A method and system for AI-based water and fertilizer variable management of tangerines based on growth prediction
By constructing an AI-based variable management method for water and fertilizer in tangerines based on growth prediction, and using data on total nitrogen content in leaves and lignification rate of branches for zoned management, the water and fertilizer supply is dynamically optimized, solving the water and fertilizer management problem during the ripening period of autumn shoots in tangerines, and improving the quality of tangerine fruits and the efficiency of resource utilization.
Patent Information
- Application Number
- CN202611233502.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-14
- Publication Date
- 2026-11-10
AI Technical Summary
Traditional water and fertilizer management models make it difficult to quantify the spatiotemporal heterogeneity of the autumn shoot maturation process of mandarin oranges, resulting in a disconnect between the timing of nitrogen fertilizer application and demand, making it difficult to predict nitrogen transport dynamics, and lacking a real-time feedback mechanism for water and fertilizer supply, which affects flower bud differentiation and fruit quality.
The AI-based water and fertilizer variable management method for mandarin oranges, based on growth prediction, constructs an evaluation matrix of maturation process by acquiring data on total nitrogen content in leaves and lignification rate of branches. It then combines density clustering algorithm partitioning to build a nitrogen transport prediction model, monitors soluble sugars in branches and flower bud differentiation markers in real time, and dynamically optimizes water and fertilizer supply programs.
It achieves precise matching of water and fertilizer supply with the maturation process of autumn shoots, reduces nitrogen leaching loss, and improves the water and fertilizer utilization efficiency and fruit quality of mandarins.
Smart Images

Figure CN123070566A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of smart agriculture technology, and in particular to an AI-based water and fertilizer variable management method and system for tangerines based on growth prediction. Background Technology
[0002] Mandarin oranges are a major citrus variety cultivated in southern my country, and the quality of their autumn shoots directly determines flower bud differentiation, yield, and quality. Traditional water and fertilizer management often adopts an "experience-based fertilization + uniform irrigation" model, which has significant limitations: First, the maturation process of autumn shoots is significantly heterogeneous in time and space due to the influence of tree nutrition and microclimate. Existing methods cannot quantify the differences in maturation delays in different regions, often leading to a disconnect between the timing of nitrogen fertilizer application and the actual needs of autumn shoots. Second, nitrogen transport is affected by multiple factors such as soil texture and root distribution. Traditional static fertilization models cannot predict the dynamics of ammonium nitrogen leaching, easily causing nitrogen loss and environmental pollution. Third, there is a lack of a real-time feedback mechanism based on physiological responses after water and fertilizer supply, making it difficult to dynamically adjust the program according to the carbon and nitrogen metabolism status of branches. This easily leads to nitrogen excess during flower bud differentiation, inhibiting the accumulation of soluble sugars and the expression of flower bud differentiation markers (such as the ABA / GA ratio), thus reducing the flowering rate.
[0003] In recent years, while smart agriculture technologies have driven the development of variable-rate fertilization, a dedicated predictive model for the maturation period of autumn shoots in tangerines remains lacking. Existing technologies largely rely on single leaf nutrient indicators, failing to integrate key growth parameters such as branch lignification rate. Furthermore, clustering and partitioning algorithms do not incorporate spatial coordinates and maturation delay indices, making it difficult to accurately identify heterogeneity within tangerine orchards. In addition, nitrogen transport predictions often employ static equilibrium methods, neglecting the dynamic impact of autumn shoot maturation on nitrogen absorption, resulting in insufficient precision in water and fertilizer synergistic regulation. Therefore, there is an urgent need for an AI-based water and fertilizer management method that integrates growth prediction, spatiotemporal partitioning, and dynamic feedback to achieve precise water and fertilizer regulation during the autumn shoot maturation period of tangerines, thereby improving fruit quality and resource utilization efficiency. Summary of the Invention
[0004] To address at least one of the aforementioned technical problems, this invention proposes an AI-based water and fertilizer variable management method and system for tangerines based on growth prediction.
[0005] The first aspect of this invention provides an AI-based water and fertilizer variable management method for tangerines based on growth prediction, comprising: Data on total nitrogen content in leaves and lignification rate of branches in each sampling unit within the mandarin orange orchard during the autumn shoot maturation period were obtained. An evaluation matrix for the autumn shoot maturation process was constructed based on the total nitrogen content in leaves and lignification rate of branches. The autumn shoot maturation delay index for each sampling unit was determined based on the evaluation matrix. Based on the density clustering algorithm and the autumn shoot maturation delay index, the sampling units of the mandarin orange orchard are partitioned to identify the spatiotemporal heterogeneity characteristics of the autumn shoot maturation process in the mandarin orange orchard, and to obtain the autumn shoot maturation partition map of the mandarin orange orchard. Based on the autumn shoot maturation zoning map and autumn shoot maturation delay index of the mandarin orange orchard, the nitrogen supply threshold for different zoning zones was determined. A nitrogen transport prediction model was constructed by combining the ammonium nitrogen leaching rate data of the root zone soil. The nitrogen transport prediction results were determined based on the nitrogen transport prediction model. Based on the nitrogen transport prediction results, water and fertilizer supply parameters for different zones were determined, and a water and fertilizer supply scheme for the mandarin orange orchard was constructed. Real-time monitoring of the change rate of soluble sugar content in branches and the content of flower bud differentiation markers in each zone after implementing the water and fertilizer supply plan within a preset time period; calculation of nitrogen supply deviation coefficient based on the change rate of soluble sugar content and the content of flower bud differentiation markers. Based on the nitrogen supply deviation coefficient, the water and fertilizer supply scheme is optimized to obtain the water and fertilizer variable management strategy for mandarin oranges.
[0006] In this scheme, the acquisition of leaf total nitrogen content data and branch lignification rate data of each sampling unit in the mandarin orange orchard during the autumn shoot maturation period, the construction of an autumn shoot maturation process evaluation matrix based on the leaf total nitrogen content data and branch lignification rate data, and the determination of the autumn shoot maturation delay index of each sampling unit based on the evaluation matrix are as follows: Based on the geographical location information of each sampling unit in the mandarin orange orchard, mature leaves from the middle of the current year's autumn shoots in each sampling unit were collected at a fixed cycle during the autumn shoot maturation period. The collected leaf samples were digested using the Kjeldahl method, and the nitrogen concentration in the digestion solution was measured by a spectrophotometer to obtain the total nitrogen content data of the leaves in each sampling unit. In each sampling unit, a preset number of current-year autumn shoots are marked, and the length of bark browning of the marked branches is measured according to a preset sampling period. The percentage of bark browning length to the total length of the marked branches is used as the branch lignification rate data. The leaf total nitrogen content data and branch lignification rate data of the same sampling unit are associated and aligned according to the sampling unit identifier, and a two-dimensional matrix is constructed with the sampling unit as the row vector and the leaf total nitrogen content and branch lignification rate as the column vector. The growth of the autumn shoots of tangerines is predicted based on the two-dimensional matrix, and the leaf total nitrogen content and branch lignification rate in the future preset time period are determined to construct an evaluation matrix for the maturation process of autumn shoots. The degree of growth and maturity of autumn shoots within a preset time period is determined based on the autumn shoot maturity process evaluation matrix. The absolute value of the difference between the degree of growth and maturity and the preset autumn shoot maturity benchmark reference value is used as the autumn shoot maturity delay index for each sampling unit.
[0007] In this scheme, the sampling units of the mandarin orange orchard are partitioned based on the density clustering algorithm and the autumn shoot maturation delay index to identify the spatiotemporal heterogeneity characteristics of the autumn shoot maturation process within the mandarin orange orchard, thereby obtaining a partition map of the autumn shoot maturation process in the mandarin orange orchard. Specifically: Using the geographical coordinates of each sampling unit as the spatial dimension input and the autumn shoot ripening delay index corresponding to each sampling unit as the attribute dimension input, a multidimensional dataset containing spatial coordinates and attribute values is constructed. The neighborhood radius parameter and the minimum number of sampling units in the neighborhood parameter are set for density clustering. Traverse each sampling unit in the multidimensional dataset, delineate a spherical neighborhood with the spatial coordinates of the currently traversed sampling unit as the center and the neighborhood radius as the radius, count the total number of sampling units in the spherical neighborhood, and when the total number of sampling units is not less than the minimum number of sampling units in the neighborhood, mark the currently traversed sampling unit as a density-reachable core point, and classify all sampling units in the spherical neighborhood into the same candidate cluster set; An iterative expansion operation is performed sequentially on the non-core point sampling units in each candidate cluster set. A new spherical neighborhood is defined with the spatial coordinates of the non-core point sampling unit as the center and the neighborhood radius as the radius. The total number of sampling units in the new spherical neighborhood is counted. When the total number of sampling units in the new spherical neighborhood is not less than the minimum number of sampling units in the neighborhood, the non-core sampling unit is marked as a core point, and all sampling units in the new spherical neighborhood of the core point that are not included in the cluster set are included in the current candidate cluster set. A unique partition identifier is marked for all candidate cluster sets that have been expanded. The spatial coordinate boundaries of all sampling units in each candidate cluster set are extracted. The polygonal spatial contours of each partition are fitted according to the spatial coordinate boundaries. The polygonal spatial contours of each partition are correlated and mapped with the corresponding partition identifier, the mean and variance of the autumn shoot maturation delay index of all sampling units in the partition, and a partition map of autumn shoot maturation in the mandarin orange orchard is constructed.
[0008] In this scheme, the nitrogen transport supply threshold for different zones is determined based on the autumn shoot maturation zoning map and the autumn shoot maturation delay index of the mandarin orange orchard. A nitrogen transport prediction model is constructed by combining this with data on the ammonium nitrogen leaching rate in the root zone soil. The nitrogen transport prediction result is then determined based on this model. Specifically: The mean value of the autumn shoot maturation delay index of each zone is extracted from the autumn shoot maturation zone map of the mandarin orange orchard. The deviation of the mean value from the autumn shoot maturation benchmark reference value is calculated. The weighted calculation is performed in combination with the spatial distribution density of the sampling units in the zone to obtain the nitrogen shift supply threshold for different zones. The nitrogen shift supply threshold is used as the upper limit constraint condition for nitrogen application. Acquire ammonium nitrogen leaching rate data in the root zone soil, divide the root zone soil into several discrete spatial grids, use the grid center coordinates as location identifiers, associate the ammonium nitrogen leaching rate data with the grid location, and construct an initial matrix for the spatial distribution of ammonium nitrogen. Using the autumn shoot maturity delay index as the state transition weight, the ammonium nitrogen concentration change process in each grid is defined as the state transition process of the Markov chain. The time step of the state transition is set as the minimum time unit of the future preset time period, and a state transition probability matrix is constructed. The initial matrix of the spatial distribution of ammonium nitrogen is used as the current state vector of the Markov chain. It is substituted into the state transition probability matrix for iterative calculation. Each iteration corresponds to a nitrogen transport simulation of a time step. During the iteration process, the nitrogen shift supply threshold is used as the cutoff boundary of nitrogen input for each grid. When the cumulative nitrogen input of a grid reaches the threshold, the nitrogen increment calculation of that grid is stopped. After a number of iterations matching the preset time period, the ammonium nitrogen concentration value and location coordinates of each grid at different time nodes are output. The trajectory of the change of ammonium nitrogen concentration over time and the spatial migration path of all grids in the same partition are integrated to obtain the nitrogen transport prediction results, which include the transport location, concentration distribution and diffusion boundary information of nitrogen in the root zone soil at each time point.
[0009] In this scheme, the step of determining the water and fertilizer supply parameters for different zones based on the nitrogen transport prediction results and constructing a water and fertilizer supply scheme for the mandarin orange orchard specifically involves: The average concentration of ammonium nitrogen in all grids within the same partition is calculated based on the nitrogen transport prediction results. The difference between the average concentration and the nitrogen shift supply threshold of the corresponding partition is calculated to obtain the nitrogen deficit of each partition at the current time node. The amount of nitrogen replenishment per session in each partition of the mandarin orange orchard is determined based on the nitrogen deficit. Based on the spatial gradient direction of ammonium nitrogen concentration and the migration trajectory of diffusion boundary in each zone, the nitrogen concentration difference between adjacent zones is determined. The nitrogen lateral migration flux is obtained by dividing the nitrogen concentration difference by the zone spacing. The nitrogen lateral migration flux is vector-superimposed with the ammonium nitrogen leaching rate data in the vertical direction to obtain the three-dimensional transport vector of nitrogen in the root zone soil. The nitrogen migration direction is determined based on the three-dimensional migration vector. When the nitrogen lateral migration flux of a certain partition points to the adjacent partition, the nitrogen single replenishment amount of the current partition is proportionally distributed to the adjacent partition according to the flux ratio to obtain the updated nitrogen single replenishment amount of each partition. Based on the autumn shoot maturation delay index of each region, the ratio of the autumn shoot maturation delay index to the baseline reference value is calculated. This ratio is used as a time decay factor to scale the amount of nitrogen replenishment per update in the time dimension, so as to obtain the amount of nitrogen replenishment per time in each region at different time nodes. Based on the ammonium nitrogen leaching rate data of each zone, the ratio of the ammonium nitrogen leaching rate to the corresponding nitrogen time-sharing replenishment amount of the zone is calculated to obtain the water-nitrogen coupling ratio of each zone. The water-nitrogen coupling ratio is used as the conversion benchmark for irrigation water consumption, and the irrigation water consumption matching the nitrogen time-sharing replenishment amount of each zone is calculated. By linking and integrating the nitrogen time-sharing replenishment amount, irrigation water volume, replenishment time node and spatial coordinates of each zone, a set of water and fertilizer variable supply parameters for each zone in the mandarin orange orchard is obtained. Based on the set of water and fertilizer variable supply parameters, a zoned water and fertilizer supply scheme for the mandarin orange orchard is constructed.
[0010] In this scheme, the real-time monitoring of the change rate of soluble sugar content in branches and the content of flower bud differentiation markers in each zone after implementing the water and fertilizer supply scheme within a preset time period, and the calculation of nitrogen supply deviation coefficient based on the change rate of soluble sugar content and the content of flower bud differentiation markers, specifically: During the implementation of the water and fertilizer supply plan, the phloem tissue of the middle section of the marked autumn shoots in each zone was collected at fixed time intervals. The anthrone colorimetric method was used to extract sugar components and perform colorimetric reactions on the collected phloem tissue samples. The absorbance value was measured and converted to obtain the soluble sugar content of the branches in each zone. The rate of change of soluble sugar content in branches of each zone was determined based on the soluble sugar content at each collection time. Terminal bud tissues of the same autumn shoots in each zone were collected, and the mass concentration ratio of abscisic acid to gibberellin in the terminal bud tissues was detected by enzyme-linked immunosorbent assay (ELISA). The mass concentration ratio was used as the content of flower bud differentiation markers. The actual nitrogen absorption of autumn shoots of mandarin oranges in each zone was determined based on the soluble sugar content and flower bud differentiation marker content of the branches. The nitrogen supply deviation coefficient was determined by comparing the actual nitrogen absorption with the amount of nitrogen replenished in a single cycle for each zone.
[0011] In this scheme, the optimization of the water and fertilizer supply scheme based on the nitrogen supply deviation coefficient yields a water and fertilizer variable management strategy for tangerines, specifically as follows: The tangerine orchards with nitrogen supply deviation coefficients greater than the preset value are marked as areas to be optimized. The ratio of the actual nitrogen absorption of autumn shoots in the areas to be optimized to the single replenishment of nitrogen in the corresponding areas is used as the nitrogen absorption benefit value. A linear correction function is constructed with deviation correction factor as weight and nitrogen absorption benefit value as input variable to calculate the nitrogen replenishment correction amount of each area to be optimized at the current time node. The nitrogen time-sharing supply amount of each partition to be optimized is corrected on a time-by-time basis according to the nitrogen supply correction amount. The ratio of the corrected nitrogen time-sharing supply amount to the ammonium nitrogen leaching rate data of the corresponding partition is calculated again to update the water-nitrogen coupling ratio of each partition to be optimized. The irrigation water consumption of each partition to be optimized is re-determined based on the updated water-nitrogen coupling ratio. Based on the corrected nitrogen time-sharing replenishment amount, updated irrigation water volume, and replenishment time nodes for each zone to be optimized, the water and fertilizer supply plan is optimized to obtain the water and fertilizer variable management strategy for mandarin oranges.
[0012] A second aspect of the present invention also provides a growth prediction-based AI water and fertilizer variable management system for tangerines. The system includes a memory and a processor. The memory includes a growth prediction-based AI water and fertilizer variable management method program for tangerines. When the growth prediction-based AI water and fertilizer variable management method program for tangerines is executed by the processor, it implements the steps of the growth prediction-based AI water and fertilizer variable management method for tangerines as described in any of the above claims.
[0013] This invention discloses an AI-based water and fertilizer variable management method and system for mandarin oranges based on growth prediction. The method acquires data on total nitrogen content in leaves and lignification rate of branches during the autumn shoot maturation period to construct an evaluation matrix for the autumn shoot maturation process and calculates a maturation delay index. It generates a mandarin orange orchard zoning map based on a density clustering algorithm combined with the delay index to identify spatiotemporal heterogeneity. A nitrogen transport prediction model is constructed by integrating the nitrogen shift supply threshold and the soil ammonium nitrogen leaching rate to determine water and fertilizer variable supply parameters and generate supply plans. The method also monitors the change rate of soluble sugars in branches and the content of flower bud differentiation markers in real time, calculates the nitrogen supply deviation coefficient, and dynamically optimizes the plan. This invention achieves precise matching of water and fertilizer supply with the autumn shoot maturation process, effectively promoting flower bud differentiation, reducing nitrogen leaching loss, and improving water and fertilizer utilization efficiency and fruit quality in mandarin oranges. Attached Figure Description
[0014] Figure 1 A flowchart of an AI-based water and fertilizer variable management method for tangerines based on growth prediction is shown in this invention. Figure 2 The flowchart illustrating the present invention for determining the autumn shoot maturity delay index for each sampling unit is shown. Figure 3 A flowchart illustrating the calculation of the nitrogen supply deviation coefficient according to the present invention is shown; Figure 4 The diagram shows a block diagram of an AI-based water and fertilizer variable management system for tangerines based on growth prediction, according to the present invention. Detailed Implementation
[0015] 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.
[0016] 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.
[0017] Figure 1 The flowchart of an AI-based water and fertilizer variable management method for tangerines based on growth prediction is shown in this invention.
[0018] like Figure 1 As shown, the first aspect of this invention provides an AI-based water and fertilizer variable management method for tangerines based on growth prediction, comprising: Data on total nitrogen content in leaves and lignification rate of branches in each sampling unit within the mandarin orange orchard during the autumn shoot maturation period were obtained. An evaluation matrix for the autumn shoot maturation process was constructed based on the total nitrogen content in leaves and lignification rate of branches. The autumn shoot maturation delay index for each sampling unit was determined based on the evaluation matrix. Based on the density clustering algorithm and the autumn shoot maturation delay index, the sampling units of the mandarin orange orchard are partitioned to identify the spatiotemporal heterogeneity characteristics of the autumn shoot maturation process in the mandarin orange orchard, and to obtain the autumn shoot maturation partition map of the mandarin orange orchard. Based on the autumn shoot maturation zoning map and autumn shoot maturation delay index of the mandarin orange orchard, the nitrogen supply threshold for different zoning zones was determined. A nitrogen transport prediction model was constructed by combining the ammonium nitrogen leaching rate data of the root zone soil. The nitrogen transport prediction results were determined based on the nitrogen transport prediction model. Based on the nitrogen transport prediction results, water and fertilizer supply parameters for different zones were determined, and a water and fertilizer supply scheme for the mandarin orange orchard was constructed. Real-time monitoring of the change rate of soluble sugar content in branches and the content of flower bud differentiation markers in each zone after implementing the water and fertilizer supply plan within a preset time period; calculation of nitrogen supply deviation coefficient based on the change rate of soluble sugar content and the content of flower bud differentiation markers. Based on the nitrogen supply deviation coefficient, the water and fertilizer supply scheme is optimized to obtain the water and fertilizer variable management strategy for mandarin oranges.
[0019] It should be noted that by constructing an evaluation matrix by integrating two indicators—total nitrogen in leaves and lignification rate in branches—the autumn shoot maturation delay index was accurately quantified, solving the error problem of judging the growth process solely based on experience. Density clustering coupled with spatial coordinates and the delay index was used to achieve heterogeneous zoning of mandarin orange orchards, overcoming the spatial blind spots of traditional uniform management. A nitrogen supply delay threshold constraint was introduced, and a transport prediction model was constructed by combining it with ammonium nitrogen leaching rate, achieving dynamic and accurate simulation of nitrogen in the root zone while avoiding the risk of nitrogen leaching. Based on the predicted water and fertilizer variable parameters, spatiotemporal matching of irrigation and nitrogen supply was achieved. By real-time monitoring of the change rate of soluble sugars in branches and the content of flower bud differentiation markers, a closed-loop feedback mechanism for water and fertilizer supply and the physiological response of carbon and nitrogen metabolism was established. Finally, the supply plan was dynamically optimized based on the deviation, significantly improving the quality of flower bud differentiation and water and fertilizer utilization efficiency.
[0020] Figure 2 The flowchart illustrating the present invention for determining the autumn shoot maturity delay index for each sampling unit is shown.
[0021] According to an embodiment of the present invention, the step of obtaining leaf total nitrogen content data and branch lignification rate data of each sampling unit in the mandarin orange orchard during the autumn shoot maturation period, constructing an autumn shoot maturation process evaluation matrix based on the leaf total nitrogen content data and branch lignification rate data, and determining the autumn shoot maturation delay index of each sampling unit based on the evaluation matrix, specifically involves: Based on the geographical location information of each sampling unit in the mandarin orange orchard, mature leaves from the middle of the current year's autumn shoots in each sampling unit were collected at a fixed cycle during the autumn shoot maturation period. The collected leaf samples were digested using the Kjeldahl method, and the nitrogen concentration in the digestion solution was measured by a spectrophotometer to obtain the total nitrogen content data of the leaves in each sampling unit. In each sampling unit, a preset number of current-year autumn shoots are marked, and the length of bark browning of the marked branches is measured according to a preset sampling period. The percentage of bark browning length to the total length of the marked branches is used as the branch lignification rate data. The leaf total nitrogen content data and branch lignification rate data of the same sampling unit are associated and aligned according to the sampling unit identifier, and a two-dimensional matrix is constructed with the sampling unit as the row vector and the leaf total nitrogen content and branch lignification rate as the column vector. The growth of the autumn shoots of tangerines is predicted based on the two-dimensional matrix, and the leaf total nitrogen content and branch lignification rate in the future preset time period are determined to construct an evaluation matrix for the maturation process of autumn shoots. The degree of growth and maturity of autumn shoots within a preset time period is determined based on the autumn shoot maturity process evaluation matrix. The absolute value of the difference between the degree of growth and maturity and the preset autumn shoot maturity benchmark reference value is used as the autumn shoot maturity delay index for each sampling unit.
[0022] It should be noted that autumn shoots are the new shoots that emerge from the tangerine tree in autumn. Their growth status directly affects the quality of flower buds and fruiting branches, and is a crucial foundation for determining the subsequent flowering and fruiting capacity of the tangerine. If the maturation of autumn shoots is delayed, it will lead to insufficient nutrient accumulation in the branches, affecting the number and quality of flower buds and the yield in the following year. Therefore, accurately judging the maturation process of autumn shoots is an important basis for formulating water and fertilizer management plans. The growth status of autumn shoots is comprehensively evaluated by two indicators: total nitrogen content in leaves and lignification rate in branches. Total nitrogen content in leaves reflects the level of nitrogen absorption and nutrient accumulation during the formation of autumn shoots, while lignification rate in branches reflects the maturation and aging process of branch tissues. These two indicators can respectively characterize the current growth stage of autumn shoots from the perspectives of nutrient supply status and tissue development status. Since there is a certain stage correspondence between the changes in leaf nitrogen content and the changes in branch lignification rate during the normal maturation process of autumn shoots, a two-dimensional matrix is constructed based on the dual-indicator data obtained from different sampling periods. By analyzing the future trends of these two indicators through growth prediction, the degree of maturation of autumn shoots within a preset time period can be obtained. By comparing the predicted growth and maturity level with the target maturity state, the deviation between the current sampling unit's autumn shoot development process and the normal maturity requirements can be determined, and the autumn shoot maturity delay index can be quantified. This index allows for the early identification of areas with insufficient autumn shoot maturity in different regions, enabling adjustments to nitrogen supply and irrigation based on future growth needs in different areas. This achieves precise matching of water and fertilizer supply with the autumn shoot maturity process, improving flower bud differentiation quality and water and fertilizer utilization efficiency. The growth prediction is based on a two-dimensional feature matrix constructed from leaf total nitrogen content data and branch lignification rate data. The dual-indicator change sequences within different sampling periods of each sampling unit are used as model input. A long short-term memory neural network algorithm is employed to construct an autumn shoot growth prediction model. This model learns the temporal relationship between leaf total nitrogen content and branch lignification rate within historical sampling periods, extracting dynamic changes in autumn shoot nutrient accumulation and tissue maturation, and establishing a mapping relationship between input indicators and future growth states. The leaf total nitrogen content and branch lignification rate data obtained in the current sampling period are input into the trained growth prediction model, and the predicted values of leaf total nitrogen content and branch lignification rate for each sampling unit in the future preset time period are output. Based on the predicted values, the future growth and maturity of autumn shoots are calculated, and an evaluation matrix for the maturity process of autumn shoots is constructed to realize dynamic prediction of the maturity status of autumn shoots in different sampling units.
[0023] According to an embodiment of the present invention, the step of partitioning the sampling units of the mandarin orange orchard based on the density clustering algorithm and the autumn shoot maturation delay index to identify the spatiotemporal heterogeneity characteristics of the autumn shoot maturation process within the mandarin orange orchard, and obtaining a partition map of the autumn shoot maturation process in the mandarin orange orchard, specifically involves: Using the geographical coordinates of each sampling unit as the spatial dimension input and the autumn shoot ripening delay index corresponding to each sampling unit as the attribute dimension input, a multidimensional dataset containing spatial coordinates and attribute values is constructed. The neighborhood radius parameter and the minimum number of sampling units in the neighborhood parameter are set for density clustering. Traverse each sampling unit in the multidimensional dataset, delineate a spherical neighborhood with the spatial coordinates of the currently traversed sampling unit as the center and the neighborhood radius as the radius, count the total number of sampling units in the spherical neighborhood, and when the total number of sampling units is not less than the minimum number of sampling units in the neighborhood, mark the currently traversed sampling unit as a density-reachable core point, and classify all sampling units in the spherical neighborhood into the same candidate cluster set; An iterative expansion operation is performed sequentially on the non-core point sampling units in each candidate cluster set. A new spherical neighborhood is defined with the spatial coordinates of the non-core point sampling unit as the center and the neighborhood radius as the radius. The total number of sampling units in the new spherical neighborhood is counted. When the total number of sampling units in the new spherical neighborhood is not less than the minimum number of sampling units in the neighborhood, the non-core sampling unit is marked as a core point, and all sampling units in the new spherical neighborhood of the core point that are not included in the cluster set are included in the current candidate cluster set. A unique partition identifier is marked for all candidate cluster sets that have been expanded. The spatial coordinate boundaries of all sampling units in each candidate cluster set are extracted. The polygonal spatial contours of each partition are fitted according to the spatial coordinate boundaries. The polygonal spatial contours of each partition are correlated and mapped with the corresponding partition identifier, the mean and variance of the autumn shoot maturation delay index of all sampling units in the partition, and a partition map of autumn shoot maturation in the mandarin orange orchard is constructed.
[0024] It should be noted that due to differences in soil fertility, topography, light conditions, microclimate, and root distribution within different areas of the mandarin orange orchard, the growth and development of autumn shoots exhibit significant spatial variations. The same management measures are insufficient to meet the maturation needs of autumn shoots in different areas, easily leading to insufficient maturation or excessive nitrogen supply in some areas. Therefore, this study identifies the spatiotemporal heterogeneity of autumn shoot maturation within the mandarin orange orchard based on the autumn shoot maturation delay index. A density clustering algorithm is employed, using the geographical coordinates of the sampling units as spatial constraints and the autumn shoot maturation delay index as a growth state attribute. By analyzing the correlation between sampling units that are spatially close and have similar growth states, the algorithm automatically identifies and divides areas with different autumn shoot maturation states within the mandarin orange orchard. Compared to traditional methods that rely on manual experience for area division, density clustering avoids subjective errors caused by manually setting fixed partition boundaries and can identify abnormal areas of autumn shoot growth with different shapes and scales, accurately reflecting the spatial distribution patterns of autumn shoot maturation within the mandarin orange orchard. By constructing a regional map of mature autumn shoots in tangerine orchards, the growth status of autumn shoots in different areas was visualized, providing a spatial basis for determining nitrogen supply thresholds, water and fertilizer variable parameters, and precision management strategies for different regions, thereby improving the targeting of water and fertilizer regulation and the efficiency of resource utilization.
[0025] According to an embodiment of the present invention, the step of determining the nitrogen transport threshold for different zones based on the autumn shoot maturation zoning map and the autumn shoot maturation delay index of the mandarin orange orchard, constructing a nitrogen transport prediction model by combining the ammonium nitrogen leaching rate data of the root zone soil, and determining the nitrogen transport prediction result based on the nitrogen transport prediction model are specifically as follows: The mean value of the autumn shoot maturation delay index of each zone is extracted from the autumn shoot maturation zone map of the mandarin orange orchard. The deviation of the mean value from the autumn shoot maturation benchmark reference value is calculated. The weighted calculation is performed in combination with the spatial distribution density of the sampling units in the zone to obtain the nitrogen shift supply threshold for different zones. The nitrogen shift supply threshold is used as the upper limit constraint condition for nitrogen application. It should be noted that the nitrogen supply threshold refers to the maximum suitable nitrogen supply determined based on the maturity status of autumn shoots and the soil nitrogen migration capacity in different zones during the late stage of autumn shoot maturation in tangerine orchards. This is to meet the nitrogen supply needs of autumn shoots during their transition from vegetative growth to maturity, while avoiding excessive nitrogen residue, leaching loss, and continued promotion of vegetative growth. Because the degree of autumn shoot maturation delay varies in different regions, the nitrogen requirements for completing tissue maturation and forming fruiting branches for the following year differ. If the same nitrogen supply standard is applied to all regions, areas with delayed maturation may fail to reach the target maturity state due to insufficient nitrogen, while normally matured areas may experience excessive nitrogen supply, leading to continued vigorous shoot growth and affecting flower bud differentiation. Therefore, it is necessary to extract the autumn shoot maturation delay index for different zones based on the autumn shoot maturation zone map of tangerine orchards, and combine this with the spatial distribution characteristics of sampling units within each zone to determine the corresponding nitrogen supply threshold for each zone. This threshold serves as an upper limit constraint for subsequent nitrogen input, ensuring that the nitrogen supply matches the autumn shoot maturation needs of different regions.
[0026] Acquire ammonium nitrogen leaching rate data in the root zone soil, divide the root zone soil into several discrete spatial grids, use the grid center coordinates as location identifiers, associate the ammonium nitrogen leaching rate data with the grid location, and construct an initial matrix for the spatial distribution of ammonium nitrogen. Using the autumn shoot maturity delay index as the state transition weight, the ammonium nitrogen concentration change process in each grid is defined as the state transition process of the Markov chain. The time step of the state transition is set as the minimum time unit of the future preset time period, and a state transition probability matrix is constructed. The state transition probability is jointly determined by the ammonium nitrogen concentration of the current grid, the difference of ammonium nitrogen concentration between adjacent grids, and the ammonium nitrogen leaching rate. The initial matrix of the spatial distribution of ammonium nitrogen is used as the current state vector of the Markov chain. It is substituted into the state transition probability matrix for iterative calculation. Each iteration corresponds to a nitrogen transport simulation of a time step. During the iteration process, the nitrogen shift supply threshold is used as the cutoff boundary of nitrogen input for each grid. When the cumulative nitrogen input of a grid reaches the threshold, the nitrogen increment calculation of that grid is stopped. After a number of iterations matching the preset time period, the ammonium nitrogen concentration value and location coordinates of each grid at different time nodes are output. The trajectory of the change of ammonium nitrogen concentration over time and the spatial migration path of all grids in the same partition are integrated to obtain the nitrogen transport prediction results, which include the transport location, concentration distribution and diffusion boundary information of nitrogen in the root zone soil at each time point.
[0027] It should be noted that, because ammonium nitrogen in the soil is affected by water migration, soil structure, and leaching after application, it undergoes spatial diffusion and vertical migration within the root zone. Therefore, its actual effective supply is not equivalent to the applied amount. Thus, nitrogen transport prediction needs to be combined with the ammonium nitrogen leaching rate in the root zone. By establishing a nitrogen transport prediction model, the changes in ammonium nitrogen concentration, migration direction, and diffusion boundary at different grid locations over a future time period are simulated. Furthermore, by using a nitrogen backward supply threshold to limit the nitrogen input range in each region, the dynamic changes in soil nitrogen can be predicted while meeting the needs of autumn shoot maturation. Through this method, a shift from static fertilization control to dynamic nitrogen regulation based on growth needs and soil migration patterns is achieved. The nitrogen transport prediction results characterize the nitrogen dissolution and diffusion process in the root zone of different regions over a predetermined time period.
[0028] According to an embodiment of the present invention, the step of determining the water and fertilizer supply parameters for different zones based on the nitrogen transport prediction results and constructing a water and fertilizer supply scheme for the mandarin orange orchard specifically includes: The average concentration of ammonium nitrogen in all grids within the same partition is calculated based on the nitrogen transport prediction results. The difference between the average concentration and the nitrogen shift supply threshold of the corresponding partition is calculated to obtain the nitrogen deficit of each partition at the current time node. The amount of nitrogen replenishment per session in each partition of the mandarin orange orchard is determined based on the nitrogen deficit. Based on the spatial gradient direction of ammonium nitrogen concentration and the migration trajectory of diffusion boundary in each zone, the nitrogen concentration difference between adjacent zones is determined. The nitrogen lateral migration flux is obtained by dividing the nitrogen concentration difference by the zone spacing. The nitrogen lateral migration flux is vector-superimposed with the ammonium nitrogen leaching rate data in the vertical direction to obtain the three-dimensional transport vector of nitrogen in the root zone soil. The nitrogen migration direction is determined based on the three-dimensional migration vector. When the nitrogen lateral migration flux of a certain zone points to the adjacent zone, the nitrogen single replenishment amount of the current zone is proportionally distributed to the adjacent zone according to the flux ratio, so that the nitrogen replenishment direction is consistent with the natural nitrogen migration direction in the soil, and the renewal nitrogen single replenishment amount of each zone is obtained. Based on the autumn shoot maturation delay index of each region, the ratio of the autumn shoot maturation delay index to the baseline reference value is calculated. This ratio is used as a time decay factor to scale the amount of nitrogen replenishment per update in the time dimension, so as to obtain the amount of nitrogen replenishment per time in each region at different time nodes. Based on the ammonium nitrogen leaching rate data of each zone, the ratio of the ammonium nitrogen leaching rate to the corresponding nitrogen time-sharing replenishment amount of the zone is calculated to obtain the water-nitrogen coupling ratio of each zone. The water-nitrogen coupling ratio is used as the conversion benchmark for irrigation water consumption, and the irrigation water consumption matching the nitrogen time-sharing replenishment amount of each zone is calculated. By linking and integrating the nitrogen time-sharing replenishment amount, irrigation water volume, replenishment time node, and spatial coordinates of each zone, a set of water and fertilizer variable supply parameters for each zone in the mandarin orange orchard is obtained. Based on the set of water and fertilizer variable supply parameters, a zoned water and fertilizer supply scheme for the mandarin orange orchard is constructed. The set of water and fertilizer variable supply parameters includes nitrogen supply intensity, irrigation quota, fertilization start time, fertilization duration, and spatial application coordinates. Based on this set, a zoned asynchronous water and fertilizer supply time series table covering the entire orchard is generated. The water and fertilizer variable supply parameters of different grids under the same time node in the zoned asynchronous water and fertilizer supply time series table are mapped to the mandarin orange orchard geographic information system layer according to spatial coordinates, thus constructing a mandarin orange orchard water and fertilizer supply scheme that includes spatial distribution characteristics and time series characteristics.
[0029] It should be noted that by calculating the difference between the current ammonium nitrogen concentration in the root zone soil and the nitrogen backward supply threshold in each region based on nitrogen transport prediction results, the actual nitrogen demand status of different regions can be accurately determined. This avoids delayed autumn shoot maturation due to insufficient nitrogen supply, while also preventing nitrogen accumulation and leaching losses caused by excessive nitrogen application. Furthermore, by analyzing the ammonium nitrogen concentration gradient and spatial migration direction between regions, and constructing a three-dimensional nitrogen transport vector by combining lateral migration flux and vertical leaching rate, the migration influence relationship of nitrogen between different regions can be identified, and the nitrogen supply amount between adjacent regions can be dynamically corrected. This ensures that the applied nitrogen fully considers the natural migration process within the soil, improving the accuracy of nitrogen spatial distribution. By introducing the autumn shoot maturation delay index as a time adjustment factor, the nitrogen supply amount can be dynamically adjusted. This allows for changes in the nitrogen supply rhythm according to the future maturation needs of autumn shoots in different regions, matching the nitrogen supply time with the plant's absorption needs, and promoting timely autumn shoot maturation and flower bud differentiation. Further analysis of ammonium nitrogen leaching rates determined the water-nitrogen coupling relationship, ensuring irrigation water volume matched nitrogen migration and root absorption processes. This prevented insufficient water leading to ineffective nitrogen transport and excessive irrigation causing nitrogen leaching. Ultimately, by integrating parameters such as timed nitrogen replenishment, irrigation water volume, replenishment timing, and spatial location for each zone, a differentiated water and fertilizer variable supply scheme was developed. This transformed water and fertilizer management in mandarin orange orchards from experience-based, uniform control to precise regulation based on growth status, soil environment, and spatial differences, improving water and fertilizer resource utilization efficiency and reducing the risk of nitrogen loss. The flux ratio refers to the proportion of lateral nitrogen migration flux from one zone to adjacent zones relative to the total lateral nitrogen migration flux of that zone, characterizing the relative contribution of nitrogen transport from the current zone to different adjacent areas.
[0030] Figure 3 A flowchart illustrating the calculation of the nitrogen supply deviation coefficient according to the present invention is shown.
[0031] According to an embodiment of the present invention, the real-time monitoring of the change rate of soluble sugar content in branches and the content of flower bud differentiation markers in each zone after implementing the water and fertilizer supply plan within a preset time period, and the calculation of the nitrogen supply deviation coefficient based on the change rate of soluble sugar content and the content of flower bud differentiation markers, specifically involves: During the implementation of the water and fertilizer supply plan, the phloem tissue of the middle section of the marked autumn shoots in each zone was collected at fixed time intervals. The anthrone colorimetric method was used to extract sugar components and perform colorimetric reactions on the collected phloem tissue samples. The absorbance value was measured and converted to obtain the soluble sugar content of the branches in each zone. The rate of change of soluble sugar content in branches of each zone was determined based on the soluble sugar content at each collection time. Terminal bud tissues of the same autumn shoots in each zone were collected, and the mass concentration ratio of abscisic acid to gibberellin in the terminal bud tissues was detected by enzyme-linked immunosorbent assay (ELISA). The mass concentration ratio was used as the content of flower bud differentiation markers. The actual nitrogen absorption of autumn shoots of mandarin oranges in each zone was determined based on the soluble sugar content and flower bud differentiation marker content of the branches. The nitrogen supply deviation coefficient was determined by comparing the actual nitrogen absorption with the amount of nitrogen replenished in a single cycle for each zone.
[0032] It should be noted that the citrus water and fertilizer supply plan is based on predictions of autumn shoot maturation and nitrogen transport. However, soil environment, root absorption capacity, and plant physiological state can lead to differences between actual nitrogen absorption and theoretical replenishment. If fertilization is continuously provided solely based on model predictions, insufficient or excessive nitrogen supply can easily occur, affecting autumn shoot maturation, carbohydrate accumulation, and flower bud differentiation. Therefore, after implementing the water and fertilizer supply plan, it is necessary to monitor the plant's physiological response to evaluate the nitrogen supply effect. Specifically, the soluble sugar content in branches reflects the accumulation level of carbon assimilation products during autumn shoot maturation. When nitrogen supply is appropriate, the plant can maintain a reasonable carbon-nitrogen metabolic balance, promoting the transport of sugars to branches and buds, providing a material basis for flower bud differentiation. Conversely, insufficient or excessive nitrogen supply can lead to an imbalance in carbon-nitrogen metabolism, causing changes in the rate of soluble sugar accumulation. The content of flower bud differentiation markers reflects the physiological state of the terminal bud transitioning from vegetative to reproductive growth. The change in the ratio of abscisic acid to gibberellin corresponds to the flower bud differentiation process and can characterize the plant's response to current nitrogen nutrition conditions. Since nitrogen participates in chlorophyll formation, protein synthesis, and carbon and nitrogen metabolism regulation, the actual nitrogen absorption and utilization level of autumn shoots directly affects the rate of change in soluble sugar accumulation in branches and the content of flower bud differentiation markers. Therefore, by establishing the correspondence between these two factors and the nitrogen absorption status, the actual utilization effect of nitrogen supply on autumn shoots of mandarin oranges in each region can be comprehensively reflected. The actual nitrogen absorption calculated based on the rate of change in soluble sugar content in branches and the content of flower bud differentiation markers, compared with the amount of nitrogen replenished in a single cycle, can determine the nitrogen supply deviation coefficient.
[0033] According to an embodiment of the present invention, the optimization of the water and fertilizer supply scheme based on the nitrogen supply deviation coefficient to obtain the water and fertilizer variable management strategy for tangerines specifically includes: The tangerine orchards with nitrogen supply deviation coefficients greater than the preset value are marked as areas to be optimized. The ratio of the actual nitrogen absorption of autumn shoots in the areas to be optimized to the single replenishment of nitrogen in the corresponding areas is used as the nitrogen absorption benefit value. A linear correction function is constructed with deviation correction factor as weight and nitrogen absorption benefit value as input variable to calculate the nitrogen replenishment correction amount of each area to be optimized at the current time node. The nitrogen time-sharing supply amount of each partition to be optimized is corrected on a time-by-time basis according to the nitrogen supply correction amount. The ratio of the corrected nitrogen time-sharing supply amount to the ammonium nitrogen leaching rate data of the corresponding partition is calculated again to update the water-nitrogen coupling ratio of each partition to be optimized. The irrigation water consumption of each partition to be optimized is re-determined based on the updated water-nitrogen coupling ratio. Based on the corrected nitrogen time-sharing replenishment amount, updated irrigation water volume, and replenishment time nodes for each zone to be optimized, the water and fertilizer supply plan is optimized to obtain the water and fertilizer variable management strategy for mandarin oranges.
[0034] It should be noted that optimizing the water and fertilizer supply plan based on the nitrogen supply deviation coefficient can solve the problem of discrepancies between the water and fertilizer supply parameters determined in the early stage based on growth prediction and nitrogen transport models and the actual absorption and utilization status of mandarin orange plants, thus achieving dynamic feedback adjustment in the water and fertilizer management process. Since different zones are affected by soil physicochemical properties, root distribution, environmental conditions, and plant growth status, their actual nitrogen absorption efficiency varies. By identifying areas where the nitrogen supply deviation coefficient exceeds the preset value, abnormal areas where water and fertilizer supply does not match plant needs can be accurately located. The actual utilization effect after the current nitrogen input is evaluated using the nitrogen absorption benefit value. A linear correction function is constructed based on the deviation correction factor to quantitatively correct the nitrogen supply amount, making the adjusted nitrogen supply amount more consistent with the actual absorption capacity of the plants. Furthermore, by recalculating the water-nitrogen coupling ratio based on the corrected nitrogen supply amount and simultaneously adjusting the irrigation water volume, the water supply, nitrogen migration, and root absorption processes can be coordinated, avoiding ineffective nitrogen transport due to insufficient irrigation and nitrogen leaching due to excessive irrigation. By continuously utilizing the physiological response results of plants to dynamically optimize water and fertilizer parameters, a shift from one-time water and fertilizer planning to closed-loop regulation based on growth feedback has been achieved. This enables the water and fertilizer supply to be adaptively adjusted according to changes in the growth status of mandarin oranges, thereby improving nitrogen fertilizer utilization efficiency and water resource utilization efficiency.
[0035] According to an embodiment of the present invention, it further includes: Acquire the canopy spectral reflectance data of mandarin orange plants in each zone, calculate the normalized vegetation index and red-edge chlorophyll index based on the canopy spectral reflectance data, compare the normalized vegetation index and red-edge chlorophyll index with the pre-established standard spectral library of healthy plants, and when the normalized vegetation index is lower than the health threshold and the red-edge chlorophyll index is higher than the health threshold, determine that there is pest and disease infestation in the zone and generate a pest and disease interference label. Based on the pest and disease interference identifier, retrieve the historical meteorological data and pest monitoring data of the corresponding zone, determine the pest and disease type and hazard level through correlation analysis, query the pre-stored pest and disease nitrogen absorption interference coefficient database based on the pest and disease type and hazard level, and obtain the nitrogen absorption interference coefficient of the zone under the current pest and disease infestation. Substitute the nitrogen absorption interference coefficient into the calculation process of the nitrogen supply deviation coefficient, and correct the actual nitrogen absorption based on the ratio of the nitrogen supply deviation coefficient to the nitrogen absorption interference coefficient to obtain the corrected actual nitrogen absorption after eliminating pest and disease interference. By comparing the actual amount of nitrogen absorbed under correction with the amount of nitrogen replenished in a single batch in each zone, the corrected nitrogen supply deviation coefficient is determined. Based on the corrected nitrogen supply deviation coefficient, the water and fertilizer supply scheme is optimized to obtain a water and fertilizer variable management strategy for tangerines that excludes the interference of pests and diseases.
[0036] According to an embodiment of the present invention, the step of querying a pre-stored database of nitrogen absorption interference coefficients for pests and diseases based on the pest and disease type and damage level to obtain the nitrogen absorption interference coefficient of the partition under the current pest and disease infestation is specifically as follows: Based on the type and severity of the pests and diseases, the corresponding basic interference coefficient is retrieved from the pre-stored database of nitrogen absorption interference coefficients for pests and diseases. The basic interference coefficient is the average percentage of inhibition of nitrogen absorption efficiency of a single pest or disease on tangerines under historical experimental conditions. Obtain the age data and current phenological period data of the mandarin orange trees in this partition, and query the corresponding age and phenological correction coefficient from the tree age and phenological correction coefficient table based on the mandarin orange tree age data and current phenological period data; Multiplying the basic interference coefficient by the tree age and phenology correction coefficient yields the nitrogen absorption interference coefficient of the region under the current pest and disease infestation, wherein the nitrogen absorption interference coefficient characterizes the ratio of the decrease in the actual nitrogen absorption efficiency of the mandarin orange plant to the theoretical absorption efficiency caused by pest and disease infestation.
[0037] It should be noted that in the practical application of water and fertilizer variable management in mandarin oranges, traditional feedback regulation mechanisms rely solely on soluble sugars and physiological markers to determine nitrogen supply. However, diseases and pests such as Huanglongbing (HLB), root rot, and spider mites in mandarin orange orchards often lead to a decline in the absorption capacity of the plant's roots or the metabolic function of its leaves. In such cases, even if the soil nitrogen is sufficient, the plant may still exhibit nitrogen deficiency symptoms. The system may misjudge this as insufficient nitrogen supply and apply additional fertilizer, creating a vicious cycle of "disease causing a false impression of nitrogen deficiency, misjudgment leading to excessive fertilization, and excessive fertilization exacerbating the disease," resulting in water and fertilizer waste and tree decline.
[0038] Therefore, by using canopy spectral reflectance to calculate the normalized vegetation index and red-edge chlorophyll index in real time, and comparing them with a spectral database of healthy plants, pest and disease infestation zones can be identified and interference markers generated before symptoms appear. By correlating meteorological and pest data, the types and severity of pests and diseases are determined. The basic interference coefficient is then corrected based on tree age and phenological stage to obtain a nitrogen absorption interference coefficient that quantitatively represents the proportion of nitrogen absorption efficiency attenuation. This coefficient is then substituted into the nitrogen supply deviation coefficient calculation process to correct the actual nitrogen absorption, eliminating false nitrogen deficiency signals caused by pests and diseases, and ensuring that the corrected deviation coefficient truly reflects the plant's nitrogen nutrition status. This optimizes the water and fertilizer supply plan, avoiding mis-adjustment and over-fertilization, and realizes a variable water and fertilizer management strategy for tangerines that accurately matches the plant's true nitrogen needs under pest and disease infestation conditions. Figure 4The diagram shows a block diagram of an AI-based water and fertilizer variable management system for tangerines based on growth prediction, according to the present invention.
[0039] A second aspect of the present invention also provides a growth prediction-based AI water and fertilizer variable management system for tangerines. The system includes: a memory 401, a processor 402, and a communication interface 403. The memory includes a growth prediction-based AI water and fertilizer variable management method program for tangerines. The communication interface is used for data connection and communication between the memory and the processor. When the growth prediction-based AI water and fertilizer variable management method program for tangerines is executed by the processor, it implements the steps of the growth prediction-based AI water and fertilizer variable management method for tangerines as described in any of the above claims.
[0040] This invention discloses an AI-based water and fertilizer variable management method and system for mandarin oranges based on growth prediction. The method acquires data on total nitrogen content in leaves and lignification rate of branches during the autumn shoot maturation period to construct an evaluation matrix for the autumn shoot maturation process and calculates a maturation delay index. It generates a mandarin orange orchard zoning map based on a density clustering algorithm combined with the delay index to identify spatiotemporal heterogeneity. A nitrogen transport prediction model is constructed by integrating the nitrogen shift supply threshold and the soil ammonium nitrogen leaching rate to determine water and fertilizer variable supply parameters and generate supply plans. The method also monitors the change rate of soluble sugars in branches and the content of flower bud differentiation markers in real time, calculates the nitrogen supply deviation coefficient, and dynamically optimizes the plan. This invention achieves precise matching of water and fertilizer supply with the autumn shoot maturation process, effectively promoting flower bud differentiation, reducing nitrogen leaching loss, and improving water and fertilizer utilization efficiency and fruit quality in mandarin oranges.
[0041] 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.
[0042] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.
[0043] 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 variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for managing water and fertilizer variables in tangerines based on growth prediction, characterized in that, Includes the following steps: Data on total nitrogen content in leaves and lignification rate of branches in each sampling unit within the mandarin orange orchard during the autumn shoot maturation period were obtained. An evaluation matrix for the autumn shoot maturation process was constructed based on the total nitrogen content in leaves and lignification rate of branches. The autumn shoot maturation delay index for each sampling unit was determined based on the evaluation matrix. Based on the density clustering algorithm and the autumn shoot maturation delay index, the sampling units of the mandarin orange orchard are partitioned to identify the spatiotemporal heterogeneity characteristics of the autumn shoot maturation process in the mandarin orange orchard, and to obtain the autumn shoot maturation partition map of the mandarin orange orchard. Based on the autumn shoot maturation zoning map and autumn shoot maturation delay index of the mandarin orange orchard, the nitrogen supply threshold for different zoning zones was determined. A nitrogen transport prediction model was constructed by combining the ammonium nitrogen leaching rate data of the root zone soil. The nitrogen transport prediction results were determined based on the nitrogen transport prediction model. Based on the nitrogen transport prediction results, water and fertilizer supply parameters for different zones were determined, and a water and fertilizer supply scheme for the mandarin orange orchard was constructed. Real-time monitoring of the change rate of soluble sugar content in branches and the content of flower bud differentiation markers in each zone after implementing the water and fertilizer supply plan within a preset time period; calculation of nitrogen supply deviation coefficient based on the change rate of soluble sugar content and the content of flower bud differentiation markers. Based on the nitrogen supply deviation coefficient, the water and fertilizer supply scheme is optimized to obtain the water and fertilizer variable management strategy for mandarin oranges.
2. The AI-based water and fertilizer variable management method for tangerines based on growth prediction according to claim 1, characterized in that, The process involves acquiring leaf total nitrogen content data and branch lignification rate data for each sampling unit within the mandarin orange orchard during the autumn shoot maturation period. Based on these data, an evaluation matrix for the autumn shoot maturation process is constructed. The autumn shoot maturation delay index for each sampling unit is then determined using this evaluation matrix. Specifically: Based on the geographical location information of each sampling unit in the mandarin orange orchard, mature leaves from the middle of the current year's autumn shoots in each sampling unit were collected at a fixed cycle during the autumn shoot maturation period. The collected leaf samples were digested using the Kjeldahl method, and the nitrogen concentration in the digestion solution was measured by a spectrophotometer to obtain the total nitrogen content data of the leaves in each sampling unit. In each sampling unit, a preset number of current-year autumn shoots are marked, and the length of bark browning of the marked branches is measured according to a preset sampling period. The percentage of bark browning length to the total length of the marked branches is used as the branch lignification rate data. The leaf total nitrogen content data and branch lignification rate data of the same sampling unit are associated and aligned according to the sampling unit identifier, and a two-dimensional matrix is constructed with the sampling unit as the row vector and the leaf total nitrogen content and branch lignification rate as the column vector. The growth of the autumn shoots of tangerines is predicted based on the two-dimensional matrix, and the leaf total nitrogen content and branch lignification rate in the future preset time period are determined to construct an evaluation matrix for the maturation process of autumn shoots. The degree of growth and maturity of autumn shoots within a preset time period is determined based on the autumn shoot maturity process evaluation matrix. The absolute value of the difference between the degree of growth and maturity and the preset autumn shoot maturity benchmark reference value is used as the autumn shoot maturity delay index for each sampling unit.
3. The AI-based water and fertilizer variable management method for tangerines based on growth prediction according to claim 1, characterized in that, The sampling units of the mandarin orange orchard are partitioned based on the density clustering algorithm and the autumn shoot maturation delay index to identify the spatiotemporal heterogeneity characteristics of the autumn shoot maturation process within the mandarin orange orchard, resulting in a partition map of the autumn shoot maturation process in the mandarin orange orchard. Specifically: Using the geographical coordinates of each sampling unit as the spatial dimension input and the autumn shoot ripening delay index corresponding to each sampling unit as the attribute dimension input, a multidimensional dataset containing spatial coordinates and attribute values is constructed. The neighborhood radius parameter and the minimum number of sampling units in the neighborhood parameter are set for density clustering. Traverse each sampling unit in the multidimensional dataset, delineate a spherical neighborhood with the spatial coordinates of the currently traversed sampling unit as the center and the neighborhood radius as the radius, count the total number of sampling units in the spherical neighborhood, and when the total number of sampling units is not less than the minimum number of sampling units in the neighborhood, mark the currently traversed sampling unit as a density-reachable core point, and classify all sampling units in the spherical neighborhood into the same candidate cluster set; An iterative expansion operation is performed sequentially on the non-core point sampling units in each candidate cluster set. A new spherical neighborhood is defined with the spatial coordinates of the non-core point sampling unit as the center and the neighborhood radius as the radius. The total number of sampling units in the new spherical neighborhood is counted. When the total number of sampling units in the new spherical neighborhood is not less than the minimum number of sampling units in the neighborhood, the non-core sampling unit is marked as a core point, and all sampling units in the new spherical neighborhood of the core point that are not included in the cluster set are included in the current candidate cluster set. A unique partition identifier is marked for all candidate cluster sets that have been expanded. The spatial coordinate boundaries of all sampling units in each candidate cluster set are extracted. The polygonal spatial contours of each partition are fitted according to the spatial coordinate boundaries. The polygonal spatial contours of each partition are correlated and mapped with the corresponding partition identifier, the mean and variance of the autumn shoot maturation delay index of all sampling units in the partition, and a partition map of autumn shoot maturation in the mandarin orange orchard is constructed.
4. The AI-based water and fertilizer variable management method for tangerines based on growth prediction according to claim 1, characterized in that, The nitrogen transport prediction model is constructed by determining the nitrogen transport supply threshold for different zones based on the autumn shoot maturation zoning map and autumn shoot maturation delay index of the mandarin orange orchard, and combining it with the ammonium nitrogen leaching rate data of the root zone soil. The nitrogen transport prediction results are then determined based on the nitrogen transport prediction model. Specifically: The mean value of the autumn shoot maturation delay index of each zone is extracted from the autumn shoot maturation zone map of the mandarin orange orchard. The deviation of the mean value from the autumn shoot maturation benchmark reference value is calculated. The weighted calculation is performed in combination with the spatial distribution density of the sampling units in the zone to obtain the nitrogen shift supply threshold for different zones. The nitrogen shift supply threshold is used as the upper limit constraint condition for nitrogen application. Acquire ammonium nitrogen leaching rate data in the root zone soil, divide the root zone soil into several discrete spatial grids, use the grid center coordinates as location identifiers, associate the ammonium nitrogen leaching rate data with the grid location, and construct an initial matrix for the spatial distribution of ammonium nitrogen. Using the autumn shoot maturity delay index as the state transition weight, the ammonium nitrogen concentration change process in each grid is defined as the state transition process of the Markov chain. The time step of the state transition is set as the minimum time unit of the future preset time period, and a state transition probability matrix is constructed. The initial matrix of the spatial distribution of ammonium nitrogen is used as the current state vector of the Markov chain. It is substituted into the state transition probability matrix for iterative calculation. Each iteration corresponds to a nitrogen transport simulation of a time step. During the iteration process, the nitrogen shift supply threshold is used as the cutoff boundary of nitrogen input for each grid. When the cumulative nitrogen input of a grid reaches the threshold, the nitrogen increment calculation of that grid is stopped. After a number of iterations matching the preset time period, the ammonium nitrogen concentration value and location coordinates of each grid at different time nodes are output. The trajectory of the change of ammonium nitrogen concentration over time and the spatial migration path of all grids in the same partition are integrated to obtain the nitrogen transport prediction results, which include the transport location, concentration distribution and diffusion boundary information of nitrogen in the root zone soil at each time point.
5. The AI-based water and fertilizer variable management method for tangerines based on growth prediction according to claim 1, characterized in that, The step of determining water and fertilizer supply parameters for different zones based on the nitrogen transport prediction results and constructing a water and fertilizer supply scheme for the mandarin orange orchard is as follows: The average concentration of ammonium nitrogen in all grids within the same partition is calculated based on the nitrogen transport prediction results. The difference between the average concentration and the nitrogen shift supply threshold of the corresponding partition is calculated to obtain the nitrogen deficit of each partition at the current time node. The amount of nitrogen replenishment per session in each partition of the mandarin orange orchard is determined based on the nitrogen deficit. Based on the spatial gradient direction of ammonium nitrogen concentration and the migration trajectory of diffusion boundary in each zone, the nitrogen concentration difference between adjacent zones is determined. The nitrogen lateral migration flux is obtained by dividing the nitrogen concentration difference by the zone spacing. The nitrogen lateral migration flux is vector-superimposed with the ammonium nitrogen leaching rate data in the vertical direction to obtain the three-dimensional transport vector of nitrogen in the root zone soil. The nitrogen migration direction is determined based on the three-dimensional migration vector. When the nitrogen lateral migration flux of a certain partition points to the adjacent partition, the nitrogen single replenishment amount of the current partition is proportionally distributed to the adjacent partition according to the flux ratio to obtain the updated nitrogen single replenishment amount of each partition. Based on the autumn shoot maturation delay index of each region, the ratio of the autumn shoot maturation delay index to the baseline reference value is calculated. This ratio is used as a time decay factor to scale the amount of nitrogen replenishment per update in the time dimension, so as to obtain the amount of nitrogen replenishment per time in each region at different time nodes. Based on the ammonium nitrogen leaching rate data of each zone, the ratio of the ammonium nitrogen leaching rate to the corresponding nitrogen time-sharing replenishment amount of the zone is calculated to obtain the water-nitrogen coupling ratio of each zone. The water-nitrogen coupling ratio is used as the conversion benchmark for irrigation water consumption, and the irrigation water consumption matching the nitrogen time-sharing replenishment amount of each zone is calculated. By linking and integrating the nitrogen time-sharing replenishment amount, irrigation water volume, replenishment time node and spatial coordinates of each zone, a set of water and fertilizer variable supply parameters for each zone in the mandarin orange orchard is obtained. Based on the set of water and fertilizer variable supply parameters, a zoned water and fertilizer supply scheme for the mandarin orange orchard is constructed.
6. The AI-based water and fertilizer variable management method for tangerines based on growth prediction according to claim 1, characterized in that, The real-time monitoring measures the rate of change in soluble sugar content and the content of flower bud differentiation markers in branches of each zone after implementing the water and fertilizer supply plan within a preset time period. The nitrogen supply deviation coefficient is calculated based on the rate of change in soluble sugar content and the content of flower bud differentiation markers. Specifically: During the implementation of the water and fertilizer supply plan, the phloem tissue of the middle section of the marked autumn shoots in each zone was collected at fixed time intervals. The anthrone colorimetric method was used to extract sugar components and perform colorimetric reactions on the collected phloem tissue samples. The absorbance value was measured and converted to obtain the soluble sugar content of the branches in each zone. The rate of change of soluble sugar content in branches of each zone was determined based on the soluble sugar content at each collection time. Terminal bud tissues of the same autumn shoots in each zone were collected, and the mass concentration ratio of abscisic acid to gibberellin in the terminal bud tissues was detected by enzyme-linked immunosorbent assay (ELISA). The mass concentration ratio was used as the content of flower bud differentiation markers. The actual nitrogen absorption of autumn shoots of mandarin oranges in each zone was determined based on the soluble sugar content and flower bud differentiation marker content of the branches. The nitrogen supply deviation coefficient was determined by comparing the actual nitrogen absorption with the amount of nitrogen replenished in a single cycle for each zone.
7. The AI-based water and fertilizer variable management method for tangerines based on growth prediction according to claim 1, characterized in that, The optimization of the water and fertilizer supply scheme based on the nitrogen supply deviation coefficient yields a water and fertilizer variable management strategy for tangerines, specifically as follows: The tangerine orchards with nitrogen supply deviation coefficients greater than the preset value are marked as areas to be optimized. The ratio of the actual nitrogen absorption of autumn shoots in the areas to be optimized to the single replenishment of nitrogen in the corresponding areas is used as the nitrogen absorption benefit value. A linear correction function is constructed with deviation correction factor as weight and nitrogen absorption benefit value as input variable to calculate the nitrogen replenishment correction amount of each area to be optimized at the current time node. The nitrogen time-sharing supply amount of each partition to be optimized is corrected on a time-by-time basis according to the nitrogen supply correction amount. The ratio of the corrected nitrogen time-sharing supply amount to the ammonium nitrogen leaching rate data of the corresponding partition is calculated again to update the water-nitrogen coupling ratio of each partition to be optimized. The irrigation water consumption of each partition to be optimized is re-determined based on the updated water-nitrogen coupling ratio. Based on the corrected nitrogen time-sharing replenishment amount, updated irrigation water volume, and replenishment time nodes for each zone to be optimized, the water and fertilizer supply plan is optimized to obtain the water and fertilizer variable management strategy for mandarin oranges.
8. A tangerine AI-based water and fertilizer variable management system based on growth prediction, characterized in that, The growth prediction-based AI water and fertilizer variable management system for tangerines includes a memory and a processor. The memory includes a growth prediction-based AI water and fertilizer variable management method program for tangerines. When the growth prediction-based AI water and fertilizer variable management method program for tangerines is executed by the processor, it implements the steps of the growth prediction-based AI water and fertilizer variable management method for tangerines as described in any one of claims 1 to 7.
