compared with the expected position along the planned flight path to calculate the lateral and longitudinal deviations. When the deviation exceeds a preset threshold (e.g., 0.5 meters), the trigger position of the command is dynamically corrected based on the real-time position.

[0076] Secondly, environmental perception is integrated for parameter compensation: real-time environmental data is obtained through airborne sensors or by receiving data from ground weather stations, mainly including real-time wind speed, wind direction, and relative altitude of the UAV; the planned seeding control command sequence is adjusted based on the real-time environmental data.

[0077] For example, crosswinds can cause seeds to drift. Based on wind speed and direction, the timing of the actual action can be advanced or delayed when the command is triggered, and reverse position compensation can be performed. Small fluctuations in flight altitude can affect the distribution range of seeds. Therefore, the parameters of the seeding actuator can be dynamically adjusted according to the altitude to maintain a constant number of seeds falling per unit area.

[0078] Finally, by combining the calibrated position and compensated parameters, the planned dissemination control command sequence is calculated in real time to generate a real-time dissemination control command sequence, which is then sent to the dissemination actuator in milliseconds.

[0079] In this embodiment, the drone integrates multiple independent seed bins with dispersing actuators (motors, seed metering devices, or solenoid valves). Each seed bin is pre-filled with a specific green manure variety and is controlled by an independent dispersing actuator.

[0080] During flight seeding, the flight control system continuously calculates and issues real-time seeding control command sequences, and the seeding actuator performs precise operations: when it receives the command "switch to seed bin number N", the corresponding motor connects the channel of the seed bin to the seeding port; at the same time, the command "set speed to Y" drives the seed metering device of the bin to work at the set speed, discharging the seeds at a precise flow rate.

[0081] In areas requiring mixed sowing, the sowing actuator cycles through different seed bins and controls them to operate at different speeds. The running time ratio of each bin strictly corresponds to the seed number ratio in the optimal mixed sowing scheme, thereby achieving nearly uniform mixed sowing in space.

[0082] Throughout the green manure application process, the working status and seed flow of each compartment are monitored in real time to ensure consistency between execution and instructions. Based on variable application prescriptions, real-time dynamic adjustments, and multi-variety collaborative operation, differentiated intelligent green manure application control for spatially heterogeneous rocky desertification areas has been achieved.

[0083] Please see Figure 2 In one optional embodiment, the present invention provides an intelligent green manure spreading control system for rocky desertification remediation. The system includes input devices, output devices, a processor, and a memory, all interconnected. The memory stores a computer program comprising program instructions. The processor is configured to invoke the program instructions to execute specific steps as described in the relevant embodiments of the intelligent green manure spreading control method for rocky desertification remediation provided by the present invention. The intelligent green manure spreading control system for rocky desertification remediation provided by the present invention has a complete and stable structure, enhancing the overall applicability and practical application capabilities of the present invention.

[0084] In summary, this invention provides an intelligent green manure spreading control method and system for rocky desertification remediation. First, surface survey data is collected using a UAV equipped with a multispectral camera and lidar. A rocky desertification surface type identification model is used to analyze micro-site types such as soil patches and rock fissures. Next, key site parameters such as average soil thickness and rocky desertification level are extracted, and quantitative ecological restoration targets are determined. Then, a resource library of ecological functional attributes of green manure varieties is established, and a set of candidate green manure varieties is selected based on key parameters and restoration targets. Subsequently, a multi-objective optimization model is constructed with the objectives of maximizing functional redundancy and minimizing interspecific competition to solve for the optimal mixed spreading scheme from the candidate set. Finally, based on the spatial distribution of micro-site types and the optimal scheme, a variable spreading prescription is generated, and the UAV executes differentiated spreading through a real-time adjusted command sequence. This invention is easy to understand, computationally simple, requires less work, and is convenient for engineering applications, providing a theoretical foundation and technical support for the further development of the rocky desertification remediation technology field.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for intelligent spreading and controlling green manure for rocky desertification remediation, characterized in that, Includes the following steps: Obtain surface survey data of the target rocky desertification remediation area, and perform spatial heterogeneity analysis based on the surface survey data to obtain micro-site types; Key site parameters of the micro-site types are extracted, and quantitative targets for ecological restoration are determined based on the key site parameters; Establish a resource database of ecological functional attributes of green manure varieties, and determine a set of candidate green manure varieties by combining the key site parameters and the quantitative targets for ecological restoration. A multi-objective optimization model is constructed by optimizing the sowing objective, and the optimal mixed sowing scheme is obtained by solving the multi-objective optimization model based on the candidate green manure variety set. By combining the micro-site type and the optimal mixed seeding scheme, a variable seeding prescription is generated. The variable seeding prescription is then executed by a drone to achieve intelligent seeding control of green manure.

2. The intelligent green manure spreading control method for rocky desertification remediation according to claim 1, characterized in that, The process involves acquiring surface survey data of the target rocky desertification remediation area, and performing spatial heterogeneous analysis based on the surface survey data to obtain micro-site types, including: The UAV is equipped with a multispectral camera and lidar as a surface surveying device. Aerial surveys of the target rocky desertification remediation area are conducted using the aforementioned surface survey equipment to collect surface survey data, including surface image data and three-dimensional topographic data. A rocky desertification surface type identification model was established, and the micro-site types were obtained by spatial heterogeneous analysis of the surface survey data, including soil patches, rock fissures and shallow depressions.

3. The intelligent green manure spreading control method for rocky desertification remediation according to claim 1, characterized in that, The extraction of key site parameters for the micro-site type includes: The average soil layer thickness of the micro-site type is obtained based on the surface survey data. Laboratory tests were conducted on soil samples from the aforementioned micro-site types to obtain soil test results, thereby determining the rocky desertification level and soil pH range. The rainy season window period of the micro-site type is obtained by using historical meteorological data of the target rocky desertification restoration area; The average soil layer thickness, the rocky desertification level, the soil pH range, and the rainy season window period are used as the key site parameters.

4. The intelligent green manure spreading control method for rocky desertification restoration according to claim 1, characterized in that, The determination of quantitative targets for ecological restoration based on the key site parameters includes: Based on the key site parameters, the first-quarter target vegetation coverage and core ecological limiting factors of the micro-site type are obtained; Based on the core ecological limiting factors, the dominant functional types of the micro-site types are determined, including rapid cover, deep soil stabilization, and continuous nitrogen supply; The target vegetation coverage rate for the first quarter and the aforementioned advantageous functional types are used as the quantitative targets for ecological restoration.

5. The intelligent green manure spreading control method for rocky desertification restoration according to claim 1, characterized in that, The establishment of a green manure variety ecological function attribute resource library, combined with the key site parameters and the ecological restoration quantitative targets, to determine the candidate green manure variety set includes: Establish a rule base for each green manure variety, which is labeled with root type, nitrogen fixation attribute, life form, ecological function and seed physical characteristics, as a resource base for the ecological function attributes of the green manure varieties; Based on the key site parameters, an initial matching of the ecological function attribute resource library of green manure varieties is performed to obtain a preliminary screening of green manure varieties. The functional fit score of the pre-screened green manure varieties to achieve the quantitative goal of ecological restoration is obtained, and the pre-screened green manure varieties are selected to determine the candidate green manure variety set.

6. The intelligent green manure spreading control method for rocky desertification remediation according to claim 5, characterized in that, The functional fit score for obtaining the initially screened green manure varieties to achieve the quantitative target of ecological restoration includes: Ecological function attribute information of the initially screened green manure varieties is obtained based on the green manure variety ecological function attribute resource library. The evaluation indicators and evaluation weights of the initially screened green manure varieties are determined based on the ecological function attribute information and the ecological restoration quantitative targets. The functional suitability score is obtained by weighting the evaluation indicators and the evaluation weights.

7. The intelligent green manure spreading control method for rocky desertification restoration according to claim 1, characterized in that, The process of constructing a multi-objective optimization model by optimizing the sowing objective, and solving the multi-objective optimization model based on the candidate green manure variety set to obtain the optimal mixed sowing scheme includes: The objectives of the seeding optimization are to maximize functional redundancy and minimize interspecific competition. A bi-objective optimization function is established based on the seeding optimization objective, and the bi-objective optimization function is used as the multi-objective optimization model. The multi-objective optimization model is solved based on the candidate green manure variety set to obtain a set of mixed sowing schemes; The optimal hybrid seeding scheme is selected from the set of hybrid seeding schemes based on the quantitative target of ecological restoration.

8. The intelligent green manure spreading control method for rocky desertification remediation according to claim 7, characterized in that, The process of solving the multi-objective optimization model based on the candidate green manure variety set to obtain a set of mixed-sowing schemes includes: The combination of green manure varieties and the seed number ratio of the candidate green manure varieties set are used as decision vectors to construct the solution space of the multi-objective optimization model. Based on the green manure variety ecological function attribute resource library, the functional redundancy index and interspecific competition index of any decision vector are obtained through the multi-objective optimization model. Using the key site parameters as constraints, the set of hybrid seeding schemes that satisfy the constraints in the solution space is obtained based on the seeding optimization objective.

9. The intelligent green manure spreading control method for rocky desertification restoration according to claim 1, characterized in that, The process of generating a variable-rate sowing prescription by combining the micro-site type and the optimal mixed-sowing scheme, and executing the variable-rate sowing prescription by a drone to achieve intelligent green manure sowing control includes: Obtain the spatial distribution information of the micro-site type, and combine the spatial distribution information with the optimal mixed seeding scheme to generate the variable seeding prescription; Based on the variable dissemination prescription, a sequence of planned dissemination control commands is generated in conjunction with the planned flight path of the UAV; Based on the real-time positioning information of the UAV, the planned dispersal control command sequence is adjusted by real-time environmental data to obtain the real-time dispersal control command sequence; The drone executes the real-time spreading control command sequence to perform differentiated application of green manure varieties, thereby achieving intelligent spreading control of green manure.

10. A smart green manure spreading control system for rocky desertification remediation, characterized in that, The system includes an input device, an output device, a processor, and a memory, which are interconnected. The memory stores a computer program, which includes program instructions. The processor is configured to invoke the program instructions to execute the intelligent green manure spreading control method for rocky desertification remediation as described in any one of claims 1-9.

Images