Variable irrigation and fertilization decision-making and autonomous wheel irrigation method and device

Abstract

The application provides a variable irrigation and fertilization collaborative decision-making and autonomous wheel irrigation method and device, comprising the following steps: determining the water deficiency diagnosis result and nutrient deficiency amount of each pixel point according to crop state data, soil state data and atmospheric state data of a farmland, and aggregating the result and the amount into representative values at the minimum irrigation unit level; determining the water requirement and fertilizer requirement of each irrigation unit according to the water deficiency diagnosis result and nutrient deficiency amount of each irrigation unit, and generating a digital prescription map; generating a water and fertilizer coupling variable prescription map according to the water deficiency diagnosis result, the nutrient deficiency amount, the digital prescription map and weather forecast data in a future rolling time domain; dividing all the irrigation units into multiple irrigation collaborative groups according to the map; and performing in-group collaborative sequencing decision-making for each irrigation collaborative group to obtain a collaborative scheduling instruction sequence in each irrigation collaborative group. The method can solve the problem that the variable irrigation scheme in the prior art is difficult to realize large-scale irrigation system collaborative scheduling.

Description

Technical Field

[0001] This invention relates to the field of precision management technology in smart agriculture, and in particular to a method and device for variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation. Background Technology

[0002] In large-scale farmland planting scenarios, the existing uniform irrigation management model carries out rotational irrigation operations according to fixed times, water volumes, and irrigation areas. Although simple to operate, it is difficult to meet the needs of precision irrigation. Variable irrigation can dynamically adjust irrigation water volume and nutrient supply according to crop growth needs, soil moisture, and weather conditions, which is of great practical significance for improving water resource utilization efficiency and ensuring stable and high yields in large-scale farmland.

[0003] However, for large-scale farmland comprising hundreds of irrigation units, existing variable irrigation schemes typically employ manual or fixed-rotation-group scheduling methods. This means the division of rotation groups and the allocation of irrigation sequences rely on manual experience or pre-defined fixed groups, making dynamic adjustments impossible based on real-time changes in demand across different areas. Consequently, achieving coordinated scheduling of large-scale irrigation systems is difficult, resulting in low overall operational efficiency. These problems ultimately prevent variable irrigation models from fully realizing their advantages, potentially leading to water and fertilizer waste, crop stress, and other issues that negatively impact yield increases and resource efficiency in large-scale farmland. Summary of the Invention

[0004] This invention provides a method and apparatus for variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation, which solves the problem that existing variable irrigation schemes are difficult to achieve large-scale coordinated scheduling of irrigation systems, and realizes the unity of water and fertilizer resource optimization and execution efficiency.

[0005] This invention provides a method for variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation, comprising the following steps: Based on crop status data, soil status data, and atmospheric status data of farmland, determine the water shortage diagnosis result and nutrient deficiency amount for each pixel; The water shortage diagnosis results and nutrient deficiency of each pixel are aggregated into a representative value at the smallest irrigation unit level to obtain the water shortage diagnosis results and nutrient deficiency of each irrigation unit. Based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, the water and fertilizer requirements of each irrigation unit are determined, and a digital prescription map is generated. Based on the water shortage diagnosis results, the nutrient deficiency, the digital prescription map, and the weather forecast data in the future rolling time domain, a water and fertilizer coupling variable prescription map is generated; the water and fertilizer coupling variable prescription map contains the water and fertilizer demand information of each irrigation unit in the future rolling time domain. Based on the water-fertilizer coupling variable prescription diagram, all irrigation units are divided into multiple irrigation coordination groups; the irrigation coordination groups satisfy the pump station flow constraints and the operation time window constraints. For each irrigation coordination group, an intra-group coordination sorting decision is executed to obtain a sequence of coordination scheduling instructions within each irrigation coordination group.

[0006] In one embodiment, determining the water deficiency diagnosis result and nutrient deficiency amount for each pixel based on crop status data, soil status data, and atmospheric status data of the farmland includes: Based on the crop status data, the soil status data, and the atmospheric status data, the crop water stress index, canopy nitrogen accumulation, and normalized vegetation index of each pixel are obtained by remote sensing inversion. Based on the crop water stress index, the canopy nitrogen accumulation, and the normalized vegetation index, calculate the water shortage diagnosis result and nutrient deficiency for each pixel.

[0007] In one embodiment, the step of aggregating the water shortage diagnosis results and nutrient deficiency amounts of each pixel into representative values ​​at the minimum irrigation unit level to obtain the water shortage diagnosis results and nutrient deficiency amounts of each irrigation unit includes: According to the field pipeline network division rules, multiple pixels are divided into multiple irrigation units, and each irrigation unit corresponds to the area where an electric valve is located. For each irrigation unit, outlier removal is performed on the water shortage diagnosis results of the pixels within the irrigation unit, and the water shortage diagnosis results of the remaining pixels after removal are aggregated to obtain the water shortage diagnosis result of the irrigation unit. For each irrigation unit, the nutrient deficiency of the pixels within the irrigation unit is aggregated to obtain the nutrient deficiency of the irrigation unit.

[0008] In one embodiment, the step of determining the water and fertilizer requirements of each irrigation unit based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, and generating a digital prescription map, includes: Based on the water shortage diagnosis results of the irrigation unit, the area of ​​the irrigation unit, and the crop coefficient, the water requirement of the irrigation unit is determined. The fertilizer requirement of the irrigation unit is determined based on the nutrient deficiency of the irrigation unit, the area of ​​the irrigation unit, and the conversion coefficient.

[0009] In one embodiment, a water-fertilizer coupling variable prescription map is generated based on the water deficiency diagnosis result, the nutrient deficiency amount, the digital prescription map, and future rolling weather forecast data, including: Based on the weather forecast data within the future rolling time domain, determine the predicted reference crop evapotranspiration and effective rainfall for each day within the future rolling time domain; For each irrigation unit, the net irrigation water requirement for each day in the future rolling time domain is determined based on the water shortage diagnosis results of the irrigation unit, the predicted reference crop evapotranspiration, and the effective rainfall. For each irrigation unit, the planned water-fertilizer ratio for each day in the future rolling time domain is determined based on the net irrigation water requirement of the irrigation unit and the fertilizer requirement of the irrigation unit in the digital prescription diagram. For each irrigation unit, the comprehensive water and fertilizer demand urgency index of the irrigation unit is determined based on the water shortage diagnosis results, nutrient deficiency, predicted reference crop evapotranspiration, and effective rainfall. The net irrigation water requirement, planned water-fertilizer ratio, and comprehensive water-fertilizer demand urgency index of each irrigation unit are used as the water-fertilizer demand information in the prescription map of the water-fertilizer coupling variable.

[0010] In one embodiment, based on the water-fertilizer coupling variable prescription diagram, all irrigation units are divided into multiple irrigation coordination groups, including: Based on the water and fertilizer coupling variable prescription map, the water and fertilizer demand characteristics and spatial location information of each irrigation unit are extracted; the water and fertilizer demand characteristics include the comprehensive urgency index of water and fertilizer demand of each irrigation unit and the planned water and fertilizer ratio; Estimate the initial cluster number K based on the total water demand of all irrigation units and the rated water supply of the pumping station; The K irrigation units with the highest comprehensive urgency index of water and fertilizer demand were selected as the initial cluster centers. For each unassigned irrigation unit, the distance between the irrigation unit and each cluster center is calculated based on the planned water-fertilizer ratio and the spatial location information. Based on the distance, the irrigation units are sequentially assigned to the irrigation coordination groups where each cluster center is located; Each time an allocation attempt is made, it is determined whether the allocated irrigation coordination group satisfies the pump station flow constraint and the operation time window constraint. If the pump station flow constraint and the operation time window constraint are satisfied, the allocation is successful. If none of the irrigation coordination groups containing all K cluster centers can satisfy the pump station flow constraint and the operation time window constraint, the number of irrigation coordination groups K is increased and the partitioning process is restarted. After each round of allocation, for each irrigation coordination group, the irrigation unit with the smallest total distance to other irrigation units in the group is selected as the new cluster center; Repeat the steps of assigning and updating cluster centers until the cluster centers no longer change, resulting in K irrigation co-groups.

[0011] In one embodiment, the step of performing intra-group collaborative sorting decision for each irrigation coordination group to obtain a collaborative scheduling instruction sequence within each irrigation coordination group includes: Each irrigation coordination group is treated as an intelligent agent; the input of the intelligent agent is the real-time observation status, which includes the remaining water demand, remaining fertilizer demand, current opening degree of each valve, actual inflow rate of each irrigation unit, and total flow rate and pressure at the group inlet; The intelligent agent outputs control commands for each valve based on the real-time observation status. The control commands include opening adjustment amount and closing / starting commands. The sequence of coordinated scheduling instructions within each irrigation coordination group is obtained based on the control instructions; The intelligent agent is trained in the following way: Construct a simulation environment that includes the hydraulic dynamics of the pipe network; In the simulation environment, the observation state, action space, and reward function of the agent are defined; The observed status includes the remaining water demand, remaining fertilizer demand, current opening degree of each valve, actual inflow rate of each irrigation unit, and total flow rate and pressure at the group inlet for each irrigation unit. The action space includes the opening adjustment amount of each valve and whether to initiate valve closure; The reward function includes a positive reward for completing irrigation on time and a negative penalty for deviations in irrigation volume; The agent is trained in the simulation environment using a reinforcement learning algorithm to obtain the trained agent.

[0012] This invention also provides a variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation device, comprising the following modules: The pixel-level diagnostic module is used to determine the water shortage diagnosis result and nutrient deficiency amount of each pixel based on crop status data, soil status data and atmospheric status data of farmland. The unit diagnosis aggregation module is used to aggregate the water shortage diagnosis results and nutrient deficiency of each pixel into a representative value at the smallest irrigation unit level, so as to obtain the water shortage diagnosis results and nutrient deficiency of each irrigation unit. The first diagram generation module is used to determine the water and fertilizer requirements of each irrigation unit based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, and generate a digital prescription diagram. The second graph generation module is used to generate a water-fertilizer coupling variable prescription map based on the water shortage diagnosis results, the nutrient deficiency amount, the digital prescription map, and the weather forecast data in the future rolling time domain; the water-fertilizer coupling variable prescription map contains the water and fertilizer demand information of each irrigation unit in the future rolling time domain. The dynamic grouping module is used to divide all irrigation units into multiple irrigation coordination groups according to the water-fertilizer coupling variable prescription diagram; the irrigation coordination groups satisfy the pump station flow constraints and the operation time window constraints. The collaborative sorting module is used to perform intra-group collaborative sorting decisions for each irrigation collaborative group to obtain a collaborative scheduling instruction sequence within each irrigation collaborative group.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method as described above.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method as described above.

[0015] The present invention provides a variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method and apparatus. Based on crop status data, soil status data, and atmospheric status data of farmland, it determines the water shortage diagnosis result and nutrient deficiency amount for each pixel, obtaining the degree of water shortage and nutrient deficiency for each pixel. The water shortage diagnosis result and nutrient deficiency amount of each pixel are aggregated into the water shortage diagnosis result and nutrient deficiency amount for each irrigation unit, obtaining the degree of water shortage and nutrient deficiency at the unit level. Based on the water shortage diagnosis result and nutrient deficiency amount of each irrigation unit, the water requirement and fertilizer requirement of each irrigation unit are determined, generating a digital prescription map to obtain the actual water and fertilizer requirements for each irrigation unit. Water and fertilizer requirements; based on water shortage diagnosis results, nutrient deficiency, digital prescription maps, and future rolling weather forecast data, a water and fertilizer coupling variable prescription map is generated, which contains water and fertilizer demand information for each irrigation unit in the future rolling time domain, so that water and fertilizer requirements and weather forecast data are uniformly expressed in the same prescription map; according to the water and fertilizer coupling variable prescription map, all irrigation units are divided into multiple irrigation coordination groups that meet pump station flow constraints and operation time window constraints, resulting in grouping results that simultaneously reflect differences in water and fertilizer demand and physical limitations; for each irrigation coordination group, intra-group coordination ranking decisions are executed to obtain the coordination scheduling instruction sequence within each irrigation coordination group. In this invention, the water-fertilizer coupling variable prescription map generated from the water and fertilizer requirements of each irrigation unit is directly used for irrigation coordination group division and intra-group ranking decisions. This allows the irrigation coordination groups to automatically adjust according to changes in water and fertilizer demand, without relying on fixed rotation irrigation groups. The irrigation coordination groups meet the pump station flow constraints and operating time window constraints, ensuring that the divided irrigation coordination groups do not exceed the pump station's water supply capacity and time window limits. The coordinated scheduling instruction sequence is automatically generated through intra-group coordinated ranking decisions, eliminating the need for manual intervention in the rotation irrigation group division. This invention can dynamically generate and automatically schedule irrigation coordination groups based on the real-time water and fertilizer requirements of each irrigation unit, realizing the dynamic division and automatic scheduling of irrigation coordination groups and solving the problem of difficult coordinated scheduling in large-scale irrigation systems. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method provided by the present invention.

[0018] Figure 2 This is a schematic diagram of the architecture of the variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation system provided by the present invention.

[0019] Figure 3 This is a schematic diagram of the generation process of the water-fertilizer coupling variable prescription diagram provided by the present invention.

[0020] Figure 4 This is a schematic diagram of the system runtime sequence and scrolling refresh provided by the present invention.

[0021] Figure 5 This is a schematic diagram of the variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation device provided by the present invention.

[0022] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0024] Figure 1 This is a flowchart illustrating the variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method provided by this invention. Variable irrigation and fertilization refers to determining the water and fertilizer requirements of each irrigation unit based on the water shortage diagnosis result and nutrient deficiency amount of each pixel, and generating a digital prescription map. Collaborative decision-making refers to directly using the water-fertilizer coupled variable prescription map for irrigation collaborative group division and intra-group collaborative ranking decisions. Autonomous rotation irrigation refers to dynamically dividing irrigation collaborative groups based on the water-fertilizer coupled variable prescription map and automatically generating a collaborative scheduling instruction sequence, without manual intervention in the rotation irrigation group division. This method is applied to the variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation system of this invention. (Refer to...) Figure 2As shown, the system comprises a four-layer architecture: a perception layer, a decision-making layer, a coordination layer, and an execution layer. The perception layer includes soil moisture sensors deployed in various irrigation units in the field, a weather station, and a water and fertilizer diagnostic interface, connected to a remote weather forecast interface. Specifically, the soil moisture sensors collect soil condition data, the weather station collects atmospheric condition data, the water and fertilizer diagnostic interface acquires crop condition data, and the remote weather forecast interface acquires future rolling weather forecast data.

[0025] The decision layer is a high-performance computing platform in the cloud or at the edge, used to perform rolling forecasting of water and fertilizer demand and collaborative scheduling optimization, outputting a prescription map of water and fertilizer coupling variables and a sequence of collaborative scheduling instructions. The coordination layer is an edge computing gateway or field controller, used to receive the sequence of collaborative scheduling instructions output by the decision layer and decompose it into a sequence of instructions executable by valves and fertilizer applicators. The execution layer consists of independently controllable field monitoring and control integrated valves, pump stations, fertilizer applicators, and other physical devices, used to execute operations such as opening and closing and fertilization according to instructions. The specific implementation steps of the present invention are described in detail below in conjunction with the above architecture.

[0026] like Figure 1 As shown, the variable-based coordinated decision-making and autonomous rotation irrigation method includes the following: S110. Based on the crop status data, soil status data, and atmospheric status data of the farmland, determine the water shortage diagnosis result and nutrient deficiency amount for each pixel.

[0027] Based on the acquired crop, soil, and atmospheric status data, a water shortage diagnosis result is calculated for each pixel. Specifically, the reference crop evapotranspiration is calculated using meteorological station data, and combined with the crop water stress index and normalized difference vegetation index obtained from remote sensing inversion, the actual water consumption of each pixel is calculated as the water shortage diagnosis result for that pixel. Simultaneously, based on the canopy nutrient accumulation obtained from remote sensing inversion, the current nutrient deficiency of each pixel can be calculated as the nutrient deficit for that pixel.

[0028] Based on atmospheric state data, crop water stress index and normalized difference vegetation index from crop state data, and soil moisture content from soil state data, the water shortage diagnosis result for each pixel is determined. The water shortage diagnosis result represents the actual water consumption of the crop at that pixel. Simultaneously, based on canopy nutrient accumulation from crop state data, the nutrient deficiency amount for each pixel is determined. The nutrient deficiency amount represents the current degree of nutrient deficiency in the crop at that pixel.

[0029] S120. The water shortage diagnosis results and nutrient deficiency of each pixel are aggregated into a representative value at the smallest irrigation unit level to obtain the water shortage diagnosis results and nutrient deficiency of each irrigation unit.

[0030] An irrigation unit refers to the smallest irrigation area in a field network controlled by a single electric valve, and each irrigation unit contains multiple remote sensing pixels. Water shortage diagnosis results refer to quantitative indicators, calculated based on parameters such as crop water stress index and reference crop evapotranspiration, that characterize the actual water consumption of the crop at that location. Nutrient deficiency refers to quantitative indicators, calculated based on parameters such as crop canopy nutrient accumulation, that characterize the current nutrient deficiency of the crop at that location; for example, it could be nitrogen deficiency.

[0031] After obtaining the water shortage diagnosis result and nutrient deficiency amount for each pixel, these are aggregated to the irrigation unit level. Based on the layout of the field pipeline network, each area controlled by an electric valve constitutes an irrigation unit, and each irrigation unit contains multiple pixels. For each irrigation unit, the water shortage diagnosis results of all pixels within it are aggregated to obtain the water shortage diagnosis result for that irrigation unit; the nutrient deficiency amounts of all pixels within it are aggregated to obtain the nutrient deficiency amount for that irrigation unit. Statistical methods can be used during aggregation, such as averaging after removing outliers or directly taking the median, to eliminate the influence of locally abnormal pixels.

[0032] S130. Based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, determine the water and fertilizer requirements of each irrigation unit and generate a digital prescription map.

[0033] A digital prescription map is a digital map created by encoding the water and fertilizer requirements of each irrigation unit according to its spatial location. After obtaining the water shortage diagnosis results and nutrient deficiency amounts for each irrigation unit, the water requirement for each unit is determined based on the water shortage diagnosis results, and the fertilizer requirement is determined based on the nutrient deficiency amounts. Water requirement is positively correlated with the water shortage diagnosis results, and fertilizer requirement is positively correlated with the nutrient deficiency amounts. The water and fertilizer requirements of each irrigation unit are encoded according to their spatial location to generate the digital prescription map.

[0034] S140. Based on the water shortage diagnosis results, the nutrient deficiency amount, the digital prescription map, and the weather forecast data in the future rolling time domain, a water-fertilizer coupling variable prescription map is generated; the water-fertilizer coupling variable prescription map contains the water and fertilizer demand information of each irrigation unit in the future rolling time domain.

[0035] A water and fertilizer coupled variable prescription map is a digital map generated by further integrating future rolling weather forecast data on the basis of a digital prescription map. It contains information on the water and fertilizer demand of each irrigation unit over a future period of time.

[0036] Based on weather forecast data, the predicted reference crop evapotranspiration and effective rainfall for each future day are determined. Based on the water deficiency diagnosis results, nutrient deficiency, predicted reference crop evapotranspiration, effective rainfall, and water and fertilizer requirements from the digital prescription map for each irrigation unit, the daily water and fertilizer requirements for each irrigation unit in the future rolling time domain are determined. This water and fertilizer requirement information for each irrigation unit is incorporated into the water and fertilizer coupling variable prescription map. The water and fertilizer coupling variable prescription map contains the daily water and fertilizer requirements for each irrigation unit in the future rolling time domain, reflecting the temporal and spatial differences in requirements among the irrigation units, and providing a data foundation for subsequent grouping and ranking.

[0037] S150. According to the water-fertilizer coupling variable prescription diagram, all irrigation units are divided into multiple irrigation coordination groups; the irrigation coordination groups satisfy the pump station flow constraints and the operation time window constraints.

[0038] An irrigation coordination group refers to an operational unit formed by aggregating multiple irrigation units based on similar water and fertilizer requirements and spatial proximity. Pump station flow constraints mean that the peak water demand of the irrigation coordination group does not exceed the rated water supply flow of the pump station or main canal; operational time window constraints mean that the total time to complete irrigation for the coordination group does not exceed a preset time window.

[0039] After generating the water and fertilizer coupling variable prescription map, all irrigation units are divided into multiple irrigation coordination groups based on this map. The division is primarily based on the water and fertilizer demand information and spatial location information of each irrigation unit. The water and fertilizer demand information reflects the water and fertilizer requirements of each irrigation unit, while the spatial location information reflects the geographical distribution of each irrigation unit in the field. During the division process, pump station flow constraints and operating time window constraints must be met; that is, the peak water demand of each irrigation coordination group must not exceed the rated water supply flow of the pump station, and the total time to complete the irrigation of that group must not exceed the preset time window. For irrigation units that cannot meet the constraints, they can be adjusted to other groups, or the number of groups can be increased and the groups re-divided if necessary. Through iterative optimization, a final set of irrigation coordination groups that satisfies all constraints is obtained.

[0040] S160. For each irrigation coordination group, perform intra-group coordination sorting decision to obtain the coordination scheduling instruction sequence within each irrigation coordination group.

[0041] Intra-group coordinated sequencing decision-making refers to the process of determining the valve opening and closing sequence and opening adjustment strategy for each irrigation coordination group. After the group is divided, intra-group coordinated sequencing decisions are executed for each irrigation coordination group, generating a sequence of coordinated scheduling instructions within each group. Specifically, each irrigation coordination group is considered a centralized decision-making entity. This entity determines the opening adjustment amount of each valve and whether to initiate a valve closing procedure based on information such as the remaining water demand, remaining fertilizer demand, current valve opening, actual inflow rate of each irrigation unit, and total flow and pressure at the group inlet. These decisions ultimately form a series of coordinated scheduling instruction sequences.

[0042] After receiving the coordinated scheduling instruction sequence within each irrigation coordination group, the decision-making level distributes the sequence to the coordination level. The coordination level interprets and decomposes the coordinated scheduling instruction sequence into a lower-level instruction sequence executable by valves and fertilizer applicators, and distributes it sequentially to the execution level according to the execution timestamps in the instruction sequence. The execution level consists of independently controllable integrated field monitoring and control valves, pump stations, and fertilizer applicators. Based on the received lower-level instructions, it automatically executes the opening and closing of valves, the adjustment of valve opening degree, and the fertilization operation of the fertilizer applicator. Throughout the entire process, the division of irrigation groups, the allocation of irrigation order, and the determination of valve opening and closing timing are all completed automatically by the system without manual intervention, thus achieving autonomous irrigation.

[0043] In actual operation, whenever new weather forecast data or new field sensing data (i.e., the aforementioned crop status data, soil status data, or atmospheric status data) is obtained, a rolling optimization is triggered, and S110-S160 is re-executed to refresh and adjust the irrigation schemes that have not yet been executed in order to adapt to the constantly changing farmland environment and meteorological conditions.

[0044] The variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method provided by this invention determines the water shortage diagnosis result and nutrient deficiency amount for each pixel based on crop status data, soil status data, and atmospheric status data of farmland, thus obtaining the water shortage degree and nutrient deficiency degree for each pixel; the water shortage diagnosis result and nutrient deficiency amount of each pixel are aggregated into the water shortage diagnosis result and nutrient deficiency amount of each irrigation unit, thus obtaining the water shortage degree and nutrient deficiency degree at the unit level; based on the water shortage diagnosis result and nutrient deficiency amount of each irrigation unit, the water requirement and fertilizer requirement of each irrigation unit are determined, and a digital prescription map is generated to obtain the actual water requirement of each irrigation unit. Based on water shortage diagnosis results, nutrient deficiency, digital prescription maps, and future rolling weather forecast data, a water-fertilizer coupling variable prescription map is generated, containing water and fertilizer demand information for each irrigation unit in the future rolling time domain. This ensures that water and fertilizer requirements are uniformly expressed in the same prescription map along with weather forecast data. According to the water-fertilizer coupling variable prescription map, all irrigation units are divided into multiple irrigation coordination groups that satisfy pump station flow constraints and operating time window constraints, resulting in grouping results that simultaneously reflect differences in water and fertilizer demand and physical limitations. For each irrigation coordination group, an intra-group collaborative ranking decision is executed to obtain a collaborative scheduling instruction sequence within each irrigation coordination group. In this invention, the water-fertilizer coupling variable prescription map generated from the water and fertilizer requirements of each irrigation unit is directly used for irrigation coordination group division and intra-group ranking decisions, allowing irrigation coordination groups to automatically adjust with changes in water and fertilizer demand without relying on fixed rotation irrigation groups. Irrigation coordination groups satisfy pump station flow constraints and operating time window constraints, ensuring that the divided irrigation coordination groups do not exceed the pump station's water supply capacity and time window limitations. The collaborative scheduling instruction sequence is automatically generated through intra-group collaborative ranking decisions, eliminating the need for manual intervention in the rotation irrigation group division. This invention can dynamically generate and automatically schedule irrigation coordination groups based on the real-time water and fertilizer needs of each irrigation unit, realizing the dynamic division and automatic scheduling of irrigation coordination groups and solving the problem of difficult coordination scheduling of large-scale irrigation systems.

[0045] In one embodiment, determining the water deficiency diagnosis result and nutrient deficiency amount for each pixel based on crop status data, soil status data, and atmospheric status data of the farmland includes: Based on the crop status data, the soil status data, and the atmospheric status data, the crop water stress index, canopy nitrogen accumulation, and normalized vegetation index of each pixel are obtained by remote sensing inversion. Based on the crop water stress index, the canopy nitrogen accumulation, and the normalized vegetation index, calculate the water shortage diagnosis result and nutrient deficiency for each pixel.

[0046] Specifically, using atmospheric state data such as temperature, wind speed, humidity, and sunshine duration, the reference crop evapotranspiration is calculated using the Penman-Monteith formula. The unit is mm / d. The reference crop evapotranspiration rate will be used. Multiply by crop coefficient , obtain pixels Potential evaporation Crop coefficient Dynamically adjusted based on the Normalized Difference Vegetation Index (NDVI), using the formula... Calculate the crop coefficient for each pixel. (The last part, "pixel number," appears to be a typo and can be left as is.) Potential evaporation Multiply by the crop water stress index of that pixel , obtain pixels Water shortage diagnosis results The unit is mm / d. When the crop water stress index... At that time, the diagnosis of water shortage This indicates that the pixel is not lacking water. Meanwhile, the canopy nitrogen accumulation amount obtained through remote sensing inversion... Calculate the amount of nitrogen missing in each pixel. This is considered as a nutrient deficiency. Specifically, the crop nitrogen requirement for that pixel is determined based on the target crop yield and recommended nitrogen fertilizer application rate for the area where that pixel is located. Nitrogen deficiency When the nitrogen accumulation in the canopy has reached or exceeded the required amount, the nitrogen deficiency is 0.

[0047] This invention uses remote sensing inversion to obtain the crop water stress index, canopy nitrogen accumulation, and normalized difference vegetation index (NDVI) for each pixel based on crop, soil, and atmospheric state data. The water shortage diagnosis and nutrient deficiency for each pixel are calculated using these metrics, allowing the diagnosis to simultaneously reflect the combined effects of crop water stress, canopy nutrient status, and vegetation cover. By using the water shortage diagnosis and nutrient deficiency for each pixel as input for subsequent aggregation steps, this invention establishes a pixel-level real-time diagnosis-based data acquisition mechanism, providing a quantitative basis for subsequent irrigation unit aggregation and grouping decisions.

[0048] In one embodiment, the step of aggregating the water shortage diagnosis results and nutrient deficiency amounts of each pixel into representative values ​​at the minimum irrigation unit level to obtain the water shortage diagnosis results and nutrient deficiency amounts of each irrigation unit includes: According to the field pipeline network division rules, multiple pixels are divided into multiple irrigation units, and each irrigation unit corresponds to the area where an electric valve is located. For each irrigation unit, outlier removal is performed on the water shortage diagnosis results of the pixels within the irrigation unit, and the water shortage diagnosis results of the remaining pixels after removal are aggregated to obtain the water shortage diagnosis result of the irrigation unit. For each irrigation unit, the nutrient deficiency of the pixels within the irrigation unit is aggregated to obtain the nutrient deficiency of the irrigation unit.

[0049] Specifically, the field network consists of main pipes, branch pipes, and electric valves installed on the branch pipes. Each electric valve controls an irrigator within a certain range, which constitutes an irrigation unit. Based on the location and control range of the electric valves, the pixels covered by the irrigation area corresponding to each electric valve are assigned to the same irrigation unit.

[0050] For each irrigation unit, calculate the first quartile of the water shortage diagnosis results for all pixels within that irrigation unit. and the third and fourth quartiles Define interquartile range The diagnosis of water shortage was lower than... or higher Pixels with outliers are removed. The arithmetic mean of the water shortage diagnosis results of the remaining pixels is calculated and used as the water shortage diagnosis result for that irrigation unit. When the number of pixels in an irrigation unit is less than 5, the median of the water shortage diagnosis results of all pixels is used directly as the water shortage diagnosis result for that irrigation unit.

[0051] For each irrigation unit, the nutrient deficiency of each pixel within that unit is aggregated to obtain the total nutrient deficiency for that unit. Specifically, the median of the nutrient deficiency of all pixels within that unit is calculated, and this median is used as the total nutrient deficiency for that unit. .

[0052] This invention divides multiple pixels into multiple irrigation units based on field pipeline network division rules, with each irrigation unit corresponding to the area where an electric valve is located. For each irrigation unit, outlier removal is performed on the water shortage diagnosis results of the pixels within that unit, and the remaining pixel water shortage diagnosis results are aggregated to obtain the water shortage diagnosis result for the irrigation unit. This avoids interference from extreme pixels such as localized waterlogging or drought on the water shortage diagnosis results, ensuring that the water shortage diagnosis results accurately reflect the overall water shortage status of the irrigation unit. For each irrigation unit, the nutrient deficiency amount of the pixels within that unit is aggregated to obtain the nutrient deficiency amount for the irrigation unit. Since the nutrient deficiency is spatially gradually distributed, the aggregation process can reflect the overall nutrient status of the irrigation unit. By aggregating pixel-level diagnostic results into representative values ​​at the irrigation unit level, this invention achieves data granularity conversion from pixels to irrigation units, enabling subsequent steps to make decisions on an irrigation unit basis, reducing data processing volume while ensuring the accuracy and robustness of the diagnostic results.

[0053] In one embodiment, the step of determining the water and fertilizer requirements of each irrigation unit based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, and generating a digital prescription map, includes: Based on the water shortage diagnosis results of the irrigation unit, the area of ​​the irrigation unit, and the crop coefficient, the water requirement of the irrigation unit is determined. The fertilizer requirement of the irrigation unit is determined based on the nutrient deficiency of the irrigation unit, the area of ​​the irrigation unit, and the conversion coefficient.

[0054] Specifically, water demand ,in, For irrigation units The diagnosis of water shortage For irrigation units area, The crop coefficient is determined based on the field network division. The area of ​​the irrigation unit is determined based on the normalized vegetation index.

[0055] Fertilizer requirement ,in, For irrigation units Nutrient deficiency For irrigation units area, The conversion factor is used. The area of ​​the irrigation unit is determined based on the field pipe network division, and the conversion factor is determined based on the fertilizer utilization rate. The value ranges from 0.6 to 0.9, and the specific value is determined based on the fertilizer utilization rate. The higher the fertilizer utilization rate, the larger the value of the conversion coefficient, and the lower the fertilizer utilization rate, the smaller the value of the conversion coefficient.

[0056] The water demand of each irrigation unit and fertilizer requirements Digital prescription maps are generated by encoding based on spatial location.

[0057] This invention determines the water requirement of an irrigation unit based on its water shortage diagnosis, area, and crop coefficient; and determines its fertilizer requirement based on its nutrient deficiency, area, and conversion coefficient. By multiplying the water shortage diagnosis and nutrient deficiency by the irrigation unit's area and corresponding coefficient, this invention converts the diagnosis results per unit area into the actual total water and fertilizer requirements of the irrigation unit, thus transforming the diagnosis results into actionable water and fertilizer requirement data. The water and fertilizer requirements of each irrigation unit are encoded spatially to generate a digital prescription map, organizing the water and fertilizer requirements in a spatially distributed manner. This provides a structured data foundation for the subsequent generation of a water-fertilizer coupled variable prescription map, and facilitates grouping and decision-making based on spatial location in subsequent steps.

[0058] In one embodiment, a water-fertilizer coupling variable prescription map is generated based on the water deficiency diagnosis result, the nutrient deficiency amount, the digital prescription map, and future rolling weather forecast data, with reference to... Figure 3 As shown, it includes: S210. Based on the weather forecast data in the future rolling time domain, determine the predicted reference crop evapotranspiration and effective rainfall for each day in the future rolling time domain.

[0059] Weather forecast data includes daily temperature, wind speed, humidity, sunshine duration, and rainfall over a future rolling time domain. The Penman-Monteith formula is used to calculate the daily forecast reference crop evapotranspiration. : ; in, The slope of the saturated water vapor pressure curve. Net radiation, For soil heat flux, This is the constant of the wet and dry meter. The average temperature The wind speed at a height of 2 meters The saturated vapor pressure, This refers to the actual water vapor pressure. The effective daily rainfall is determined based on the daily rainfall data in the weather forecast. ,in, This refers to the daily rainfall. For the effective rainfall coefficient, when ≤5mm Take 1.0, when 5mm < When ≤50mm Take 0.8, when When >50mm Take 0.7.

[0060] S220. For each irrigation unit, based on the water shortage diagnosis results of the irrigation unit, the predicted reference crop evapotranspiration rate, and the effective rainfall, determine the net irrigation water requirement of the irrigation unit per day in the future rolling time domain.

[0061] Obtain field diagnostic data, including water shortage diagnosis results. and nutrient deficiency .

[0062] ; in, For irrigation units In the Net irrigation water requirement per day This is based on the current reference crop evapotranspiration deficit determined in real-time, i.e., the water shortage diagnosis results mentioned above. For the first The forecast for the day is based on crop evapotranspiration. For the first Effective rainfall per day, For crop coefficients, For water use efficiency, This serves as an index for the number of days in the future rolling time domain.

[0063] S230. For each irrigation unit, the planned water-fertilizer ratio for each irrigation unit is determined daily in the future rolling time domain based on the net irrigation water requirement of the irrigation unit and the fertilizer requirement of the irrigation unit in the digital prescription diagram.

[0064] Calculate irrigation unit On the day of the decision ( The initial planned water-to-fertilizer ratio And the planned water-to-fertilizer ratio for each day in the future. .in, For irrigation units Net irrigation water demand on the day of decision-making For irrigation units The amount of fertilizer required, For irrigation units In the future The planned water-to-fertilizer ratio for the day For irrigation units In the future Net irrigation water requirement per day. The planned water-to-fertilizer ratio represents the ratio of water and fertilizer requirements. This is a key feature for subsequent dynamic clustering to reduce fertilizer applicator formula switching.

[0065] S240. For each irrigation unit, the comprehensive water and fertilizer demand urgency index of the irrigation unit is determined based on the water shortage diagnosis results, nutrient deficiency, predicted reference crop evapotranspiration, and effective rainfall.

[0066] To quantify the immediate demand for irrigation resources for each irrigation unit and to provide a prioritization basis for dynamic grouping, for each irrigation unit... exist Define a comprehensive urgency index for water and fertilizer demand. The calculation process is as follows.

[0067] For irrigation units Calculate the expected net mitigation effect of future weather on its water shortage situation. .

[0068] ; In the formula, For future rolling time domain day index, It means tomorrow. This means the day after tomorrow, and so on. For the first Effective rainfall per day, For irrigation units In the The forecast for the day is based on crop evapotranspiration. The conversion factor is used for Tengfa; the entire formula represents the first... Daily net contribution of weather to soil moisture (positive value is replenishment, negative value is depletion). Time decay factor (0 < ≤1). It emphasizes that recent weather has a greater impact on alleviating the current urgency than long-term weather. Use a value between 0.8 and 0.95. For example, take... =0.8, tomorrow's rainfall will have 1.25 times the effect of the rainfall the day after tomorrow in alleviating the current drought.

[0069] Calculate the weighted sum of the attenuation of the net weather replenishment effect, i.e., the net water urgency after future weather correction. ,in, For irrigation units The water shortage diagnosis results indicate the current water shortage amount.

[0070] The current water shortage is Subtracting the amount of water that is expected to be replenished by future weather, we obtain the net water urgency, which is the rigid water shortage that cannot be resolved by future weather and must be compensated for by irrigation. If the amount of water that is expected to be replenished by future weather is greater than the current water shortage, the net water urgency is zero, indicating that the irrigation unit does not need irrigation.

[0071] Calculate the comprehensive urgency index of water and fertilizer demand : ; in, For irrigation units The urgent need for net water This is the maximum net water demand urgency across all irrigation units, used for normalization. For irrigation units Nutrient deficiency This represents the maximum nutrient deficiency across all irrigation units, used for normalization. , For dynamic weighting coefficients, + =1, which can be adjusted according to the crop's growth stage; the water weight is increased during the flowering stage. Increase fertilizer application during the fruit expansion period. .

[0072] S250. The net irrigation water demand, planned water-fertilizer ratio, and comprehensive water-fertilizer demand urgency index of each irrigation unit are used as the water-fertilizer coupling variable prescription map as the water-fertilizer demand information.

[0073] The net irrigation water demand of each irrigation unit Planned water-fertilizer ratio Combined urgency index of water and fertilizer demand The water and fertilizer demand information, used as a water-fertilizer coupling variable in the prescription map, forms a multi-dimensional structured list. The time dimension of this list is a continuous time domain in days. The spatial dimension covers all N irrigation units ( The list of work instructions includes daily net irrigation water demand, planned water-fertilizer ratio, and a comprehensive urgency index of water and fertilizer demand. The number of days for the future rolling time domain is set according to actual needs, for example... =7 indicates a forecast of irrigation demand for the next 7 days.

[0074] This invention determines the predicted reference crop evapotranspiration and effective rainfall for each day within a future rolling time domain based on weather forecast data. For each irrigation unit, the net irrigation water requirement for each day within the future rolling time domain is determined based on the water shortage diagnosis results, predicted reference crop evapotranspiration, and effective rainfall. This combines the current water shortage situation with future weather forecasts, ensuring that the net irrigation water requirement reflects the mitigating or aggravating effect of future weather on irrigation demand. For each irrigation unit, the planned water-fertilizer ratio for each day within the future rolling time domain is determined based on the net irrigation water requirement and the fertilizer requirement of the irrigation unit in the digital prescription map. This allows the water-fertilizer ratio to dynamically change with irrigation demand and is based on the actual total fertilizer requirement of the irrigation unit, providing a key feature for reducing prescription switching in subsequent dynamic grouping. For each irrigation unit, a comprehensive water and fertilizer demand urgency index is determined based on the unit's water shortage diagnosis results, nutrient deficiency, predicted reference crop evapotranspiration, and effective rainfall. This index is weighted by the attenuation of the net weather recharge effect and by allowing future weather to mitigate water shortages over time, thus assigning higher weight to recent weather conditions. The amount of water that future weather can replenish is removed from the net water demand urgency, resulting in the rigid water shortage that must be compensated by irrigation. The urgency index is then calculated by weighting the normalized net water demand urgency and the normalized nutrient deficiency, allowing the urgency index to simultaneously reflect the combined water and fertilizer shortage situation. The net irrigation water demand, planned water-fertilizer ratio, and comprehensive water and fertilizer demand urgency index of each irrigation unit are used as water and fertilizer demand information in the prescription map of water and fertilizer coupling variables, providing a complete spatiotemporal data foundation for subsequent dynamic grouping and intra-group ranking decisions.

[0075] In one embodiment, based on the water-fertilizer coupling variable prescription diagram, all irrigation units are divided into multiple irrigation coordination groups, including: S310. Based on the water and fertilizer coupling variable prescription map, extract the water and fertilizer demand characteristics and spatial location information of each irrigation unit; the water and fertilizer demand characteristics include the comprehensive urgency index of water and fertilizer demand of each irrigation unit and the planned water and fertilizer ratio; S320. Estimate the initial cluster number K based on the total water demand of all irrigation units and the rated water supply of the pumping station; S330. Select the K irrigation units with the highest comprehensive urgency index of water and fertilizer demand as the initial cluster centers; S340. For each unassigned irrigation unit, calculate the distance between the irrigation unit and each cluster center based on the planned water-fertilizer ratio and the spatial location information. S350. Based on the distance, attempt to assign the irrigation units to the irrigation coordination groups where each cluster center is located in turn. S360. During each allocation attempt, determine whether the allocated irrigation coordination group satisfies the pump station flow constraint and the operation time window constraint. If the pump station flow constraint and the operation time window constraint are satisfied, the allocation is successful. If all K cluster centers in the irrigation coordination group cannot satisfy the pump station flow constraint and the operation time window constraint, increase the number of irrigation coordination groups K and restart the partitioning process. S370. After each round of allocation, for each irrigation coordination group, select the irrigation unit with the smallest total distance to other irrigation units in the group as the new cluster center. S380. Repeat the steps of assigning and updating cluster centers until the cluster centers no longer change, resulting in K irrigation co-groups.

[0076] Specifically, based on the water and fertilizer coupling variable prescription map, the water and fertilizer demand characteristics and spatial location information of each irrigation unit are extracted. The water and fertilizer demand characteristics include the comprehensive urgency index of water and fertilizer demand for each irrigation unit. Water-to-fertilizer ratio as planned Spatial location information is determined based on the branch pipe or topological node number to which the irrigation unit belongs, and is used to calculate the spatial connectivity between units.

[0077] Based on the total water demand of all irrigation units and the rated water supply of the pumping station Estimate the initial number of clusters K. Specifically, sum the water requirements of all irrigation units (i.e., the water requirements in the digital prescription map) to obtain the total water flow required for simultaneous irrigation of all irrigation units, and divide this total water flow by the rated water supply flow of the pumping station. Then, round up to obtain the minimum number of groups required to satisfy the flow constraints of the pumping station, which is used as the initial cluster number K.

[0078] Select the comprehensive urgency index of water and fertilizer demand The K highest irrigation units are used as the initial cluster centers. .

[0079] For each unassigned irrigation unit According to the plan, the water-to-fertilizer ratio Based on spatial location information, calculate the irrigation unit. with each cluster center Distance between The distance is calculated using a multidimensional distance metric function, which is defined as: ; in, For irrigation units With irrigation unit (i.e., cluster center) The multidimensional distance between the irrigation units is such that the smaller the value, the better the irrigation unit matches the irrigation co-group to which the cluster center belongs. Distance for water and fertilizer mode For spatial distance, For urgency distance, , , These are the weighting coefficients.

[0080] ; in, and Representing irrigation units and irrigation units The planned water-to-fertilizer ratio vector, and Representing irrigation units and irrigation units The planned water-fertilizer ratio vector is the result after standardization. The water-fertilizer pattern distance is the calculated distance between the planned water-fertilizer ratio vectors of two irrigation units. and After standardization, the Euclidean distance is calculated to characterize the overall similarity of water and fertilizer ratio requirements between two irrigation units during the irrigation cycle. The smaller the distance, the more likely they are to be grouped together, allowing the fertilizer applicator to use the same or similar fertilizer stock solution formula, requiring only minor adjustments to the injection ratio, thus reducing the frequency and time spent on fertilizer changes and equipment cleaning.

[0081] ; in, The shortest hydraulic path between the two irrigation units. This represents the maximum value of the shortest hydraulic path between all irrigation units. A smaller distance indicates that the groups are relatively concentrated in space, which can reduce ineffective head loss in the irrigation main.

[0082] ; in, and This is a comprehensive urgency index of water and fertilizer demand for the two irrigation units. and These represent the maximum and minimum values ​​of the comprehensive water and fertilizer demand urgency index for all irrigation units. A smaller distance indicates that units with similar urgency tend to be grouped together, facilitating priority scheduling of high-urgency groups and the adoption of a consistent scheduling strategy in subsequent rankings.

[0083] Weighting coefficients of the three distance components , '、 This can be adjusted according to management objectives. If the primary goal is to reduce the operation of fertilizer applicators, then the amount should be increased. If the energy consumption of water transmission through the pipeline network is the primary concern, then the increase will be significant. '.

[0084] Based on distance, try the irrigation unit in turn. Assigned to each cluster center The irrigation coordination group Each time an allocation attempt is made, the irrigation coordination group is determined after the allocation. Does it meet the pump station flow constraint and operating time window constraint? The pump station flow constraint means that the peak total water demand for simultaneous irrigation by the irrigation coordination group after allocation must not exceed the pump station's rated water supply flow. The operation time window constraint means that the total time required to complete the irrigation of all irrigation units in the irrigation coordination group must not exceed the preset time window. If the pump station flow rate constraint and the operation time window constraint are satisfied, the allocation is successful. If none of the irrigation coordination groups containing all K cluster centers satisfy the pump station flow rate constraint and the operation time window constraint, the number of irrigation coordination groups K is increased and the partitioning process is restarted.

[0085] After each round of allocation is completed, for each irrigation coordination group The irrigation unit with the smallest total distance to other irrigation units within the group is selected as the new cluster center. This ensures that the cluster center is located at the core of the group in the feature space, so as to reflect the common characteristics of each irrigation unit in the group in terms of planned water-fertilizer ratio, spatial location and comprehensive urgency index of water and fertilizer demand to the greatest extent.

[0086] Repeat the steps of assigning and updating cluster centers until the cluster centers no longer change, resulting in K irrigation coordination groups. If any unassigned irrigation units cannot be assigned to any existing irrigation coordination group during the allocation process, these unassigned units will be grouped separately, or reassigned after fine-tuning the flow constraints within the time window. If there are too many unassigned units, it indicates that the initial K value is too small. In this case, the K value will be increased, and the clustering will be re-executed, returning to step S330. Ultimately, K irrigation coordination groups that satisfy the pump station flow constraints, the operation time window constraints, and minimize intra-group distance will be obtained.

[0087] This invention extracts the comprehensive urgency index of water and fertilizer demand and the planned water-fertilizer ratio for each irrigation unit as water and fertilizer demand characteristics based on the water-fertilizer coupling variable prescription map, and extracts spatial location information. The initial cluster number K is estimated based on the total water demand of all irrigation units and the rated water supply flow of the pumping station, allowing the number of groups to be automatically determined according to the system's water supply capacity without manual pre-setting. The K irrigation units with the highest comprehensive urgency index of water and fertilizer demand are selected as initial cluster centers, prioritizing high-urgency units and ensuring that key areas receive higher priority during the grouping process. For each unassigned irrigation unit, the distance between the irrigation unit and each cluster center is calculated based on the planned water-fertilizer ratio and spatial location information. This ensures that grouping is based on both water-fertilizer ratio similarity and spatial proximity. Grouping units with similar water-fertilizer ratios together reduces the number of fertilizer applicator prescription switching times, and grouping units with similar spatial locations together reduces ineffective head loss in the irrigation main. The irrigation groups are assigned sequentially based on distance, and each assignment checks whether the irrigation coordination groups meet the pump station flow constraints and operating time window constraints. Only assignments that meet these constraints are accepted, ensuring that the grouping results reflect both the similarity and spatial proximity of water and fertilizer needs, while also conforming to the physical limitations of pump station water supply capacity and time windows. If none of the irrigation coordination groups corresponding to any cluster center meet the constraints, the number of groups K is increased and the division process restarts. Dynamically adjusting the value of K addresses the issue of a fixed number of groups not meeting the constraints. After each round of allocation, the irrigation unit with the smallest total distance to other irrigation units within the group is selected as the new cluster center. This ensures that the cluster center is located at the center of the group's feature space, maximizing the reflection of the common characteristics of each unit within the group in terms of water-fertilizer ratio, spatial location, and demand urgency. The steps of allocating and updating cluster centers are repeated until the cluster centers no longer change, allowing the grouping structure to approach the optimal solution, until a local optimum is reached. Through the above dynamic clustering process, this invention aggregates irrigation units with similar water and fertilizer requirements and spatial proximity into the same irrigation coordination group, reducing the number of times fertilizer applicator formulas need to be switched. At the same time, the grouping results automatically meet the pump station flow constraints and operation time window constraints, realizing the dynamic division and automatic scheduling of irrigation coordination groups, and solving the problem of difficult coordination scheduling of large-scale irrigation systems.

[0088] The water-fertilizer coupling variable prescription diagram generated by this invention is shown in Table 1 below: Table 1

[0089] In one embodiment, the step of performing intra-group collaborative sorting decision for each irrigation coordination group to obtain a collaborative scheduling instruction sequence within each irrigation coordination group includes: Each irrigation coordination group is treated as an intelligent agent; the input of the intelligent agent is the real-time observation status, which includes the remaining water demand, remaining fertilizer demand, current opening degree of each valve, actual inflow rate of each irrigation unit, and total flow rate and pressure at the group inlet; The intelligent agent outputs control commands for each valve based on the real-time observation status. The control commands include opening adjustment amount and closing / starting commands. The control commands are used to obtain the sequence of coordinated scheduling commands within each irrigation coordination group.

[0090] Specifically, each irrigation coordination group is treated as an agent. For K irrigation coordination groups, there are K agents, each independently responsible for the opening and closing timing and opening degree adjustment of each valve within its group. Each agent observes the local state within the group, including the prescription information of each irrigation unit and the state of its neighbors. The prescription information of each irrigation unit includes its water requirement, fertilizer requirement, and a comprehensive water and fertilizer urgency index. and the pressure of the pipe section where the irrigation unit is located. ,flow The neighbor status includes the current pressure and flow rate of the pipe segment containing the irrigation unit adjacent to the irrigation unit, as well as the comprehensive urgency index of water and fertilizer demand of the adjacent irrigation units. By observing the pressure and flow rate of the adjacent irrigation units, the agent can perceive local changes in the hydraulic state of the pipeline network; by observing the comprehensive urgency index of water and fertilizer demand of the adjacent irrigation units, the agent can understand the urgency of demand in the adjacent area, thereby considering the impact on adjacent irrigation units when making decisions and achieving intra-group collaboration.

[0091] Each irrigation coordination group is treated as an agent. The agent's input is the real-time observation status, which includes the remaining water demand, remaining fertilizer demand, current valve opening, actual inflow rate of each irrigation unit, and total flow rate and pressure at the group inlet. The initial values ​​for the remaining water demand and fertilizer demand are derived from the digital prescription map. The current valve opening ranges from 0 to 1 (0 for fully closed, 1 for fully open), obtained from valve position sensors or control feedback. The actual inflow rate of each irrigation unit is measured by a flow meter or estimated based on valve opening and upstream pressure. The total flow rate and pressure at the group inlet refer to the total water flow and pressure at the upstream water supply pipeline inlet, reflecting the upstream water supply capacity and pressure status for the irrigation coordination group. Furthermore, the real-time observation status also includes information on currently closed valves (recording which valve is being closed and its remaining closing time), the total irrigation time within the group (cumulative time from the start of irrigation to the current moment), and changes in valve openings during the most recent decisions.

[0092] The intelligent agent outputs control commands for each valve based on real-time observation. These commands include opening adjustment amounts and shut-off start commands. The opening adjustment amount refers to the increase or decrease in the opening of each valve from its current position, with a single adjustment not exceeding 10% of the current opening. The shut-off start command selects an action from either not initiating shut-off or initiating a valve shut-off procedure. Once shut-off is initiated, the valve automatically executes the shutdown according to a preset curve.

[0093] The coordinated scheduling instruction sequence within each irrigation coordination group is obtained based on the control commands. This sequence converts the control commands into a format directly recognizable by the execution layer devices. Each instruction includes the following fields: target valve identifier, instruction type (opening adjustment or closing start), instruction value (opening adjustment amount or closing identifier), and execution timestamp. When the instruction type is opening adjustment, the execution layer adjusts the opening of the corresponding valve according to the instruction value; when the instruction type is closing start, the execution layer executes the valve closing procedure according to a preset closing curve. The coordinated scheduling instruction sequence is organized in the order of decision times, forming a complete instruction timetable from the current time to the end of the irrigation cycle. After being sent to the coordination layer, it is decomposed into a sequence of instructions executable by the valves.

[0094] The intelligent agent is trained in the following way: S410. Construct a simulation environment that includes the hydraulic dynamics of the pipeline network.

[0095] The simulation environment is constructed as follows: First, a pipeline network topology model is established. Based on the layout of main pipes, branch pipes, valves, and emitters in the actual irrigation system, the pipeline network is abstracted as a graph structure of nodes and pipe segments. Nodes include water source points, pipe segment connection points, valve installation points, and irrigation unit inlet points. Pipe segments include hydraulic parameters such as pipe length, pipe diameter, and friction coefficient. In practice, an adjacency list or adjacency matrix is ​​used to store the connection relationships between nodes and pipe segments. Each pipe segment records its starting node, ending node, length, diameter, and friction coefficient. This model is used to determine the connection relationships between nodes in the pipeline network and the water flow path, providing a topological basis for hydraulic calculations.

[0096] Secondly, a hydraulic calculation model is established. The Heizen-Williams formula is used to calculate the head loss along the pipe section: ; in, For head loss along the route, The roughness coefficient is the Hezen-Williams roughness coefficient. For the flow rate of the pipe section, For the length of the pipe section, This refers to the diameter of the pipe section.

[0097] Local head loss Calculated using the following formula: ; in, This is the local drag coefficient. The velocity inside the pipe (unit: m / s) This is the acceleration due to gravity. This model is used to calculate the pressure loss as water flows through each pipe section.

[0098] Next, a valve flow model is established. The relationship between valve opening and flow rate adopts the equal percentage flow characteristic: ,in, The current flow rate of the valve. This represents the maximum flow rate when the valve is fully open. This refers to the valve's adjustable ratio (usually 30 or 50). The valve opening is represented by a value ranging from 0 to 1, where 0 indicates fully closed and 1 indicates fully open. The valve closing process is simulated as the opening linearly decreasing to 0 from its current value. The closing time is preset according to the valve type and actuation method; typically, the closing time for a solenoid valve is 5 to 10 seconds, and for an electric valve, it is 30 to 60 seconds. This model is used to determine the flow rate through the valve based on its opening.

[0099] Next, a flow distribution model is established. Based on the node connection relationships in the pipeline network topology model, node flow balance equations and loop pressure drop balance equations are established. The node flow balance equation represents the total flow into the node. Equal to the total flow of the outflowing node The loop pressure drop balance equation is the algebraic sum of the total head loss of each pipe section. Equal to the algebraic sum of the pump heads within the loop The total head loss is the sum of friction head loss and local head loss, both calculated using a hydraulic calculation model. The Newton-Raphson iterative method is used to solve the hydraulic equations of the pipe network. Initial values ​​for each node are set to 50% of the inlet pressure. The convergence criterion is set to an absolute value of the pressure correction less than 0.001 m, and the maximum number of iterations is set to 100. If the iteration fails to converge, the damping factor method or a relaxation of the convergence criterion is used for recalculation. This model is used to solve for the pressure at each node and the flow rate of each pipe segment in the pipe network under given valve opening conditions.

[0100] The calling process of each model in the simulation environment is as follows: At each decision step, the simulation environment receives the valve opening adjustment amount and the closing / starting command output by the agent, and updates the current valve opening of each valve. Update the valve opening. Input the valve flow model and calculate the theoretical flow rate of each valve (i.e., the above). The theoretical flow rates of each valve are used as boundary conditions and input into the pipe network topology model. Based on the node connection relationships and pipe segment parameters stored in the model, all pipe segments are traversed, and the hydraulic calculation model is called for each pipe segment to calculate its head loss along the pipe. and local head loss Then, the node flow balance equations and loop pressure drop balance equations in the flow distribution model are invoked, and the Newton-Raphson iterative method is used to solve for the pressure distribution and flow allocation of the entire pipeline network, obtaining the actual inflow rate of each irrigation unit and the total flow and pressure at the group inlet. The updated remaining water demand, remaining fertilizer demand, actual inflow rate, pressure, and other states are returned to the agent. The update rule for the remaining water demand is: Remaining water demand = Remaining water demand Actual inflow rate × decision step duration. The update rule for the remaining fertilizer requirement is: Remaining fertilizer requirement = Remaining fertilizer requirement Actual inflow rate × decision step duration × fertilizer concentration, where the fertilizer concentration is determined by the planned water-to-fertilizer ratio and the current fertilizer injection ratio.

[0101] This simulation environment simulates the changes in flow and pressure within a pipeline during valve opening and closing. Specifically, a larger valve opening results in a larger flow rate, but also consumes more pressure. When multiple valves are open simultaneously, the total flow rate is limited by the total water supply capacity of the group's inlet. If all valves are opened very wide simultaneously, the inlet pressure will drop, causing the actual flow rate of each valve to be less than the design value. When a valve begins to close, its flow rate gradually decreases, and the released flow capacity is distributed to other valves that remain open, causing their pressure and flow to increase. The agent can offset this increase by reducing the opening of other valves to maintain a stable irrigation rate for each irrigation unit. A simplified hydraulic simulator is embedded in the simulation environment, calculating the flow distribution once per second, ensuring that the training results reflect the real physical process.

[0102] S420. In the simulation environment, define the observation state, action space, and reward function of the agent; The observed status includes the remaining water demand, remaining fertilizer demand, current opening degree of each valve, actual inflow rate of each irrigation unit, and total flow rate and pressure at the group inlet for each irrigation unit. The action space includes the opening adjustment amount of each valve and whether to initiate valve closure; The reward function includes a positive reward for completing irrigation on time and a negative penalty for deviations in irrigation volume.

[0103] Specifically, positive rewards include: a positive reward value is given when all irrigation units complete irrigation on time within the time window and the irrigation uniformity meets the standard, awarded only once at the end of the round; a positive reward value is also given for each completed irrigation unit without causing other irrigation units to be under-irrigated or over-irrigated during the shutdown process. Negative penalties include: points are deducted proportionally when the actual irrigation volume of any irrigation unit deviates from the required water volume beyond the allowable range; a one-time negative penalty is given if an agent closes the valve of an irrigation unit before it has reached the required water volume, resulting in under-irrigation; and a negative penalty is given based on the delay time if an agent fails to close the valve of an irrigation unit after it has reached the required water volume, resulting in over-irrigation. The above reward and penalty values ​​can be set according to actual needs. For example, a reward of 100 points is given when all irrigation units complete irrigation on time and the irrigation uniformity meets the standard, 10 points are given for each completed irrigation unit without disturbance, 5 points are deducted for every 1% deviation in actual irrigation volume, 30 points are deducted for under-irrigation, and 5 points are deducted for every minute of delay in over-irrigation. The reward for each step is equal to the sum of positive rewards minus the sum of negative penalties.

[0104] S430. The agent is trained in the simulation environment using a reinforcement learning algorithm to obtain the trained agent.

[0105] An intelligent agent consists of a policy network and a value network. The policy network outputs actions based on observed states, while the value network evaluates the value of states to aid in the training of the policy network. After training, only the policy network is used for reasoning and decision-making.

[0106] The group-level agent is trained using the action-space proximal policy optimization (PPO) algorithm. The agent collects experience through interaction with the environment and stores it in a priority experience replay pool. The policy network is continuously updated until the water-filling uniformity reaches a certain level for several consecutive rounds and the average reward no longer increases, resulting in the trained agent. The specific implementation is as follows: First, initialize the policy network parameters. and value network parameters The policy network outputs the probability distribution of actions, while the value network evaluates the value of states. Both the policy and value networks employ neural network structures, and the specific network parameters can be set according to actual needs. For the action space, the policy network outputs the mean and standard deviation of a Gaussian distribution for sampling continuous actions, i.e., the opening adjustment amount, and simultaneously outputs a categorical probability distribution for discrete actions, i.e., the selection of whether to initiate valve closing.

[0107] Secondly, set the hyperparameters of the PPO algorithm, such as the learning rate and discount factor. Generalized Advantage Estimation (GAE) parameters Cutting parameters Update round E, batch size B, experience replay pool capacity M, etc.

[0108] Then, refer to Figure 4 As shown, the following training loop is executed: Step 1: Interact with the environment to gather experience. (In the current strategy) Below, the agent starts from the initial observation state. Beginning, at each decision step Based on the observed state Sample actions from the action distribution output by the policy network. Execute actions Afterwards, the environment returns a reward. Next observation state . Empirical tuples Store the experience in the replay pool. Repeat this process until the end of the round or the replay pool is full.

[0109] In this invention, Refers to the intelligent agent in the decision-making step The sensed environmental information specifically includes: the remaining water demand of each irrigation unit in the group, the remaining fertilizer demand of each irrigation unit in the group, the current opening degree of each valve in the group, the actual inflow rate of each irrigation unit in the group, and the total flow rate and pressure at the group inlet.

[0110] Refers to the intelligent agent in the decision-making step The output control commands specifically include: the opening adjustment amount of each valve in the group and whether to initiate valve closure. The opening adjustment amount is a continuous value, representing the increase or decrease of each valve based on its current opening; whether to initiate valve closure is a discrete selection, indicating whether to select one valve from multiple valves to execute the closure procedure or not to initiate closure.

[0111] Refers to the intelligent agent in the decision-making step Execute action Subsequently, the environment calculates and returns an immediate reward value based on the reward function. The reward function includes a positive reward for completing irrigation on time and a negative penalty for deviations in irrigation volume. Specifically, a positive reward is given when all irrigation units complete irrigation on time within the time window and the irrigation uniformity meets the standard; a positive reward is also given when each irrigation unit completes irrigation and the shutdown process does not cause other irrigation units to be under-irrigated or over-irrigated. When any deviation between the actual irrigation volume and the required water volume exceeds the allowable range, a negative penalty is accumulated proportionally; when an agent closes its valve before an irrigation unit has reached its required water volume, resulting in under-irrigation, a one-time negative penalty is given; when an agent fails to close its valve after an irrigation unit has reached its required water volume, resulting in over-irrigation, a negative penalty is given based on the delay time.

[0112] Refers to the actions performed by the intelligent agent After a decision-making step interval, the environment updates the remaining water demand, remaining fertilizer demand, current valve opening, actual inflow rate of each irrigation unit, and total inlet flow and pressure of the group based on the hydraulic calculation model, valve flow model, and flow distribution model, forming the observation state for the next decision step. Specifically, the remaining water demand decreases based on the actual inflow rate, the remaining fertilizer demand decreases based on the actual inflow rate and fertilizer concentration, the valve opening is updated based on the opening adjustment amount and closing procedure, and the actual inflow rate and total inlet flow and pressure of the group are obtained through the flow distribution model.

[0113] It should be noted that, Figure 4 The numbers 11-46 are marked in the grid, with each grid representing an irrigation unit.

[0114] Step 2: For each empirical tuple, calculate the advantage function estimate using the GAE method. and discount cumulative returns The specific calculation method can be found in existing technologies for those skilled in the art, and will not be repeated here. The formula for calculating GAE is: , , in, , For value network to state , The evaluation value. For TD error, For value network to state The evaluation value. This is the decision step at the end of the round. It is the reward value returned by the environment after the action is performed in this decision step. , From the decision-making step The cumulative discount reward from the start to the end of the round. As a discount factor, Indicates the current decision-making step The initial offset, At that time, it corresponds to the last time step T. For decision-making steps TD error.

[0115] Step 3: Sample small batches of experience from the experience replay pool according to priority, with the priority set to... ,in To prevent zero priority for small constants, sampling yields a mini-batch set containing B empirical tuples, each empirical tuple including ( ) and their corresponding priorities and pre-calculated advantage function estimates Discount cumulative rewards .

[0116] Step 4: Calculate the strategy loss : ; in, As the current strategy, The strategy before the update is as follows. For the parameterized action space, a Gaussian strategy is used for the continuous action part, and a classification strategy is used for the discrete action part. The proportions of each are calculated separately. The action probability distribution output by the policy network represents the action to be taken under the current policy. The probability of. To take action under the previous strategy The probability, These are the trimming parameters. It expresses expectation.

[0117] Step 5: Calculate the value loss .

[0118] Step 6: Calculate entropy loss .

[0119] Step 7: Calculate the total loss. ,in, and These are weighting parameters, which can be set according to actual needs, such as... =0.5, =0.01. Update the policy network parameters using the Adam optimizer. and value network parameters .

[0120] Step 8: Repeat steps 1 to 7, for each sample The network is updated every step, and this process is repeated for E rounds.

[0121] Step 9: When the irrigation uniformity meets the standard for several consecutive rounds and the average reward tends to stabilize, stop training and obtain the trained agent.

[0122] Irrigation uniformity This refers to the degree of consistency in irrigation across all irrigation units within the entire irrigation coordination group.

[0123] ; in, For irrigation units The actual amount of irrigation water, For irrigation units The water demand.

[0124] Irrigation uniformity meets the standard when the deviation between the actual irrigation volume and the water demand of all irrigation units is within the allowable range, i.e., the irrigation uniformity is greater than the first preset threshold. The actual irrigation volume is calculated by the simulation environment, and the water demand is the initial water demand in the digital prescription map. After each round, the sum of the rewards for all decision steps in that round is calculated as the total reward for that round. The average reward is the value obtained by averaging the total reward of each round over the most recent rounds. The average reward tends to stabilize when the average reward of the most recent rounds changes less than the average reward of the previous rounds, which is less than the second preset threshold.

[0125] This invention treats each irrigation coordination group as an intelligent agent, using the remaining water demand, remaining fertilizer demand, current valve opening, actual inflow rate of each irrigation unit, and total inflow and pressure at the group inlet as inputs for real-time observation. The agent outputs valve opening adjustment amounts and closing / starting commands based on the real-time observation status. The opening adjustment amount cannot exceed 10% of the current opening in a single adjustment, and the closing / starting command selects one action from either not starting a valve closure or starting a specific valve closure procedure. A sequence of coordinated scheduling commands within each irrigation coordination group is obtained based on the control commands. During training, a simulation environment incorporating the hydraulic dynamics of the pipeline network is constructed. This simulation environment simulates the physical relationship between valve opening, flow rate, and pressure, including dynamic processes such as total flow rate being limited by inlet water supply capacity and flow redistribution when valves close. The simulation environment defines the agent's observation state, action space, and reward function. The reward function includes a positive reward for timely irrigation and a negative penalty for irrigation deviation. The penalty function sets different deduction rules for under-irrigation and over-irrigation. A reinforcement learning algorithm is used to train the agent in a simulation environment. Through experience replay and policy network updates, the process continues until the irrigation uniformity meets the target and the average reward no longer increases. Through this training, the agent learns the optimal valve regulation strategy, ensuring that the opening and closing sequence of each actuator in the group meets the hydraulic coupling constraints of the pipeline network, smoothing pipeline pressure fluctuations, and shortening the overall irrigation time within the group.

[0126] This invention treats each irrigation coordination group as an intelligent agent, using the remaining water and fertilizer requirements of each irrigation unit within the group, the current opening degree of each valve, the actual inflow rate of each irrigation unit, and the total flow rate and pressure at the group inlet as inputs for real-time observation. This allows the agent to comprehensively perceive the water volume, fertilizer quantity, valve status, and hydraulic conditions within the group. Based on the real-time observation status, the agent outputs valve opening adjustment amounts and closing / starting commands, directly outputting valve control commands without intermediate conversion. The control commands generate a sequence of coordinated scheduling commands within each irrigation coordination group, achieving a direct mapping from decision-making to execution. During training, a simulation environment incorporating the hydraulic dynamics of the pipe network is constructed. Within this environment, the agent's observation state, action space, and reward function are defined. The reward function includes a positive reward for timely irrigation and a negative penalty for irrigation volume deviation, enabling the agent to balance irrigation efficiency and irrigation accuracy during the learning process. A reinforcement learning algorithm was used to train an agent in a simulation environment. Through continuous interaction with the simulation environment, the agent learned to coordinate the opening and closing timing and opening degree of all valves within the group while satisfying hydraulic constraints. This smoothed out pressure fluctuations in the pipeline network and shortened the overall irrigation time within the group, resulting in the trained agent. Through this training, the agent can directly output optimal valve control commands based on real-time observations in the actual irrigation system, without human intervention, achieving autonomous and coordinated valve opening and closing control within the group.

[0127] This invention determines the water shortage diagnosis and nutrient deficiency of each pixel based on crop status data, soil status data, atmospheric status data, and future rolling weather forecast data. These values ​​are then aggregated to form the water shortage diagnosis and nutrient deficiency of each irrigation unit. Based on this, a water-fertilizer coupling variable prescription map is generated, incorporating water shortage diagnosis, nutrient deficiency, digital prescription map, and future rolling weather forecast data. This method is driven by real-time sensing data and rolling weather forecasts, dynamically generating the water-fertilizer coupling variable prescription map at each decision point, rather than relying on static historical experience or fixed-cycle plans, thus achieving real-time response to dynamic environments.

[0128] This invention decomposes the optimization process into three closely linked levels. First, based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, the water and fertilizer requirements of each irrigation unit are determined, generating a digital prescription map. Then, based on the water shortage diagnosis results, nutrient deficiency, digital prescription map, and future rolling weather forecast data, a water-fertilizer coupling variable prescription map is generated. Second, based on the water-fertilizer coupling variable prescription map, all irrigation units are divided into multiple irrigation coordination groups that meet the pump station flow constraints and operating time window constraints. This allows irrigation units with similar water and fertilizer requirements and spatial proximity to be grouped together, greatly reducing the number of fertilizer applicator formula switching times. Finally, for each irrigation coordination group, a group-wide collaborative sorting decision is executed to obtain a collaborative scheduling instruction sequence within each irrigation coordination group. This ensures that the start and stop times of each actuator within the group meet the hydraulic coupling constraints of the pipeline network, smoothing pipeline network pressure fluctuations and shortening the overall irrigation time within the group. Through the sequential connection of these three levels, this invention integrates water and fertilizer demand optimization and field execution scheduling into a unified framework, achieving a balance between water and fertilizer resource optimization and execution efficiency.

[0129] This invention dynamically divides irrigation coordination groups based on a water-fertilizer coupling variable prescription map. The division is based on water and fertilizer demand information from the prescription map, rather than fixed geographical areas or manually designated rotation irrigation groups. This allows the grouping results to be automatically adjusted according to the real-time water and fertilizer needs of each irrigation unit. Simultaneously, the irrigation coordination groups satisfy pump station flow constraints and operating time window constraints, ensuring the grouping results are physically executable. By integrating a dynamic grouping logic based on a comprehensive water and fertilizer demand urgency index, the system can automatically identify and prioritize high-urgency units without manual intervention in rotation irrigation group division and valve opening / closing plans, achieving dynamic division and automatic scheduling of irrigation coordination groups.

[0130] The variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation device provided by the present invention is described below. The variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation device described below can be referred to in correspondence with the variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method described above.

[0131] The variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation device provided by this invention refers to... Figure 5 As shown, it includes the following modules: The pixel-level diagnostic module 510 is used to determine the water shortage diagnosis result and nutrient deficiency amount of each pixel based on crop status data, soil status data and atmospheric status data of farmland. The unit diagnosis aggregation module 520 is used to aggregate the water shortage diagnosis results and nutrient deficiency of each pixel into a representative value at the minimum irrigation unit level, so as to obtain the water shortage diagnosis results and nutrient deficiency of each irrigation unit. The first diagram generation module 530 is used to determine the water and fertilizer requirements of each irrigation unit based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, and generate a digital prescription diagram. The second graph generation module 540 is used to generate a water-fertilizer coupling variable prescription map based on the water shortage diagnosis results, the nutrient deficiency amount, the digital prescription map, and the weather forecast data in the future rolling time domain; the water-fertilizer coupling variable prescription map contains the water and fertilizer demand information of each irrigation unit in the future rolling time domain. The dynamic grouping module 550 is used to divide all irrigation units into multiple irrigation coordination groups according to the water-fertilizer coupling variable prescription diagram; the irrigation coordination groups satisfy the pump station flow constraints and the operation time window constraints. The collaborative sorting module 560 is used to perform intra-group collaborative sorting decisions for each irrigation collaborative group to obtain a collaborative scheduling instruction sequence within each irrigation collaborative group.

[0132] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include: a processor 610, a communication interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communication interface 620, and the memory 630 communicate with each other through the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method, which includes: Based on crop status data, soil status data, and atmospheric status data of farmland, determine the water shortage diagnosis result and nutrient deficiency amount for each pixel; The water shortage diagnosis results and nutrient deficiency of each pixel are aggregated into a representative value at the smallest irrigation unit level to obtain the water shortage diagnosis results and nutrient deficiency of each irrigation unit. Based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, the water and fertilizer requirements of each irrigation unit are determined, and a digital prescription map is generated. Based on the water shortage diagnosis results, the nutrient deficiency, the digital prescription map, and the weather forecast data in the future rolling time domain, a water and fertilizer coupling variable prescription map is generated; the water and fertilizer coupling variable prescription map contains the water and fertilizer demand information of each irrigation unit in the future rolling time domain. Based on the water-fertilizer coupling variable prescription diagram, all irrigation units are divided into multiple irrigation coordination groups; the irrigation coordination groups satisfy the pump station flow constraints and the operation time window constraints. For each irrigation coordination group, an intra-group coordination sorting decision is executed to obtain a sequence of coordination scheduling instructions within each irrigation coordination group.

[0133] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0134] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the variable irrigation and fertilization coordinated decision-making and autonomous rotation irrigation method provided by the above methods, the method comprising: Based on crop status data, soil status data, and atmospheric status data of farmland, determine the water shortage diagnosis result and nutrient deficiency amount for each pixel; The water shortage diagnosis results and nutrient deficiency of each pixel are aggregated into a representative value at the smallest irrigation unit level to obtain the water shortage diagnosis results and nutrient deficiency of each irrigation unit. Based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, the water and fertilizer requirements of each irrigation unit are determined, and a digital prescription map is generated. Based on the water shortage diagnosis results, the nutrient deficiency, the digital prescription map, and the weather forecast data in the future rolling time domain, a water and fertilizer coupling variable prescription map is generated; the water and fertilizer coupling variable prescription map contains the water and fertilizer demand information of each irrigation unit in the future rolling time domain. Based on the water-fertilizer coupling variable prescription diagram, all irrigation units are divided into multiple irrigation coordination groups; the irrigation coordination groups satisfy the pump station flow constraints and the operation time window constraints. For each irrigation coordination group, an intra-group coordination sorting decision is executed to obtain a sequence of coordination scheduling instructions within each irrigation coordination group.

[0135] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0136] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0137] 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 of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for coordinated decision-making and autonomous rotational irrigation based on variable irrigation and fertilization, characterized in that, include: Based on crop status data, soil status data, and atmospheric status data of farmland, determine the water shortage diagnosis result and nutrient deficiency amount for each pixel; The water shortage diagnosis results and nutrient deficiency of each pixel are aggregated into a representative value at the smallest irrigation unit level to obtain the water shortage diagnosis results and nutrient deficiency of each irrigation unit. Based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, the water and fertilizer requirements of each irrigation unit are determined, and a digital prescription map is generated. Based on the water shortage diagnosis results, the nutrient deficiency, the digital prescription map, and the weather forecast data in the future rolling time domain, a water and fertilizer coupling variable prescription map is generated; the water and fertilizer coupling variable prescription map contains the water and fertilizer demand information of each irrigation unit in the future rolling time domain. Based on the water-fertilizer coupling variable prescription diagram, all irrigation units are divided into multiple irrigation coordination groups; the irrigation coordination groups satisfy the pump station flow constraints and the operation time window constraints. For each irrigation coordination group, an intra-group coordination sorting decision is executed to obtain a sequence of coordination scheduling instructions within each irrigation coordination group.

2. The variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method according to claim 1, characterized in that, The process of determining the water deficiency diagnosis result and nutrient deficiency amount for each pixel based on crop status data, soil status data, and atmospheric status data includes: Based on the crop status data, the soil status data, and the atmospheric status data, the crop water stress index, canopy nitrogen accumulation, and normalized vegetation index of each pixel are obtained by remote sensing inversion. Based on the crop water stress index, the canopy nitrogen accumulation, and the normalized vegetation index, calculate the water shortage diagnosis result and nutrient deficiency for each pixel.

3. The variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method according to claim 1, characterized in that, The process of aggregating the water shortage diagnosis results and nutrient deficiency amounts of each pixel into a representative value at the minimum irrigation unit level, to obtain the water shortage diagnosis results and nutrient deficiency amounts of each irrigation unit, includes: According to the field pipeline network division rules, multiple pixels are divided into multiple irrigation units, and each irrigation unit corresponds to the area where an electric valve is located. For each irrigation unit, outlier removal is performed on the water shortage diagnosis results of the pixels within the irrigation unit, and the water shortage diagnosis results of the remaining pixels after removal are aggregated to obtain the water shortage diagnosis result of the irrigation unit. For each irrigation unit, the nutrient deficiency of the pixels within the irrigation unit is aggregated to obtain the nutrient deficiency of the irrigation unit.

4. The variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method according to claim 1, characterized in that, The process of determining the water and fertilizer requirements of each irrigation unit based on the water shortage diagnosis results and nutrient deficiency amounts, and generating a digital prescription map, includes: Based on the water shortage diagnosis results of the irrigation unit, the area of ​​the irrigation unit, and the crop coefficient, the water requirement of the irrigation unit is determined. The fertilizer requirement of the irrigation unit is determined based on the nutrient deficiency of the irrigation unit, the area of ​​the irrigation unit, and the conversion coefficient.

5. The variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method according to claim 1, characterized in that, Based on the water shortage diagnosis results, the nutrient deficiency amount, the digital prescription map, and future rolling weather forecast data, a water-fertilizer coupling variable prescription map is generated, including: Based on the weather forecast data within the future rolling time domain, determine the predicted reference crop evapotranspiration and effective rainfall for each day within the future rolling time domain; For each irrigation unit, the net irrigation water requirement for each day in the future rolling time domain is determined based on the water shortage diagnosis results of the irrigation unit, the predicted reference crop evapotranspiration, and the effective rainfall. For each irrigation unit, the planned water-fertilizer ratio for each day in the future rolling time domain is determined based on the net irrigation water requirement of the irrigation unit and the fertilizer requirement of the irrigation unit in the digital prescription diagram. For each irrigation unit, the comprehensive water and fertilizer demand urgency index of the irrigation unit is determined based on the water shortage diagnosis results, nutrient deficiency, predicted reference crop evapotranspiration, and effective rainfall. The net irrigation water requirement, planned water-fertilizer ratio, and comprehensive water-fertilizer demand urgency index of each irrigation unit are used as the water-fertilizer demand information in the prescription map of the water-fertilizer coupling variable.

6. The variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method according to claim 1, characterized in that, Based on the aforementioned water-fertilizer coupling variable prescription diagram, all irrigation units are divided into multiple irrigation coordination groups, including: Based on the water and fertilizer coupling variable prescription map, the water and fertilizer demand characteristics and spatial location information of each irrigation unit are extracted; the water and fertilizer demand characteristics include the comprehensive urgency index of water and fertilizer demand of each irrigation unit and the planned water and fertilizer ratio; Estimate the initial cluster number K based on the total water demand of all irrigation units and the rated water supply of the pumping station; The K irrigation units with the highest comprehensive urgency index of water and fertilizer demand were selected as the initial cluster centers. For each unassigned irrigation unit, the distance between the irrigation unit and each cluster center is calculated based on the planned water-fertilizer ratio and the spatial location information. Based on the distance, the irrigation units are sequentially assigned to the irrigation coordination groups where each cluster center is located; Each time an allocation attempt is made, it is determined whether the allocated irrigation coordination group satisfies the pump station flow constraint and the operation time window constraint. If the pump station flow constraint and the operation time window constraint are satisfied, the allocation is successful. If none of the irrigation coordination groups containing all K cluster centers satisfy the pump station flow constraint and the operation time window constraint, the number of irrigation coordination groups K is increased and the partitioning process is restarted. After each round of allocation, for each irrigation coordination group, the irrigation unit with the smallest total distance to other irrigation units in the group is selected as the new cluster center; Repeat the steps of assigning and updating cluster centers until the cluster centers no longer change, resulting in K irrigation co-groups.

7. The variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method according to claim 1, characterized in that, For each irrigation coordination group, an intra-group coordination ranking decision is executed to obtain a sequence of coordination scheduling instructions within each irrigation coordination group, including: Each irrigation coordination group is treated as an intelligent agent; the input of the intelligent agent is the real-time observation status, which includes the remaining water demand, remaining fertilizer demand, current opening degree of each valve, actual inflow rate of each irrigation unit, and total flow rate and pressure at the group inlet; The intelligent agent outputs control commands for each valve based on the real-time observation status. The control commands include opening adjustment amount and closing / starting commands. The sequence of coordinated scheduling instructions within each irrigation coordination group is obtained based on the control instructions; The intelligent agent is trained in the following way: Construct a simulation environment that includes the hydraulic dynamics of the pipe network; In the simulation environment, the observation state, action space, and reward function of the agent are defined; The observed status includes the remaining water demand, remaining fertilizer demand, current opening degree of each valve, actual inflow rate of each irrigation unit, and total flow rate and pressure at the group inlet for each irrigation unit. The action space includes the opening adjustment amount of each valve and whether to initiate valve closure; The reward function includes a positive reward for completing irrigation on time and a negative penalty for deviations in irrigation volume; The agent is trained in the simulation environment using a reinforcement learning algorithm to obtain the trained agent.

8. A variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation device, characterized in that, include: The pixel-level diagnostic module is used to determine the water shortage diagnosis result and nutrient deficiency amount of each pixel based on crop status data, soil status data and atmospheric status data of farmland. The unit diagnosis aggregation module is used to aggregate the water shortage diagnosis results and nutrient deficiency of each pixel into a representative value at the smallest irrigation unit level, so as to obtain the water shortage diagnosis results and nutrient deficiency of each irrigation unit. The first diagram generation module is used to determine the water and fertilizer requirements of each irrigation unit based on the water shortage diagnosis results and nutrient deficiency of each irrigation unit, and generate a digital prescription diagram. The second graph generation module is used to generate a water-fertilizer coupling variable prescription map based on the water shortage diagnosis results, the nutrient deficiency amount, the digital prescription map, and the weather forecast data in the future rolling time domain; the water-fertilizer coupling variable prescription map contains the water and fertilizer demand information of each irrigation unit in the future rolling time domain. The dynamic grouping module is used to divide all irrigation units into multiple irrigation coordination groups according to the water-fertilizer coupling variable prescription diagram; the irrigation coordination groups satisfy the pump station flow constraints and the operation time window constraints. The collaborative sorting module is used to perform intra-group collaborative sorting decisions for each irrigation collaborative group to obtain a collaborative scheduling instruction sequence within each irrigation collaborative group.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method as described in any one of claims 1 to 6.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the variable irrigation and fertilization collaborative decision-making and autonomous rotation irrigation method as described in any one of claims 1 to 6.

Images