Intelligent irrigation dynamic decision-making method and device

Abstract

The present invention relates to the field of agricultural irrigation technology, specifically to a method and device for intelligent irrigation dynamic decision-making. By integrating multi-source remote sensing data with precipitation data and combining it with the water requirements of crops at different growth stages, a method for intelligent irrigation fuzzy reasoning and dynamic decision-making is constructed. This method, based on regularly updated remote sensing data, dynamically monitors crop moisture status and generates plot-level irrigation decision recommendations. Based on these recommendations, managers can combine local water resource conditions and irrigation schedules to scientifically formulate irrigation strategies, achieve precise irrigation control, and improve irrigation efficiency and scientific decision-making.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural irrigation, and in particular to an intelligent irrigation dynamic decision-making method and equipment. Background Art

[0002] Intelligent irrigation decision-making technology is crucial for improving water resource utilization efficiency, ensuring crop growth, and increasing agricultural yields. However, current intelligent irrigation decision-making technology has many limitations.

[0003] Current intelligent irrigation decision-making technologies primarily include: Irrigation decision-making based on soil moisture data: This uses sensors to monitor soil moisture changes and combines this with crop water demand patterns to generate irrigation decision results. Irrigation decision-making based on crop growth models and weather forecasts: This uses crop growth models to simulate the water requirements and growth status of crops at different stages, and combines this with meteorological data to generate irrigation decision results. Irrigation decision-making based on remote sensing data: This uses the spectral information of drone remote sensing images combined with measured moisture data to identify water-deficit areas and generate irrigation decisions. Furthermore, a single index (such as NDVI) is calculated based on satellite remote sensing data, and combined with a surface energy balance model to predict water consumption and generate irrigation decisions.

[0004] However, the above intelligent irrigation decision-making technologies generally have the following problems:

[0005] Poor adaptability: Irrigation decision-making technologies based on soil moisture data, whose sensor-monitored soil moisture data only reflects single-point conditions and is not representative of overall field moisture conditions. Different soil types vary significantly in water retention and permeability, and fields vary widely in spatial scale. This technology cannot dynamically adapt to these changes. Irrigation decision-making technologies based on crop growth models and weather forecasts utilize complex crop growth models and require numerous input parameters. However, some parameters are difficult to accurately obtain across different regions and planting conditions, and parameter calibration is challenging, leading to numerous uncertainties and poor adaptability to diverse environments. Irrigation decision-making technologies based on remote sensing data, however, use drone-based remote sensing data that is susceptible to environmental interference. For example, windy and rainy weather can affect drone flight stability and data collection quality, leading to increased data processing complexity. When satellite remote sensing data is combined with evapotranspiration models, both scales rely on ground-based data. This makes them less adaptable given the varying terrain, climate, and other conditions across regions, limiting their scope of application.

[0006] Poor accuracy and stability: Irrigation decision-making technologies based on soil moisture data rely on sensor accuracy, but sensors are easily affected by multiple factors such as temperature and soil texture. For example, in high-temperature environments, sensor errors may occur, resulting in deviations in the acquired soil moisture data, which in turn leads to erroneous irrigation decisions. Irrigation decision-making technologies based on crop growth models and weather forecasts rely on weather data and the accuracy of parameter calibration. However, weather forecast data itself is uncertain, and the complexity of model parameterization can also introduce cumulative errors, ultimately leading to erroneous irrigation decisions. Irrigation decision-making technologies based on remote sensing data, at the drone scale, rely on high-resolution remote sensing imagery, but weather factors can severely impact data quality, such as cloud cover causing loss of image information. At the satellite scale, a single index combined with a surface energy balance model is often used. This fails to address data lag issues, limiting real-time performance and potentially leading to incorrect irrigation decisions.

[0007] Difficulty in Popularization: Most existing technologies require soil sensors, drones, or model development and optimization. These devices and software require regular maintenance and updates, which consumes significant manpower and material resources and requires specialized technicians to operate. Model development also requires significant research and debugging, which increases implementation costs and limits the widespread adoption of these technologies. Summary of the Invention

[0008] In view of this, the purpose of the present invention is to provide an intelligent irrigation dynamic decision-making method and device to solve the problems of poor adaptability, poor accuracy and stability, and difficulty in promotion of intelligent irrigation decision-making in the existing technology.

[0009] According to a first aspect of an embodiment of the present invention, a method for dynamic decision-making in intelligent irrigation is provided, comprising:

[0010] Get the target plot boundary, and obtain the Sentinel-2 L2A remote sensing data of the target plot for the current date and the same period in history based on the target plot boundary;

[0011] generating a vector mask file according to the sentinel-2 L2A remote sensing data, removing data of interference pixels in the sentinel-2 L2A remote sensing data according to the vector mask file, and calculating the normalized vegetation index of the target plot using the denoised sentinel-2 L2A remote sensing data;

[0012] Obtain the Sentinel-3 Level-2 LST data and precipitation data corresponding to the Sentinel-2 L2A remote sensing data for the same period in history; reproject the Sentinel-3 Level-2 LST data to the coordinate system corresponding to the Sentinel-2 L2A remote sensing data; resample the Sentinel-3 Level-2 LST data and precipitation data to the preset resolution;

[0013] The moisture state index of each pixel is calculated based on the denoised Sentinel-2 L2A remote sensing data of the current date and the same period in history. The vegetation state index of each pixel is calculated based on the normalized vegetation index of the current date and the same period in history. The temperature state index of each pixel is calculated based on the surface temperature data in the Sentinel-3 Level-2 LST data of the current date and the same period in history.

[0014] According to the crop growth period on the current date, the water state index linguistic variable, vegetation state index linguistic variable and temperature state index linguistic variable of each pixel are obtained;

[0015] According to the linguistic variables corresponding to the water state index, vegetation state index and temperature state index, the irrigation decision level of each pixel is obtained;

[0016] The irrigation decision of the target plot is generated according to the irrigation decision level of all pixels in the target plot; the decision prompt is generated according to the past precipitation data of the current date and the future predicted precipitation data.

[0017] Preferably, before acquiring the sentinel-2 L2A remote sensing data for the same period in history, the method further includes: downloading the sentinel-2 L2A remote sensing data for 2 days selected from a preset historical time period during the peak crop growth period each year;

[0018] After calculating the normalized vegetation index of the target plot based on the downloaded sentinel-2 L2A remote sensing data, the method further includes: if the normalized vegetation index of the target plot is greater than or equal to the planting determination threshold, acquiring the sentinel-2 L2A remote sensing data for the remaining days of the year.

[0019] Preferably, after acquiring the Sentinel-2 L2A remote sensing data, the method further includes:

[0020] The cloud, snow, and ice coverage ratio is calculated based on the SLC band data of the Sentinel-2 L2A remote sensing data. If the cloud, snow, and ice coverage ratio is less than or equal to the preset threshold, the Sentinel-2 L2A remote sensing data is marked as valid data; otherwise, it is considered invalid data.

[0021] Preferably, the water state index of each pixel is calculated based on the denoised Sentinel-2 L2A remote sensing data of the current date and the same period in history, including:

[0022] The normalized difference water index (NDWI) of each pixel is calculated based on the near-infrared and short-wave infrared reflectances of each pixel in the denoised Sentinel-2 L2A remote sensing data of the current date and the same period in history.

[0023] The moisture state index of each pixel is calculated based on the normalized difference moisture index of each pixel on the current date and the maximum and minimum values ​​of the normalized difference moisture index in the same period in history.

[0024] Preferably, the vegetation state index of each pixel is calculated based on the normalized vegetation index of the current date and the same period in history, including:

[0025] The vegetation status index of each pixel is calculated based on the normalized vegetation index of each pixel on the current date and the maximum and minimum values ​​of the normalized vegetation index in the same historical period.

[0026] Preferably, the temperature state index of each pixel is calculated based on the surface temperature data of the current date and the historical same period, including:

[0027] The temperature state index of each pixel is calculated based on the surface temperature data of each pixel on the current date and the maximum and minimum surface temperature data in the same historical period.

[0028] Preferably, generating a decision prompt includes:

[0029] If the irrigation decision generated for the target plot is that irrigation is required, then based on the precipitation data, it is determined whether effective precipitation occurs within a preset time period before and after the current date. If so, after the irrigation decision is generated, a decision prompt is generated that no irrigation is required due to the occurrence of precipitation.

[0030] Preferably, generating a decision prompt further includes:

[0031] Get the irrigation operation data that the user has performed;

[0032] If the irrigation decision generated for the target plot is that irrigation is required, it is determined based on the irrigation operation data whether an irrigation operation has been performed within a preset time period before the current date. If so, after the irrigation decision is generated, a decision prompt is generated indicating that irrigation is not required because the irrigation operation has been performed.

[0033] Preferably, the method further comprises:

[0034] If there is no effective precipitation within the preset time period before and after the current date, the predicted precipitation data for the next three days based on the current date is retrieved from the meteorological forecast database, and whether there is effective precipitation is determined based on the predicted precipitation data; if there is effective precipitation in the next three days, a decision prompt is generated that no irrigation is required due to the expected future precipitation.

[0035] According to a second aspect of an embodiment of the present invention, there is provided an intelligent irrigation dynamic decision-making device, comprising:

[0036] A main controller, and a memory connected to the main controller;

[0037] The memory stores program instructions;

[0038] The main controller is used to execute program instructions stored in the memory and perform any of the above methods.

[0039] The technical solutions provided by the embodiments of the present invention may have the following beneficial effects:

[0040] It is understandable that the technical solution shown in the present invention integrates multi-source data, introduces fuzzy reasoning, and provides flexible output through the input of multiple variables. It is more suitable for agricultural scenarios under different technical and environmental conditions and has stronger generalization capabilities. In the face of different growth periods of crops, this technical solution is more in line with the growth laws of crops by accurately designing the language variables and membership functions in fuzzy reasoning. It can better handle the uncertainty and complexity of water demand. At the same time, it constructs dynamic rules and uses precipitation data to correct the impact of remote sensing data lag on irrigation decisions in real time, thereby improving the dynamics and accuracy of irrigation decisions. Relying on remote sensing satellite data and precipitation data, it intelligently provides plot-level irrigation decision recommendations without the need for sensor deployment, manual sampling, and regular equipment calibration, significantly reducing the cost of hardware investment and long-term operation and maintenance. In addition, managers can flexibly adjust according to actual water resource conditions and irrigation systems, enhancing the scientific nature and flexibility of irrigation decisions and making them easier to promote and use.

[0041] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0043] Figure 1 This is a schematic diagram showing the steps of a smart irrigation dynamic decision-making method according to an exemplary embodiment;

[0044] Figure 2is a flowchart of data acquisition and preprocessing according to an exemplary embodiment;

[0045] Figure 3 The figure is a flow chart of a dynamic decision-making system for intelligent irrigation of corn fields according to an exemplary embodiment. DETAILED DESCRIPTION

[0046] Exemplary embodiments will be described in detail herein, examples of which are illustrated in the accompanying drawings. In the following description, when referring to the drawings, like numbers in different figures represent like or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present invention. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present invention, as detailed in the appended claims.

[0047] In one embodiment, Figure 1 This is a schematic diagram of a method for dynamic decision-making in intelligent irrigation according to an exemplary embodiment. Figure 1 , provides an intelligent irrigation dynamic decision-making method, including:

[0048] Step S11: Obtain the boundary of the target plot, and obtain the Sentinel-2 L2A remote sensing data of the target plot on the current date and the same period in history based on the boundary of the target plot.

[0049] Step S12: Generate a vector mask file based on the sentinel-2 L2A remote sensing data, remove data of interfering pixels in the sentinel-2 L2A remote sensing data based on the vector mask file, and calculate the normalized vegetation index of the target plot using the denoised sentinel-2 L2A remote sensing data.

[0050] Step S13: Acquire Sentinel-3 Level-2 LST data and precipitation data corresponding to Sentinel-2 L2A remote sensing data for the same period in history; reproject the Sentinel-3 Level-2 LST data to the coordinate system corresponding to the Sentinel-2 L2A remote sensing data; and resample the Sentinel-3 Level-2 LST data and precipitation data to a preset resolution.

[0051] Step S14: Calculate the water state index of each pixel based on the denoised Sentinel-2 L2A remote sensing data of the current date and the same period in history; calculate the vegetation state index of each pixel based on the normalized vegetation index of the current date and the same period in history; and calculate the temperature state index of each pixel based on the surface temperature data in the Sentinel-3 Level-2 LST data of the current date and the same period in history.

[0052] Step S15: According to the crop growth period on the current date, the water state index linguistic variable, the vegetation state index linguistic variable and the temperature state index linguistic variable of each pixel are obtained.

[0053] Step S16: derive the irrigation decision level of each pixel according to the linguistic variables corresponding to the water state index, the vegetation state index, and the temperature state index.

[0054] Step S17: Generate an irrigation decision for the target plot based on the irrigation decision levels of all pixels in the target plot; and generate a decision prompt based on the past precipitation data of the current date and the future predicted precipitation data.

[0055] It is understandable that the technical solution shown in the present invention integrates multi-source data, introduces fuzzy reasoning, and provides flexible output through the input of multiple variables. It is more suitable for agricultural scenarios under different technical and environmental conditions and has stronger generalization capabilities. In the face of different growth periods of crops, this technical solution is more in line with the growth laws of crops by accurately designing the language variables and membership functions in fuzzy reasoning. It can better handle the uncertainty and complexity of water demand. At the same time, it constructs dynamic rules and uses precipitation data to correct the impact of remote sensing data lag on irrigation decisions in real time, thereby improving the dynamics and accuracy of irrigation decisions. Relying on remote sensing satellite data and precipitation data, it intelligently provides plot-level irrigation decision recommendations without the need for sensor deployment, manual sampling, and regular equipment calibration, significantly reducing the cost of hardware investment and long-term operation and maintenance. In addition, managers can flexibly adjust according to actual water resource conditions and irrigation systems, enhancing the scientific nature and flexibility of irrigation decisions and making them easier to promote and use.

[0056] In specific practice, the technical solution shown in this embodiment can be divided into three stages: data acquisition and preprocessing, key feature extraction, and intelligent irrigation fuzzy reasoning decision generation.

[0057] Data acquisition and preprocessing are reflected in steps S11, S12 and S13. In a preferred embodiment, the specific process of data acquisition and preprocessing is shown in FIG. Figure 2 .

[0058] First, we determine the target plot's boundaries and then retrieve Sentinel-2 L2A remote sensing data for the current date based on the target plot's boundaries. We then need to determine whether the Sentinel-2 L2A data is valid based on the percentage of cloud, snow, and ice coverage.

[0059] It should be noted that the SLC band data (scene classification map) is obtained, and the cloud, snow, and ice coverage area ratio is calculated based on the SLC band data of the Sentinel-2 L2A remote sensing data. If the cloud, snow, and ice coverage area ratio is less than or equal to the preset threshold, the Sentinel-2 L2A remote sensing data is marked as valid data; otherwise, it is considered invalid data.

[0060] Calculate the cloud, snow / ice cover percentage (CSI) using the following formula:

[0061]

[0062]

[0063]

[0064] in, is the percentage of cloud, snow / ice coverage area in the corn field, unit is %; is the number of pixels marked as clouds in the SCL scene classification map; is the number of pixels marked as snow or ice in SCL; is the total number of pixels in the corn field.

[0065] The validity of Sentinel-2 L2A remote sensing data is determined by comparing the cloud, snow, and ice cover ratio (CSI) with a preset threshold.

[0066] Assuming the preset threshold is 50%, the judgment conditions are as follows:

[0067] If the CSI is ≤50%, the Sentinel-2 L2A remote sensing data for the cornfield is retained and marked as valid. A vector mask file, Plot_No_CSI, is generated for denoising and subsequent index calculations. Vector mask files define spatial extents using geometric shapes such as points, lines, and polygons. When processing remote sensing data related to cornfields, a polygonal vector mask is constructed based on the cornfield boundaries to precisely delineate the target area. Furthermore, the vector mask file marks pixels with valid data and sets pixels covered by cloud, snow, or ice as invisible / rejected. The vector mask file is overlaid with the Sentinel-2 L2A remote sensing data to achieve data filtering and denoising.

[0068] If CSI>50%, it means that the land parcel is greatly affected by clouds, snow / ice and does not meet the requirements of subsequent calculations. In this case, the calculation is terminated at this point and no subsequent steps are performed. The remote sensing data for this period is marked as invalid data.

[0069] When the remote sensing data for the current date is valid, extract the date of the Sentinel-2 L2A remote sensing data, that is, the year, month, and day.

[0070] Then, obtain the Sentinel-2 L2A remote sensing data for the same period in history. It should be noted that, first, select two days of Sentinel-2 L2A remote sensing data from the preset historical time period (which can be the past six years) during the peak crop growth period of each year and download them.

[0071] Assuming the current date is 2024, obtain Sentinel-2 L2A remote sensing data for two days from the past six years (2018-2023) during the peak crop growth period (for example, corn, the peak growth period is July 1st to July 31st). Similarly, the validity of the Sentinel-2 L2A remote sensing data needs to be determined using the cloud, snow, or ice cover percentage (CSI).

[0072] If it is judged to be valid, the Sentinel-2 L2A remote sensing data is combined with the vector mask file to calculate the normalized vegetation index of the plot after removing the interference pixels. The calculation formula is as follows:

[0073]

[0074] in, is the normalized vegetation index of the corn plot; and are the near-infrared and red band reflectances of the i-th pixel (derived from sentinel-2 L2A remote sensing data). is the total number of valid pixels in the corn plot (invalid pixels are excluded according to the vector mask file Plot_No_CSI).

[0075] If the normalized vegetation index of the target plot If the value is greater than or equal to the planting threshold, Sentinel-2 L2A remote sensing data for the same period of the year will be obtained. If historical data for the same date does not exist or is unavailable during the acquisition process, the remaining available Sentinel-2 L2A remote sensing data within 5 days will be queried daily to serve as the historical Sentinel-2 L2A remote sensing data for the same period.

[0076] For example, assuming that the planting judgment threshold is 0.35, if the sentinel-2 L2A remote sensing data of two days in 2018 are selected If both values ​​are below the set planting threshold of 0.35, it means that corn was not planted on the plot in 2018, and further download of Sentinel-2 and Sentinel-3 data for 2018 will not be performed. Conversely, if the Sentinel-2 L2A remote sensing data from two days in 2019 are both above the set planting threshold of 0.35, it means that corn was planted on the plot in 2019, and further download of remote sensing data for the same period of that year can be performed.

[0077] Preferably, a loop can be used to determine whether corn was planted each year from 2018 to 2023. The years in which corn was planted are saved in a planting year statistical table. Sentinel-2 L2A remote sensing data for the same period is then obtained based on the valid historical years saved in the planting year statistical table. For each acquired Sentinel-2 L2A remote sensing data set, its validity is determined by the cloud, snow, or ice cover (CSI) percentage, and invalid data is discarded.

[0078] It is understandable that by designing a non-planted year detection function to filter, only the planting year data is downloaded, invalid data processing is avoided to optimize computing resources, and SCL is used to remove noise, thereby improving data accuracy and processing efficiency.

[0079] After the above processing, the Sentinel-3 level-2 LST data and precipitation data for the preset historical time period are downloaded. The three data sources need to be processed to make them consistent in time and space to achieve multi-source remote sensing data fusion. Specifically, the Sentinel-3 level-2 LST data is reprojected to a coordinate system consistent with the Sentinel-2 L2A remote sensing data; and the Sentinel-3 level-2 LST data and precipitation data are resampled to the preset resolution (for example, 10m).

[0080] After data acquisition and preprocessing, the key feature extraction stage is carried out.

[0081] Key feature extraction is to calculate the water state index, vegetation state index and temperature state index based on the acquired sentinel-2 L2A remote sensing data and Sentinel-3 level-2LST data.

[0082] It should be noted that the water status index (WCI) is mainly calculated through the normalized difference moisture index (NDMI) and is used to assess the moisture content of corn. The calculation of the water status index for each pixel includes: calculating the normalized difference moisture index for each pixel based on the near-infrared band reflectance and short-wave infrared band reflectance of each pixel in the denoised Sentinel-2 L2A remote sensing data of the current date and the same period in history; and calculating the water status index of each pixel based on the normalized difference moisture index of each pixel on the current date and the maximum and minimum values ​​of the normalized difference moisture index in the same period in history. The calculation formula is as follows:

[0083] 00

[0084]

[0085] in, is the normalized difference moisture index of the pixel on this date in the current year; 、 They are respectively the pixel's historical period Minimum and maximum values; is the near-infrared band reflectivity of the pixel; is the shortwave infrared band reflectivity of the pixel. is the water status index of the pixel.

[0086] It should be noted that the vegetation condition index (VCI) mainly represents the relative measurement of corn growth status. The higher the VCI value, the better the vegetation growth status. The vegetation condition index of each pixel is calculated, including:

[0087] The vegetation status index of each pixel is calculated based on the normalized vegetation index of each pixel on the current date and the maximum and minimum values ​​of the normalized vegetation index in the same period in history. The calculation formula is as follows:

[0088] 00

[0089]

[0090] in, is the normalized vegetation index of the pixel on this date in the current year; 、 They are respectively the pixel's historical period Minimum and maximum values. is the vegetation status index of the pixel.

[0091] It should be noted that the Temperature State Index (LSTCI) reflects the degree of temperature stress on corn by comparing the difference between the current year's surface temperature on that date and the historical temperature data for the same period. The temperature state index of each pixel is calculated, including:

[0092] The temperature state index of each pixel is calculated based on the current date's surface temperature data and the maximum and minimum values ​​of the surface temperature data during the same period in history. The calculation formula is as follows:

[0093] 00

[0094]

[0095] in, The surface temperature of the pixel on the date of the current year, in degrees Celsius. You need to first convert the surface temperature in the Sentinel-3level-2 LST data to Convert to degrees Celsius; 、 They are respectively the pixel's historical period Minimum and maximum values, in °C. is the temperature state index of the pixel.

[0096] After the key feature extraction is completed, the intelligent irrigation fuzzy reasoning decision-making stage is entered, which includes steps S15, S16 and S17.

[0097] In step S15, the water state index linguistic variable, the vegetation state index linguistic variable and the temperature state index linguistic variable of each pixel are obtained according to the growth period of the crops on the current date.

[0098] Preferably, the water state index linguistic variable, vegetation state index linguistic variable and temperature state index linguistic variable of each pixel are obtained by comparing with the pre-built linguistic variable database in combination with the growth period of crops on the current date.

[0099] In specific practice, taking corn as an example, by designing an intelligent irrigation fuzzy inference system, the linguistic variables and membership functions of the input variables in different corn growth stages are constructed, and a linguistic variable database is constructed to obtain the water state index linguistic variables, vegetation state index linguistic variables and temperature state index linguistic variables of each pixel.

[0100] Corn has varying water requirements at different stages of growth. During the seedling stage from sowing to jointing, the plant is small and its overall water requirement is low. Care should be taken to avoid excessive watering during this stage, as this can hinder seedling growth. From jointing to tasseling, growth enters its most vigorous phase, with gradually rising temperatures and a faster evaporation rate, increasing overall water consumption. From tasseling to grain filling, water requirements are higher. During this period, metabolism is active, and water demand is high. Only adequate water can meet growth and development needs, making it a critical stage for yield formation. During the grain filling and maturity phase, when the kernels are essentially set, water consumption decreases.

[0101] Based on the water requirement characteristics of corn throughout its growth period, corresponding linguistic variables were constructed for three key indices, namely, the water status index (WCI), the vegetation status index (VCI), and the temperature status index (LSTCI). Trigonometric functions were selected as the membership functions. The comparison of the linguistic variable database is shown in Table 1.

[0102] Table 1 Language variable comparison table

[0103]

[0104] Since the present embodiment uses a trigonometric function as the membership function, the trigonometric membership function needs to have an intersection in range.

[0105] By constructing the linguistic variables and membership functions of the input variables in different corn growth stages, combining the growth stage of crops on the current date, as well as the calculated moisture state index, vegetation state index and temperature state index on the current date, the corresponding linguistic variables can be obtained.

[0106] It is understandable that a growth period-oriented fuzzy rule library is constructed for different growth stages of corn (sowing-jointing, jointing-tasting, tasting-filling, filling-maturity). Through dynamic rule design combined with crop growth laws, the irrigation decision results of the dynamic decision model for intelligent irrigation of corn plots are more scientific and adaptable.

[0107] Step S16: derive the irrigation decision level of each pixel according to the linguistic variables corresponding to the water state index, the vegetation state index, and the temperature state index.

[0108] In practice, a fuzzy-reasoning-based rule base for corn irrigation decision-making was designed. Based on measured soil volumetric moisture content (0-60 cm) in the root zone across multiple regions throughout the corn growth period and combined with expert experience, 27 fuzzy rules were constructed as part of the corn irrigation decision-making hierarchy (see Table 2). For example, if the linguistic variable for the water state index is low, the linguistic variable for the temperature state index is low, and the linguistic variable for the vegetation state index is low, then the irrigation decision level is high.

[0109] Table 2 Corn irrigation decision-making level rule library

[0110]

[0111] By constructing a corn irrigation decision-making level rule library, the irrigation decision level of each pixel can be obtained based on the linguistic variables corresponding to the moisture state index, vegetation state index and temperature state index.

[0112] Step S17: Generate an irrigation decision for the target plot based on the irrigation decision levels of all pixels in the target plot.

[0113] In practice, pixel-level fuzzy reasoning and plot-level decision aggregation are performed. Using the Mamdani model, the current date's growth period and key characteristics are used as input variables. Combined with the "Irrigation Decision Level Rule Library," a fuzzy set of irrigation decision levels is derived. This is then defuzzified using the centroid method to obtain the defuzzified value for each pixel. This value is then mapped to the corresponding irrigation decision category based on a predefined range of values. Based on this, the percentage of pixels within a plot with the irrigation decision category "needing irrigation" is counted. If this percentage exceeds 60%, the plot is marked as "needing irrigation"; if it is 30% or less and is less than 60%, the plot is marked as "basically satisfied"; and if it is less than 30%, the plot is marked as "not requiring irrigation."

[0114] At this point, staff are able to use the constructed intelligent irrigation dynamic decision-making method to generate irrigation decisions.

[0115] In practical applications, the intelligent irrigation dynamic decision-making method shown in this embodiment can be integrated into an intelligent irrigation dynamic decision-making system for user convenience.

[0116] See also Figure 3 The specific implementation process of the dynamic decision-making of intelligent irrigation for corn plots is as follows:

[0117] Before starting the dynamic decision-making system for intelligent irrigation of corn fields, the required user data should be obtained in advance, including the date when corn planting starts in the field, the field boundary information (latitude and longitude), the user's irrigation farming operation time, and the expected harvest date.

[0118] Upon system startup, the system first compares the user-entered corn planting start date with the current date. If the user entered data before planting and the current date is before the planting date, the system pauses and automatically restarts when the current date matches the entered planting date. If the current date is on or after the planting date, the model operates normally. The system then retrieves the required databases, including the corn growing period database and meteorological forecast data. Based on the current date and plot boundary information (latitude and longitude), the system retrieves information from the corn growing period database and performs a matching operation to obtain the actual corn growing period for the current date. The system then enters the "Data Acquisition and Preprocessing Phase." After this phase completes, it checks for valid data output. If no data is available, it indicates that Sentinel-2 L2A product data may not be available that day (due to satellite revisit cycles) or that cloud cover does not meet the required data requirements, making the data unusable. The system then terminates and restarts the next day. If data is available, the system proceeds to the "Key Feature Extraction Phase." The extracted features and growing period information are input into the "Intelligent Irrigation Fuzzy Inference Phase," where fuzzy reasoning is performed to generate irrigation decisions.

[0119] After generating the irrigation decision for the target plot, a decision prompt is generated based on the past precipitation data of the current date and the future predicted precipitation data.

[0120] If the irrigation decision output by the intelligent irrigation dynamic decision-making system is "basically satisfied", the final decision prompt output to the user is "Under current conditions, the corn in this plot is growing well and does not require irrigation for the time being. Continuous monitoring is recommended."

[0121] If the irrigation decision output by the "intelligent irrigation fuzzy reasoning system" is "no irrigation required", the final decision prompt output to the user is "at the current stage of the land, the corn growth is in a state of sufficient water and no irrigation measures are required."

[0122] It should be noted that generating a decision prompt includes: if the irrigation decision generated for the target plot is that irrigation is required, then based on the precipitation data, determining whether effective precipitation occurs within a preset time period before and after the current date; if so, after generating the irrigation decision, generating a decision prompt that no irrigation is required due to the occurrence of precipitation.

[0123] Generating a decision prompt also includes: obtaining irrigation operation data that the user has performed; if the irrigation decision generated for the target plot is that irrigation is required, determining whether an irrigation operation has been performed within a preset time period before the current date based on the irrigation operation data; if so, after generating the irrigation decision, generating a decision prompt indicating that irrigation is not required because the irrigation operation has been performed.

[0124] If there is no effective precipitation within the preset time period before and after the current date, the predicted precipitation data for the next three days based on the current date is retrieved from the meteorological forecast database, and whether there is effective precipitation is determined based on the predicted precipitation data; if there is effective precipitation in the next three days, a decision prompt is generated that no irrigation is required due to the expected future precipitation.

[0125] In practice, if the "intelligent irrigation fuzzy inference system" outputs "irrigation required," precipitation data processed during the "data acquisition and preprocessing phase" is retrieved. Because remote sensing satellite data products have a certain lag in acquisition, precipitation data between two dates must be checked. If significant precipitation occurs (i.e., daily precipitation >3 mm) or the user performs an irrigation operation, the fuzzy inference result is intelligently adjusted, ultimately outputting a decision prompt to the user: "The plot was detected as requiring irrigation on YYYY-MM-DD (satellite acquisition date), but due to subsequent precipitation or irrigation operations, continued monitoring is recommended, and irrigation is not currently required." If there is no effective precipitation, the forecast precipitation data for the next three days based on the current date will be retrieved from the meteorological forecast database. If there is no effective precipitation in the forecast three days and the user has not implemented irrigation operations, the final decision prompt will be output to the user: "It is recommended to start irrigation on YYYY-MM-DD_1 (i.e., the current date)." Otherwise, the user will be prompted that "it was monitored on YYYY-MM-DD that the land needs irrigation, but precipitation is expected in the next three days. It is recommended to continue observation and no irrigation is required for the time being."

[0126] The corn field intelligent irrigation dynamic decision-making system is executed once a day to dynamically output irrigation decision recommendations at the field scale until the harvest date, at which time the system is stopped.

[0127] This embodiment constructs a dynamic, multi-level, and multi-scale decision-making logic, aggregates pixel-level fuzzy reasoning to plot-level decisions, and corrects the lag of remote sensing data. It combines historical precipitation, weather forecasts, and multi-dimensional conditional judgments of user irrigation operations to generate personalized suggestions, realizing the integrated optimization of dynamic decision-making and user feedback, improving the real-time and accuracy of irrigation decisions, and helping users achieve accurate and efficient irrigation optimization.

[0128] Compared with the prior art, the technical solution shown in the present invention is:

[0129] Greater generalization: Existing technologies rely on ground sensors, drones, field data, or crop growth models. The accuracy and representativeness of these data sources, coupled with the complexity of these models, make them difficult to adapt to diverse environmental conditions. This invention, however, integrates multi-source data to avoid reliance on a single data source. Furthermore, it incorporates a fuzzy inference system. Compared to traditional methods that rely on fixed thresholds, this system uses multiple variables to generate flexible outputs, making it more applicable to agricultural scenarios under diverse technical and environmental conditions and possessing greater generalization capabilities.

[0130] Higher Precision: Existing irrigation decision-making methods are mostly generalized and fail to consider the water requirements of specific crops at different growth stages. This invention addresses the different growth stages of corn by precisely designing the linguistic variables and membership functions within the fuzzy inference system, which better aligns with crop growth patterns. The introduction of a fuzzy inference system better addresses the uncertainty and complexity of water requirements. It also establishes dynamic rules to correct for the impact of remote sensing data lag on irrigation decisions in real time, improving the dynamism and accuracy of irrigation decisions.

[0131] Stronger application and promotion: This invention relies on remote sensing satellite data and meteorological data, and adopts a dynamic decision-making model for intelligent irrigation of corn plots to intelligently provide plot-level irrigation decision-making suggestions. It does not require sensor deployment, manual sampling, and regular equipment calibration, which significantly reduces hardware investment and long-term operation and maintenance costs. In addition, managers can flexibly adjust according to actual water resource conditions and irrigation systems, which enhances the scientific nature and flexibility of irrigation decisions and is easier to promote and use.

[0132] In another embodiment, a smart irrigation dynamic decision-making device is provided, comprising:

[0133] A main controller, and a memory connected to the main controller;

[0134] The memory stores program instructions;

[0135] The main controller is used to execute program instructions stored in the memory and perform any of the above methods.

[0136] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0137] It should be noted that, in the description of the present invention, the terms "first", "second", etc. are used for descriptive purposes only and should not be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "plurality" is at least two.

[0138] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention pertain.

[0139] It should be understood that various components of the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods may be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof may be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0140] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0141] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.

[0142] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0143] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0144] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. An intelligent irrigation dynamic decision-making method, characterized in that: include: Get the target plot boundary, and obtain the Sentinel-2 L2A remote sensing data of the target plot for the current date and the same period in history based on the target plot boundary; generating a vector mask file according to the sentinel-2 L2A remote sensing data, removing data of interference pixels in the sentinel-2 L2A remote sensing data according to the vector mask file, and calculating the normalized vegetation index of the target plot using the denoised sentinel-2 L2A remote sensing data; Obtain the Sentinel-3 Level-2 LST data and precipitation data corresponding to the Sentinel-2 L2A remote sensing data for the same period in history; reproject the Sentinel-3 Level-2 LST data to the coordinate system corresponding to the Sentinel-2 L2A remote sensing data; resample the Sentinel-3 Level-2 LST data and precipitation data to the preset resolution; The moisture state index of each pixel is calculated based on the denoised Sentinel-2 L2A remote sensing data of the current date and the same period in history. The vegetation state index of each pixel is calculated based on the normalized vegetation index of the current date and the same period in history. The temperature state index of each pixel is calculated based on the surface temperature data in the Sentinel-3 Level-2 LST data of the current date and the same period in history. Derivatives of a water state index linguistic variable, a vegetation state index linguistic variable, and a temperature state index linguistic variable for each pixel based on the current crop growth period include: presetting different membership functions for mapping the water state index, vegetation state index, and temperature state index to corresponding linguistic variables for different crop growth periods; selecting a corresponding membership function based on the current crop growth period, and deriving the linguistic variable for each pixel based on the selected membership function and the index value of each pixel; According to the linguistic variables corresponding to the water state index, vegetation state index and temperature state index, the irrigation decision level of each pixel is obtained; The irrigation decision of the target plot is generated according to the irrigation decision level of all pixels in the target plot; the decision prompt is generated according to the past precipitation data of the current date and the future predicted precipitation data.

2. The method according to claim 1, characterized in that Before obtaining the historical Sentinel-2 L2A remote sensing data for the same period, it also includes: downloading Sentinel-2 L2A remote sensing data for 2 days selected from the preset historical time period during the peak crop growth period each year; After calculating the normalized vegetation index of the target plot based on the downloaded sentinel-2 L2A remote sensing data, the method further includes: if the normalized vegetation index of the target plot is greater than or equal to the planting determination threshold, acquiring the sentinel-2 L2A remote sensing data for the remaining days of the year.

3. The method according to claim 1, characterized in that After acquiring Sentinel-2 L2A remote sensing data, it also includes: The cloud, snow, and ice coverage ratio is calculated based on the SLC band data of the Sentinel-2 L2A remote sensing data. If the cloud, snow, and ice coverage ratio is less than or equal to the preset threshold, the Sentinel-2 L2A remote sensing data is marked as valid data; otherwise, it is considered invalid data.

4. The method according to claim 1, wherein Based on the denoised Sentinel-2 L2A remote sensing data for the current date and the same period in history, the water status index of each pixel is calculated, including: The normalized difference water index (NDWI) of each pixel is calculated based on the near-infrared and short-wave infrared reflectances of each pixel in the denoised Sentinel-2 L2A remote sensing data of the current date and the same period in history. The moisture state index of each pixel is calculated based on the normalized difference moisture index of each pixel on the current date and the maximum and minimum values ​​of the normalized difference moisture index in the same period in history.

5. The method according to claim 1, wherein Calculate the vegetation status index of each pixel based on the normalized vegetation index of the current date and the same period in history, including: The vegetation status index of each pixel is calculated based on the normalized vegetation index of each pixel on the current date and the maximum and minimum values ​​of the normalized vegetation index in the same historical period.

6. The method according to claim 1, characterized in that Calculate the temperature state index of each pixel based on the current date and historical surface temperature data of the same period, including: The temperature state index of each pixel is calculated based on the surface temperature data of each pixel on the current date and the maximum and minimum surface temperature data in the same historical period.

7. The method according to claim 1, characterized in that Generate decision prompts, including: If the irrigation decision generated for the target plot is that irrigation is required, then based on the precipitation data, it is determined whether effective precipitation occurs within a preset time period before the current date. If so, after the irrigation decision is generated, a decision prompt is generated that no irrigation is required due to the occurrence of precipitation.

8. The method according to claim 7, characterized in that Also includes: Get the irrigation operation data that the user has performed; If the irrigation decision generated for the target plot is that irrigation is required, it is determined based on the irrigation operation data whether an irrigation operation has been performed within a preset time period before the current date. If so, after the irrigation decision is generated, a decision prompt is generated indicating that irrigation is not required because the irrigation operation has been performed.

9. The method according to claim 7, characterized in that Also includes: If there is no effective precipitation within the preset time period before the current date, the predicted precipitation data for the next three days based on the current date is retrieved from the meteorological forecast database, and whether there is effective precipitation is determined based on the predicted precipitation data; if there is effective precipitation in the next three days, a decision prompt is generated that no irrigation is required due to the expected future precipitation.

10. An intelligent irrigation dynamic decision-making device, characterized in that: include: A main controller, and a memory connected to the main controller; The memory stores program instructions; The main controller is used to execute program instructions stored in the memory and perform the method according to any one of claims 1 to 9.

Images