Intelligent water conservancy monitoring method for irrigation area and related equipment thereof

Abstract

The invention relates to the technical field of water conservancy monitoring, and provides an intelligent water conservancy monitoring method for an irrigation area and related equipment thereof. Water body boundary extraction is carried out on remote sensing image data to obtain an initial water body boundary data set, time sequence interpolation processing is carried out on the initial water body boundary data set to obtain a water body change trend data set, and soil moisture content prediction deduction is carried out according to a satellite cloud picture and the water body change trend data set to generate a soil moisture content prediction data set. Performing water level and area fitting on the regional DEM according to the water body change trend data set to construct a water storage capacity model, and performing crop water demand calculation according to regional data and crop data to obtain a crop water demand matrix data set; and performing constrained programming according to the soil moisture content prediction data set, the crop water demand matrix data set and the water storage capacity model to obtain a multi-target irrigation scheduling decision set. According to the application, dynamic balance between irrigation requirements and water resource supply is realized through multi-level and multi-source data fusion and intelligent scheduling.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of water conservancy monitoring, in particular to an intelligent water conservancy monitoring method for an irrigation area and a related device thereof. BACKGROUND

[0002] In an irrigation area, the spatiotemporal distribution of water resources has a high degree of uncertainty, extreme weather such as drought and heavy rain occurs frequently, and the soil moisture content and crop water requirement change dramatically with the season and growth stage. Traditional manual patrol and regular irrigation cannot timely discover the dynamic changes of water level, water volume and soil moisture content, resulting in waste of water resources or stagnation of crops due to lack of water. Therefore, a water conservancy monitoring system based on remote sensing, ground sensing and intelligent scheduling can realize real-time sensing of water body range, water level change, soil moisture and weather information, and achieve dynamic balance and precise allocation of water resources supply and demand, which has important practical significance for improving the risk resistance capacity of drought and flood and ensuring stable grain output.

[0003] At present, water conservancy monitoring in an irrigation area mainly relies on regular inspection of water channels and reservoirs by manual labor, and water level sensors are arranged at key nodes to measure static water level and flow, and water release scheduling is performed in combination with historical experience; in some areas, soil moisture content is roughly calculated according to weather forecasts, and crop water requirement is estimated by manually measuring soil water content and combining with an empirical formula. These methods have low investment in equipment and labor cost, and can meet the needs of regular irrigation, but due to sparse monitoring points, lagging information update and lack of fusion of multi-source data, they can only be adjusted after the event, and it is difficult to accurately guide global irrigation. SUMMARY

[0004] Therefore, the present application provides an intelligent water conservancy monitoring method for an irrigation area and a related device thereof to solve the problem of poor timeliness of dynamic analysis of soil moisture content.

[0005] The first aspect of the present application provides an intelligent water conservancy monitoring method for an irrigation area, which comprises: extracting the water body boundary of the target irrigation area from the collected remote sensing image data to obtain an initial water area boundary data set, and performing time series interpolation processing on the initial water area boundary data set to obtain a water body change trend data set; performing soil moisture content prediction and deduction according to the real-time acquired satellite cloud image and the water body change trend data set to generate a soil moisture content prediction data set; performing water level and area fitting on a preset regional DEM according to the water body change trend data set to construct a water storage capacity model; performing crop water requirement calculation according to the real-time collected regional data and crop data of the target irrigation area to obtain a crop water requirement matrix data set; Constrained programming is performed according to the soil moisture prediction dataset, the crop water requirement matrix dataset, and the water storage capacity model to obtain a multi-objective irrigation scheduling decision set.

[0006] In an optional embodiment, the initial water area boundary dataset includes pixel area and vector element data; and the water body boundary extraction of the target irrigation area from the collected remote sensing image data to obtain the initial water area boundary dataset includes: Coordinate fusion and correction screening are performed on the collected remote sensing image data to obtain a multispectral image; Green band pixel value and near-infrared band pixel value extraction are performed on the multispectral image, and normalization calculation is performed on the green band pixel value and the near-infrared band pixel value to obtain an NDWI image; The NDWI image is binarized and segmented according to a preset empirical threshold to obtain a water body candidate area, and morphological opening operation and morphological closing operation are performed on the water body candidate area to obtain a binary water body image; Connected domain analysis and contour coordinate extraction conversion are performed on the binary water body image to obtain map coordinate sequences and pixel areas of each connected water body, and the map coordinate sequences are converted into vector element data by a preset vertex simplification algorithm.

[0007] In an optional embodiment, the time series interpolation processing of the initial water area boundary dataset to obtain a water body change trend dataset includes: The initial water area boundary dataset is grouped and sorted according to a preset water body ID and shooting time to obtain a polygon of each water body; Polar coordinate conversion is performed on each vertex in the polygon to obtain a polar coordinate sequence of each polygon, and an angle and radius association dataset of each water body at each shooting time is constructed according to the polar coordinate sequence; According to a preset sampling angle set and the angle and radius association dataset, linear sample interpolation processing is performed on the radius corresponding to each sampling angle at each time to obtain the radius corresponding to each sampling angle of each water body at each time; The polar coordinate sequence is updated according to the radius corresponding to each sampling angle at each time, and Cartesian coordinate system conversion is performed on the updated polar coordinate sequence to obtain a water body contour polygon of each water body at each time; The water body contour polygons are arranged in time sequence to obtain a continuity water body change trend dataset corresponding to each water body.

[0008] In an optional embodiment, the soil moisture prediction dataset is generated by performing soil moisture prediction deduction according to the real-time acquired satellite cloud image and the water body change trend dataset. projecting and cloud feature recognition extracting of the real-time acquired satellite cloud image according to a preset irrigation area coordinate system to obtain continuous regionalized cloud data; predicting cloud layer moving speed and direction of the continuous regionalized cloud data by a preset cloud field tracking algorithm to obtain cloud band prediction data of each grid in the irrigation area coordinate system; estimating precipitation of each plot in the target irrigation area in future time periods according to preset historical precipitation and cloud top temperature statistical data and the cloud band prediction data to obtain precipitation prediction data of each plot; carrying out irrigation area surface recharge prediction according to the water body change trend data set to obtain surface recharge prediction data; deducing soil layer water content of the target irrigation area according to the precipitation prediction data and the surface recharge prediction data to obtain predicted water content; obtaining corresponding field water holding capacity according to the current crop growth stage, and calculating water supplement state data according to the field water holding capacity and the predicted water content, to generate the soil moisture prediction data set according to the water supplement state data and the predicted water content.

[0009] In an optional implementation, the fitting water level and area according to the water body change trend data set to construct a water storage capacity model includes: rasterizing water body contour polygons in the water body change trend data set to obtain a binary mask at each time; performing per-pixel overlay and elevation value mode statistics on the binary mask and a preset regional DEM to obtain an average water level at each time; calculating a horizontal water area corresponding to each time according to the number of pixels of all water bodies in the binary mask and the area of a single pixel in the regional DEM; constructing an area-height correlation function according to the average water level and the horizontal water area by a preset polynomial fitting algorithm, to perform numerical integral operation on the area-height correlation function according to a preset water level threshold to obtain a volume-height correlation function; encapsulating the area-height correlation function and the volume-height correlation function to construct the water storage capacity model.

[0010] In an optional implementation, the regional data includes real-time vegetation index, original soil water content, soil temperature, environmental humidity, and weather element data, and the crop data includes crop type and current crop growth stage; the crop water requirement calculation according to the real-time collected regional data and crop data of the target irrigation area to obtain a crop water requirement matrix data set includes: correcting the original soil moisture content data according to the soil temperature through a preset temperature correction formula to obtain standardized soil moisture content data; calculating a reference evapotranspiration according to the standardized soil moisture content data, the environmental humidity, and the weather element data through a preset Penman-Monteith algorithm; obtaining an original crop coefficient according to a crop type and a current crop growth stage of the target irrigation area, and dynamically correcting the original crop coefficient according to the real-time vegetation index to obtain a real-time crop coefficient; calculating a water requirement depth per unit area according to the reference evapotranspiration and the real-time crop coefficient, and calculating a current water supplement amount according to a preset field water holding capacity, areas of each plot in the target irrigation area, the standardized soil moisture content data, and the water requirement depth per unit area to obtain a crop water requirement matrix data set.

[0011] In an optional implementation, the constraint programming according to the soil moisture condition prediction data set, the crop water requirement matrix data set, and the water storage capacity model to obtain a multi-objective irrigation scheduling decision set comprises: performing water requirement prediction and extraction according to the soil moisture condition prediction data set and the crop water requirement matrix data set to obtain water requirement data of a current period and a plurality of future periods; performing water storage amount mapping value calculation on the water storage capacity model to obtain available water resource data of the current period and the plurality of future periods; constructing a quadratic weighted optimization objective function of each plot in each period according to a preset plot weight coefficient and the water requirement data; performing multi-objective constraint solving on the quadratic weighted optimization objective function according to the water requirement data and the available water resource data to obtain a multi-objective irrigation scheduling decision set.

[0012] The second aspect of the application provides an intelligent water conservancy monitoring device for an irrigation area, the device comprising: a water area analysis module configured to extract a water body boundary of a target irrigation area from collected remote sensing image data to obtain an initial water area boundary data set, and perform time series interpolation processing on the initial water area boundary data set to obtain a water body change trend data set; a soil moisture condition prediction module configured to perform soil moisture condition prediction and deduction according to a real-time acquired satellite cloud image and the water body change trend data set to generate a soil moisture condition prediction data set; a water storage model module configured to perform water level and area fitting on a preset regional DEM according to the water body change trend data set to construct a water storage capacity model; a water demand measurement module configured to calculate crop water demand based on real-time collected regional data of the target irrigation area and crop data to obtain a crop water demand matrix dataset; an irrigation decision module configured to perform constraint programming based on the soil moisture prediction dataset, the crop water demand matrix dataset, and the water storage capacity model to obtain a multi-objective irrigation scheduling decision set.

[0013] The third aspect of the present application 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 implements the steps of the intelligent water conservancy monitoring method for an irrigation area as described above when executing the computer program.

[0014] The fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the steps of the intelligent water conservancy monitoring method for an irrigation area as described above.

[0015] In summary, the present application has at least the following beneficial technical effects: 1. By combining the collected remote sensing images with image segmentation and edge extraction algorithms, and performing time series interpolation processing, continuous water body change trend data can be generated, greatly improving the response speed and prediction accuracy of water body dynamic changes.

[0016] 2. By combining the real-time satellite cloud image with the water body change trend data, the soil moisture prediction can be deduced, the soil moisture change trend of the irrigation area under different weather scenarios can be estimated, the evolution of soil water content in the future period can be more accurately simulated, the soil moisture reference under multiple scenarios can be provided, and the prediction reliability can be improved.

[0017] 3. Through multi-objective optimization programming, multiple objectives such as "optimal overall soil moisture of the region", "maximization of water storage capacity", "crop water demand satisfaction", "water saving and energy saving", and "drainage and waterlogging risk" can be considered comprehensively to form a multi-objective irrigation scheduling decision set, which can take into account the interests of all parties, and can quickly re-plan in the event of sudden weather or water quantity mutation, ensuring the robustness and feasibility of the irrigation scheme. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1is a flow chart of an irrigation area intelligent water conservancy monitoring method provided by an embodiment of the present application; Figure 2 is a functional module diagram of an irrigation area intelligent water conservancy monitoring device provided by an embodiment of the present application; Figure 3 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0021] As shown in Figure 1 is a flow chart of an irrigation area intelligent water conservancy monitoring method provided by an embodiment of the present application. The irrigation area intelligent water conservancy monitoring method provided by the embodiment of the present application includes the following steps.

[0022] Step S1, performing water body boundary extraction on the collected remote sensing image data to obtain an initial water area boundary data set, and performing time series interpolation processing on the initial water area boundary data set to obtain a water body change trend data set.

[0023] It should be understood that the remote sensing image data in the embodiments of the present application can be divided into satellite remote sensing images and unmanned aerial vehicle remote sensing images according to the collection sources. In order to ensure the consistency of the images obtained by different sensors or at different times in spatial position and projection reference, firstly, the obtained remote sensing images need to be fused and corrected with the pre-defined geographic coordinate system of the target irrigation area. For example, if the oblique shooting image of the unmanned aerial vehicle and the satellite orthographic image appear to be offset in the same area, it is necessary to unify them to the latitude and longitude or projection coordinate system through control point registration and affine transformation, so as to form a multi-spectral layer with the same geographic reference. Then, in order to eliminate the influence of atmospheric scattering and absorption on different wave bands, so that the reflectivity obtained under different time and different climate conditions is comparable. The atmospheric correction algorithm is applied to the fused multi-spectral image to convert the digital quantization value of the image into the ground reflectivity. For example, the green light and near-infrared reflectivity obtained after correction of the image obtained in cloudy weather matches the measurement results under sunny conditions, which is convenient for subsequent water body identification.

[0024] Subsequently, according to the corrected multi-spectral image, pixel values of the green light band G(x, y) and the near-infrared band N(x, y) are extracted, so as to obtain a water body index (Normalized Difference Water Index, NDWI) by using a classical normalization calculation formula. Specifically, according to the corrected reflectivity G(x, y) and N(x, y) of each pixel (x, y) in the multi-spectral image in the green light and near-infrared bands, the water body index NDWI(x, y) of the pixel (x, y) is obtained by calculating the ratio of the difference between the green light band G(x, y) and the near-infrared band N(x, y) and the sum of the green light band G(x, y) and the near-infrared band N(x, y). The difference between the green light band G(x, y) and the near-infrared band N(x, y) is used to highlight the high reflection of water body to green light and the low reflection of water body to near-infrared, and the sum of the green light band G(x, y) and the near-infrared band N(x, y) ensures that the index is between [-1, 1]. For example, for the same pond area, the green light visibility is high above the water surface when shooting, and the near-infrared is strongly absorbed by water, and the typical NDWI(x, y) value can reach more than 0.7; on the contrary, the green light reflection is relatively weak in the dense vegetation area, and the near-infrared reflection is strong, and the NDWI(x, y) can be reduced to negative value. Therefore, by calculating the water body index of each pixel in the multi-spectral image, an NDWI image capable of significantly distinguishing water body and non-water body pixels is obtained.

[0025] After generating the NDWI image, it needs to be binarized to extract the water body candidate region. The binarization threshold τ is not simply fixed, but is dynamically determined in combination with the statistical characteristics of the multi-spectral image. That is, assuming μ NDWI is the mean of the entire NDWI image, σ NDWI is the standard deviation, and therefore τ is the sum between μ NDWI and ησ NDWI . Among them, the coefficient η is used to control whether the water body recognition is conservative or aggressive, for example, when η = 0.2, the threshold is slightly high, which can reduce the misjudgment of shadows; when η =-0.1, the threshold is low, which can take into account the details of the water body edge. The binarization threshold τ obtained in this way can automatically adapt to the optimal threshold in different scenes, avoiding inconsistent recognition results caused by artificially fixing the value. For example, in sunny images, the mean μ NDWI is high, and if the standard deviation σ NDWIDynamic adjustment is likely to misjudge the shoal as non-water body. Further, the pixel of NDWI(x, y) > τ is marked as water body candidate area, and the morphological open operation and then the morphological close operation are performed on the binary image, which can remove the noise spots smaller than the set kernel size and fill the holes in the water body. The morphological open operation first erodes to remove isolated pixels, and then dilates to restore large area shape, which effectively eliminates the false positive caused by shadow or reflection; the morphological close operation first dilates and fills small holes, and then erodes to eliminate boundary jitter, so as to make the water body edge smooth and continuous. For example, the shadow area of the stone on the river bank, the morphological open operation can remove the small spots therein, and the morphological close operation can repair the contour break caused by the vegetation shelter.

[0026] The binary water body image generated after morphological processing is labeled by using the connected domain analysis algorithm to analyze the neighborhood connected region, and the contour pixel sequence of each water body is extracted. The edge tracking method is used for contour extraction, and the coordinates of each point are recorded along the pixel chain code from a certain boundary starting point. Then the extracted pixel coordinates (x i ,y i ) are mapped to map coordinates (X i ,Y i ) through affine transformation. Specifically, in the process of converting pixel coordinates (x i ,y i ) to map coordinates (X i ,Y i ), the pixel coordinates recorded in the image are first taken as reference points in a local plane. In order to make the reference points correspond to a real position in the actual geographic coordinate system, linear scaling, rotation and / or offset operations are performed on the pixel coordinates, and finally the map coordinates consistent with the horizontal and latitude / longitude directions of the earth's surface are obtained. For example, if the pixel coordinates of a lake in the image are (2000, 1500), and the real map coordinates of the lake are (102000.5m, 231500.2m), the system will first transform (2000, 1500) to a value close to (102000.5m, 231500.2m) according to the scaling, rotation parameters calculated by all control points, and then accurately to the final result (102000.5, 231500.2) by adding a small translation of a few meters or a few hundred meters. Meanwhile, the total number of pixels P in each connected domain is counted, and the approximate planar projection area S of the water body can be obtained by taking the single-pixel area A P (e.g., 4 square meters) as an example. When a small water hole contains 500 pixels in the mask, and the single-pixel area is 4m 2 , the calculated area of the water hole is about 2000m 2 .

[0027] To reduce the subsequent storage and spatial analysis burden, a vertex simplification algorithm (e.g., Douglas-Peucker algorithm) is further applied to the sequence of map coordinates of each contour, and the polygon is compressed by a preset distance tolerance ε. For example, if ε = 1 m, the key inflection points are retained while the redundant vertices are removed, so that the number of vertices after simplification is only a small part of the original, but the deviation from the actual contour shape does not exceed 1 m. This process greatly reduces the data volume while ensuring the integrity of the contour geometric features, making it easier for GIS database management.

[0028] Finally, each simplified vector element data and its corresponding area value S are packaged to form an initial water boundary data set, wherein each record in the initial water boundary data set includes but is not limited to water body number, water body shooting timestamp, vectorized closed polygon vertex sequence, and area attribute. The initial water boundary data set can be directly used for subsequent time series interpolation, dynamic capacity calculation, and water quantity scheduling analysis, ensuring that the entire monitoring system has clear and traceable records of the boundary information of each water body in the irrigation area.

[0029] In the process of time series interpolation of the initial water boundary data set, each water body contour from different time nodes is first grouped and sorted according to its unique identifier and shooting time to ensure that subsequent operations can seamlessly connect the geometric information of the same water body at multiple time points. In fact, the core significance of grouping and sorting is to treat the contours of a water body at different times as a continuous curve evolution with time as the axis, similar to splicing multiple static pictures into a dynamic video in chronological order, so that smooth estimation and reconstruction can be made in the middle frame. For example, the contour (i.e., polygon) of a river section taken at 8 o'clock and the contour taken at 10 o'clock can be reasonably interpolated and inferred to the transition polygon shape between flooding and shrinking at 9 o'clock after sorting.

[0030] After grouping and sorting, the polygon contour at each time point is further converted to polar coordinates. The polar coordinate conversion takes the centroid (X c ,Y c ) of the polygon at the current time as the origin, and maps the Cartesian coordinates (X i ,Y i ) of each vertex on the polygon to the angle θ i and the radius r i . The angle θ i can be obtained by the arctangent function arctan2(Y i -Y c ,X i -X c ), and the radius r iis equal to the Euclidean distance from the vertex to the centroid. By converting the polar coordinates, the originally arbitrary shaped closed polygon is transformed into a radius function curve that oscillates with angle. Subsequently, a uniform set of sampling angles {θ k |k = 1, …, M} is defined in the polar coordinate domain, and each discrete polar coordinate sequence at each time is padded according to the preset angle. For example, the angles can be equally divided into every 2° or 5° to take a sample, so that θ k is an equally spaced sequence. For a vertex that does not exactly fall on these angle points at a certain time, linear interpolation is performed on the two adjacent known (θ i , r i ) to obtain the corresponding radius r(θ k , t k ) at each θ j . Specifically, assuming that the time points of two known curves are t j and t j+1 , at a fixed angle θ k , first, the radius value corresponding to time t i is obtained by linear interpolation according to the adjacent two vertices (θ i , r j (t i+1 )) and (θ i+1 , r j (t j )). Through linear interpolation calculation, no matter how the original contour is distributed, a simple straight line segment can be used to approximate the local curve on a uniform angle grid, thereby establishing a set of angle-radius mapping functions (i.e., angle and radius association data set), laying the foundation for shape alignment at different time points, while ensuring that interpolation does not produce self-intersection or discontinuity problems. After the above linear interpolation operation is performed on each time, a batch of {θ k , r raw (θ k , t j )} data points are obtained, corresponding to different times t j . If the radius at any time t ∈ [t j , t j+1 ] between two adjacent times is to be reconstructed, linear interpolation needs to be performed again in the time dimension. The discrete shapes at multiple time phases are straightened along the time dimension into differentiable line segments, thereby generating a radius value that changes continuously with time for each angle point.

[0031] After obtaining the interpolation radius r lim (θ k , t) corresponding to each sampling angle at each time, the polar coordinates (θ k , r lim (θ kt)) back to Cartesian coordinates, so that at any time t that we want to reconstruct, we can get the water body boundary polygon by connecting all the (X, Y) points in the angle grid k The (X, Y) coordinate points are calculated, and these points are sequentially connected to obtain a complete and seamless closed polygon contour (i.e., a water body contour polygon). By restoring the function image in the polar coordinate domain back to the actual map plane, the shape change and the accurate spatial position are combined.

[0032] Finally, the reconstructed contours at each time are indexed and encapsulated according to the shooting time or the interpolation time, and they are corresponded to the corresponding time stamps one by one and packaged as continuous water body change trend data sets. Each contour in the water body change trend data set is labeled with a time label. In this way, when water volume calculation or water level calculation is performed subsequently, the boundary shape at the corresponding time can be directly obtained by time query. For example, if the trend of the river width during 9:30 to 10:30 is to be verified, all the contours in this time period are read and secondary operation is performed, so that a smooth dynamic image can be obtained. Through the above dynamic interpolation processing, not only the contour at any intermediate time can be reasonably estimated outside the key time that has been shot, but also the entire curve evolution process is continuous and does not lose the original geometric characteristics, thereby laying a solid foundation for subsequent water level and water volume analysis.

[0033] Step S2, soil moisture condition prediction data sets are generated by performing soil moisture condition prediction and deduction according to the real-time acquired satellite cloud image and the water body change trend data sets.

[0034] Before implementing soil moisture prediction and extrapolation, it is necessary to first project cloud images from meteorological satellites onto a spatial reference frame consistent with the geographic coordinate system of the irrigation area. This ensures that subsequent gridded analysis based on latitude, longitude, or planar projection coordinates can correspond one-to-one with ground features. For example, if a cloud image from the Himawari-8 satellite uses geographic latitude and longitude projection in the original data, but the Digital Elevation Model (DEM) or water body contour data of the irrigation area uses UTM projection, then the cloud image must be remapped using an interpolation algorithm based on the regional reference ellipsoid to obtain the latitude, longitude, or planar coordinates corresponding to each pixel in the same coordinate system. This ensures that the spectral information of the cloud image can be fully utilized for overlaying with ground features, preventing distortion of prediction conclusions due to coordinate mismatch. After projection, the spatial distribution of clouds needs to be identified in each cloud image. Cloud feature extraction is usually accomplished by analyzing cloud top temperature in the infrared band and texture in the visible light band: lower cloud top temperatures correspond to higher cloud heights, and vice versa; visible light texture can reflect the contours and thickness differences of cloud boundaries. By first converting the image into a brightness temperature map, then performing threshold segmentation on the brightness temperature distribution, and combining local texture gradient operators to extract the edges of cloud bodies, continuous regionalized cloud data is finally obtained. For example, at the same time, all pixels are divided into two categories: "cloud bodies" and "non-cloud bodies". Then, adjacent pixels are merged according to the threshold neighborhood to achieve an accurate depiction of the cloud distribution within the entire irrigation area.

[0035] After obtaining continuous regionalized cloud maps, a cloud field tracking algorithm is applied to estimate the cloud movement speed and direction at adjacent time points. This is significant because it can predict the timing of rainfall in each grid cell within a short future period. The cloud field tracking algorithm here first uses local autocorrelation coefficients in the spatial domain to determine the movement direction of cloud patches, and then combines this with the gradient trend of cloud top temperature over time for vector fusion, forming the position of each grid point (i,j) at adjacent time points (t). k ,t k+1 The velocity vector between ) =(u i,j ,v i,j This is used to predict the location of cloud bands in the future. Where u i,j v i,j These represent the east-west and north-south movement speeds of the clouds, respectively. The data is based on motion vectors. Using the velocity-displacement formula, the cloud distribution location for each grid can be directly obtained at the next time step. For example, if a cloud mass is observed moving northeast at a speed of approximately 5 km / h at the current time, it can be predicted that the center of that grid will be covered by clouds again half an hour later. This information is particularly crucial for rainfall estimation, as precipitation is only possible when cloud bands cover the ground.

[0036] After obtaining the cloud band prediction data, it is necessary to combine it with the statistical relationship between historical precipitation and cloud top temperature, so as to estimate the rainfall of each plot in the future period, and obtain the precipitation prediction data of each plot. Among them, the precipitation prediction data includes but is not limited to precipitation probability and predicted precipitation. Specifically, based on the long-term accumulated cloud top temperature-precipitation intensity empirical data set, a nonlinear mapping function is used to map the predicted cloud top temperature T i,j (t) at the current grid to the corresponding precipitation probability P i,j (t) and the corresponding precipitation amount p i,j (t). It should be understood that the lower the cloud top temperature, the greater the water vapor content and the thicker the cloud layer, and thus the more likely it is to produce more rainfall. In order to put this physical judgment into specific numerical values, it is necessary to analyze a large number of historical observation data in advance to obtain an empirical curve: when the cloud top temperature decreases, the corresponding precipitation amount increases exponentially; and once the temperature exceeds a certain threshold, it is difficult to form obvious precipitation. Based on this law, a mapping mechanism can be constructed: first, take a reference value representing the maximum possible precipitation brought by the thickest cloud layer, and then let the corresponding precipitation amount increase significantly for every degree of cloud top temperature decrease; for example, if the cloud top temperature is 5°C lower than the reference threshold, the precipitation amount at that time may almost reach 90% of the historical maximum; if it is only 2°C lower, the precipitation amount may only reach half of the historical maximum. This exponential mapping can ensure that the precipitation amount increases rapidly in the early stage of temperature decrease, and tends to be flat when approaching the maximum value, avoiding the situation that the calculated value exceeds the physical limit. At the same time, the estimation of the precipitation probability is also processed as the second step: if the estimated precipitation amount of a certain grid point is already very considerable, the corresponding precipitation occurrence probability is close to 100%; on the contrary, if the estimated rainfall is small, the probability will be low. Finally, both a specific precipitation numerical value and a corresponding probability can be obtained for each grid point in a certain period.

[0037] Meanwhile, the surface recharge of the irrigation area also needs to be predicted to reflect the replenishment effect of the surface water system on the soil moisture comprehensively, combining the data set from the water body trend. Specifically, when allocating the recharge water from the river and reservoir to each farmland plot, the influence of the spatial relationship between the plot and the water body and the slope factor needs to be considered comprehensively. First, the area change of all water bodies in the prediction period is converted into volume change, which is often obtained through known DEM information and the like to obtain a rough new recharge water. If the water surface of a reservoir expands from 20 hectares to 22 hectares from 9 o'clock to 10 o'clock, according to the average depth change of the elevation band in the DEM, about 40,000 cubic meters of new recharge water can be calculated. Then, in order to reasonably distribute the 40,000 cubic meters of water to the entire irrigation area, a spatial response coefficient needs to be calculated for each farmland, which comprehensively considers the horizontal distance between the plot and each water body, the soil permeability of the stratum, and the slope of the terrain and other factors. For example, the plot closest to the reservoir and relatively low in terrain is more likely to obtain recharge due to gravity, so the response coefficient will be larger; if a plot is far away from the main river or in an upward slope direction, the response coefficient will be smaller. Adding up the spatial response coefficients of all plots and horizontally distributing them in proportion to the proportion of each plot's coefficient in the total coefficient can achieve: new recharge water × (plot's coefficient ÷ total coefficient). Through this water distribution logic, it is ensured that plots close to water bodies and with good permeability conditions can obtain recharge in priority, and the problem of drought in remote plots caused by simple equal distribution of water is avoided, so that each plot obtains reasonable infiltration recharge water (i.e., surface recharge prediction data) according to its water source accessibility and terrain conditions.

[0038] After the predicted precipitation ρ i (t) and the surface recharge prediction data Q i 补 (t) are determined, the soil moisture content of each plot can be deduced according to the layered soil mode. Taking the soil moisture content θ i,d (t) of the depth layer d as the state variable, a mass conservation model as shown below can be established: wherein, D d represents the crop evapotranspiration water consumption of each depth layer of the plot in the prediction period. D d is the soil volume conversion coefficient corresponding to the thickness of the soil layer, so that the moisture content with a unit of percentage can be mapped with the cubic millimeter volume of water. If is greater than the evapotranspiration consumption, The soil moisture content at different depths can be estimated accurately by layering the rainfall and recharge amount of each plot. For example, if the first layer (0-10 cm) of a plot has a moisture content of 15% at 9:00, and the predicted rainfall for the period is 5 mm, the surface recharge is 2 mm, and the evapotranspiration consumption is 3 mm, then the soil moisture content of the layer can be calculated by the mass conservation model as follows: θ i,1 (t+1h)=15%+(5+2-3) / 100=19%, i.e., the soil moisture content of the layer is greatly increased to 19%.

[0039] Finally, the irrigation status can be inferred from the predicted soil moisture content in combination with the current crop growth stage and the corresponding field capacity data. If the field capacity corresponding to the current crop growth stage is θ FC , and the predicted soil moisture content is θ < θ FC , then it is considered that the soil of the layer needs to be irrigated. The irrigation status is defined as a level value as follows in the embodiments of the present application: wherein S i,d =1 indicates that irrigation is needed, and S i,d =0 indicates that irrigation is not needed. During the data set packaging stage, the predicted soil moisture content of each depth layer of each plot is packaged together with the corresponding irrigation status to form the final soil moisture prediction data set. Through this data set, it can be known clearly which plots will have a risk of drought due to insufficient rainfall and surface recharge in the future period, so that scientific irrigation arrangements can be made in advance.

[0040] Step S3, fitting the water level and area to construct a water storage capacity model according to the water body change trend data set.

[0041] When rasterizing each water body contour polygon, the closed contour in vector form needs to be first mapped to the same grid as the regional DEM according to the known projection coordinates, and the pixels inside the contour are marked as "water body" and the remaining pixels are marked as "non-water body". The abstract geometric boundary is converted into a pixel-level binary mask, so that spatial analysis can be performed according to the resolution of the DEM to perform superposition operations pixel by pixel. For example, if the resolution of the DEM is 2 m x 2 m, when a closed polygon covers 150 pixels, it can be directly inferred that the current horizontal water surface of the water body is about 150 x 4 = 600 m 2 , which provides input for the next step of area calculation.

[0042] After obtaining the binary mask, it is overlaid with the regional DEM on a pixel-by-pixel basis to extract the elevation distribution of the water-covered area at the current time. To further obtain a more accurate average water level, the mode of the DEM elevation values corresponding to all pixels within the mask can be calculated. Here, the mode is used instead of the simple mean to eliminate the interference of local outliers. For example, if the outline of a water body just crosses a dug pit or a small bridge tunnel, the extremely low or extremely high DEM value can lower or raise the average value, and the mode statistics can select the elevation with the highest frequency, taking the intermediate elevation as the best estimate of the water surface elevation. The specific method is to traverse the DEM elevation value of each pixel within the mask, record the number of each elevation in the form of a histogram, and finally select the one with the highest frequency h mode as the average water level. For example, when most of the pixels on the plane of a reservoir correspond to an elevation of 120.5 m, 120.5 m is taken as the water surface height at that time.

[0043] The product of the number of all water body pixels in the binary mask and the area of the DEM pixel can directly calculate the horizontal water area at that time. If the area of a single pixel is A P and there are P pixels marked as "1" in the binary mask, then the horizontal area S = P × A P。 The significance of this calculation is to accurately correspond the pixel count to the real ground area, rather than just estimating based on the geometric outline. For example, when a water body contains 250 pixels after being rasterized at a certain time, and the DEM resolution is 2 m × 2 m, the horizontal water surface area can be determined as 1000 m 2 This result is directly used for subsequent function fitting.

[0044] After obtaining a series of water surface elevations h i and corresponding horizontal areas S i , the discrete {(h i , S i )} data points need to be described by a continuous area-height association function f(h) so that subsequent area and volume estimates for any water level can be made. To meet both fitting accuracy and ensure the smoothness of the function in different elevation intervals, a least squares spline fitting with a smoothing regularization term can be used. The core idea is to find a function f(h) that approximately matches the given data points {(h i , S i )} i=1 N while penalizing excessive fluctuations in the second derivative of the function globally, thereby preventing overfitting. The fitting objective is represented by the following formula: where μ is the smoothing parameter, used to balance the data approximation accuracy and the smoothness of the function. It is the second derivative of the function f at height u, reflecting the change in curvature. [h] min ,h max The value represents the current water body elevation range. After performing this fitting, a continuous and differentiable area-height function S=f(h) is obtained. For example, when a reservoir is at 118m, 120m, and 122m, the measured horizontal area is 5000m². 2 6000m 2 7200m 2 The fitting process will make the second derivative of the function relatively small while keeping f(118)≈5000, f(120)≈6000, and f(122)≈7200, so that a smooth and reasonable area estimate can be given at non-observation heights such as 119m and 121m.

[0045] After constructing the area-height correlation function, it is further transformed into a volume-height correlation function to map the water surface elevation to the cumulative water volume. Given the area-height curve S=f(h), the volume-height relationship V(h) can be regarded as the volume-height relationship starting from the lowest water storage elevation h. min The volume-height relationship is expressed by the cumulative surface area integrated up to the current elevation h. Here, α(u) is a scale factor that maps height to vertical distance, used to align elevation units on the DEM with actual water depth units. If the aspect ratio of DEM pixels is independent of horizontal area, α(u) can be set to 1. For example, when the area-height correlation function f(u) is approximated by a continuous curve to show the change in the cross-sectional area of ​​the water surface with elevation, integrating it from 118m to 122m yields the total volume accumulated in the reservoir when the water level reaches 122m. If the water depth mapping coefficient for this area is known to be 0.001, then f(u) at 120m is 6000m. 2 This means that a cross-sectional thickness of 1m corresponds to a thickness of 6m. 3 The entire integral, by summing up all elevations, restores the actual water volume of the body.

[0046] Finally, the fitted area-height correlation function S=f(h) and the integrated volume-height correlation function are combined. The entire process is encapsulated to generate a complete description of the water storage capacity model. This model can directly retrieve the horizontal area and volume by calling a specific water level value during real-time or future water level forecasting phases, thus providing an immediate estimate of available water for subsequent water supply scheduling. For example, if the current real-time reservoir water level is input as 121.3m into the scheduling system, the model can output the corresponding water area of ​​approximately 6800m². 2and a water storage volume of about 9 x 10 6 m 3 , so as to help the system to determine whether the downstream irrigation water requirement is met or whether the floodgate needs to be opened for flood storage when making decisions.

[0047] Step S4, crop water requirement calculation is performed according to the real-time collected regional data and crop data of the target irrigation area, so as to obtain a crop water requirement matrix data set.

[0048] In the field automatic monitoring system, soil volume water content sensors and soil temperature sensors are first arranged in each farmland plot, which are usually installed at multiple layers at different depths (for example, 0-10 cm, 10-30 cm, 30-60 cm) to obtain the original soil water content and temperature values of each layer of soil in real time. At the same time, air temperature and humidity sensors and radiation sensors are installed on the ground to continuously collect key meteorological elements such as environmental humidity, air temperature, solar radiation intensity and wind speed. When the sensor device reports data at a fixed time, the system has mastered the vegetation index determined by the current vegetation growth of the target irrigation area, the surface temperature and humidity conditions, and the temperature and water content of each depth layer of soil. This all-round and multi-level perception can accurately reflect the water conditions of different depth soils and crop canopies. For example, if the upper layer soil water content of a plot is less than 10%, the air humidity is continuously below 30%, and the wind speed is large, it can be inferred that the evapotranspiration intensity of the plot has been very high, and the water requirement degree has risen sharply.

[0049] Then, for each collected raw soil moisture value, a temperature correction model is needed to convert it into a normalized volumetric water content. The significance of this correction model is that the soil moisture sensor output is often affected by temperature drift: when the temperature rises, the sensor's resistance or capacitance readings will drift, resulting in a lower moisture reading, and vice versa. In order to eliminate this error, the temperature correction model correlates the raw soil moisture value measured by the sensor with its corresponding soil temperature, automatically offsetting the impact of temperature on the measurement. Specifically, whenever the sensor reads a raw soil moisture, say 20%, and the soil temperature at that point is 30°C, the temperature is compared with a pre-set reference temperature (usually the sensor's calibration temperature in a standard laboratory environment). The difference between the two is first squared to amplify it, and then multiplied by a temperature-sensitive correction coefficient, which is derived from a special experimental calibration of the performance drift of this model sensor at different temperatures. Thus, an amplification factor is obtained, which will significantly correspond the temperature deviation to the moisture, to reflect the sensor's reading deviation at that temperature. Finally, dividing the raw reading by this amplification factor, a standardized soil moisture data after temperature correction can be obtained. The significance of this is that whether the soil temperature rises or falls, the raw soil moisture will not be systematically high or low. For example, if the soil temperature is 10°C higher than the reference temperature, the temperature difference will be expanded to one hundred after squaring, and multiplied by a small coefficient, so that the amplification factor exceeds one, making the final output of the standardized soil moisture data significantly lower than the original value, thus offsetting the false high reading of the sensor at high temperature. Conversely, if the temperature is very close to the reference temperature, the amplification factor is close to one, the correction amplitude is small, and the original value is maintained. In this way, by amplifying and normalizing the temperature drift twice, the moisture value at any time can be close to the actual water content of the soil itself, without distortion due to external temperature changes.

[0050] After obtaining the standardized soil moisture data, the system will calculate the reference evapotranspiration of each plot according to the Penman-Monteith algorithm, which is the amount of water needed to evaporate per unit area under standard grass conditions. The reference evapotranspiration depends on both environmental humidity and weather elements such as solar radiation, wind speed and air temperature, which can be represented by the following formula: where R n,i (t) represents the net radiation received by the i i th plot at time t. G i (t) is the soil heat flux of the ground surface. Δ is the slope of the saturated water vapor pressure curve with temperature. γ is the dry constant. T i (t) is the air temperature. u2,i (t) represents the wind speed at a height of 2m. e s,i (t) and e a,i (t) represents the saturated vapor pressure and the actual vapor pressure, respectively. By combining the known soil depth and temperature data at the current moment with a thermal conductivity model, the amount of heat conducted downwards per unit area per hour is estimated, thus more accurately determining the portion of surface energy available for evapotranspiration. Its significance lies in the fact that energy changes in the surface and soil layers significantly impact evapotranspiration, regardless of whether it's cloudy / rainy weather or intense sunlight. This algorithm provides a reference evapotranspiration rate that better reflects the actual environment.

[0051] After calculating the reference evapotranspiration, it is also necessary to combine it with the actual evapotranspiration coefficient K of the crop. c,i (t) is used to obtain the actual crop evapotranspiration ET. c,i (t)=K c,i (t)×ET o,i (t). To make K c,i (t) As crop growth is adjusted in real time, an initial crop coefficient K can be obtained by looking up a table based on the crop type and current crop growth stage for each field. c,tab,i Simultaneously, the original crop coefficient K was compared with the real-time vegetation index (Normailzed Difference Vegetation Index, NDVI) obtained by drones or ground-based multispectral cameras to reflect vegetation greenness. c,tab,i Dynamic adjustments are made. Specifically, the actual greenness information and historical lower and lower limits are normalized. If the vegetation index approaches full green, it indicates that the crop is growing very well, and the coefficient may increase by 20% to 30% above the base value. Conversely, if the vegetation index is low or the leaves are sparse, it means that the vegetation cover is not good enough, and the actual evapotranspiration multiple needs to be reduced. Through this dynamic adjustment, the evapotranspiration used for irrigation decisions can take into account both the standard grassland model and the differences in the crop's own growth status. For example, when the NDVI of a wheat field obtained from satellite imagery navigation at the jointing stage is close to 0.90, and the preset full green and initial thresholds are 0.95 and 0.30 respectively, the crop coefficient for that plot will be taken from the base value in the menu and then scaled up by about 10% according to the vegetation index ratio. This ensures that when the vegetation is growing vigorously, the actual evapotranspiration multiple exceeds the coefficient in the table, thus reflecting the crop's stronger water requirement under the same reference conditions. The actual evapotranspiration multiple (ET) is then obtained. c,i (t) after which the actual evaporation rate ET c,i (t) represents the water depth required per unit area ΔW i (t). This unifies the water consumption per unit area with the soil moisture retention capacity in terms of dimensions, making it easier to use in conjunction with volumetric water content and land area.

[0052] Finally, combined with the known field water holding capacity θ FC,i,d and the standardized moisture content θ at different depths for each plot of land s,i,d (t) is used to calculate the actual amount of irrigation water needed for each plot during the current time period. For the d-th depth layer of the i-th plot, if the root activity depth is Z... d The water capacity of this soil layer is approximately [θ]. FC,i,d -θ s,i,d (t)]×Z d This represents the water depth required for the soil layer to recover from its current measured moisture content to field capacity. If the soil layer has already reached or exceeded field capacity, then the available water supply for that layer is zero. The cumulative water recharge depth required for the i-th plot during the current time period is obtained by summing the water requirements at all root-accessible depths. By calculating the cumulative required water replenishment depth D i (t) The moisture difference between each layer in the multi-layered soil is summed to obtain a total water requirement depth. For example, if the moisture content of the first layer is 15%, the field capacity is 25%, and the layer thickness is 200 ms, then this layer requires 0.10 × 200 = 20 ms of water. If the second layer is close to saturation, its water requirement is not included. If the moisture content of the third layer is also lower than the field capacity, it is calculated again. Finally, the three layers are added together to obtain the total water requirement of the current plot. Then, this water requirement depth is multiplied by the plot area A. i This gives us the total amount of water W that needs to be replenished to the plot within a unit of time period. i (t)=D i (t)×A i If the plot area is 5 hectares and the total required water depth is 20 m³, then the required water replenishment is approximately 1000 m³. 3 Finally, the calculated water replenishment amount for each plot in the current time period is sequentially filled into the crop water requirement matrix, forming a two-dimensional array structure. The rows in the crop water requirement matrix represent plot numbers, and the columns represent different depths or the current total water requirement, for use in subsequent scheduling and optimization.

[0053] Step S5: Perform constrained planning based on the soil moisture prediction dataset, the crop water requirement matrix dataset, and the water storage capacity model to obtain a multi-objective irrigation scheduling decision set.

[0054] Before conducting multi-objective constrained programming, the water demand information for each plot of land in the current and several future time periods is first extracted from the soil moisture prediction dataset and the crop water requirement matrix dataset. The significance of this process is to clarify the irrigation water required by each plot of land at each moment, so as to coordinate supply and demand at different time points in the future. For example, if it is predicted that a cornfield will require 100 m³ of water in the next hour, the solution is to... 3 80m³ of water is needed in the next two hours 3 The rice paddies will need 150 cubic meters of water in the next hour.3 , water demand 120m in the next two hours 3 , these numbers form a two-dimensional demand matrix, where the row index corresponds to "each plot", and the column index corresponds to "the current period and each subsequent predicted period". Through this step, the water demand of each plot at the present and future can be grasped in a space-time framework, providing input data for the optimization algorithm.

[0055] At the same time, the water level of the current period and each future period is mapped using the storage capacity model to obtain the corresponding available water resources. The storage capacity model is essentially a function that converts "a certain water level in the reservoir or channel" to "the volume of water storage or the amount of water that can be regulated". When executed, it will bring the water level value h(t j ) of each predicted period into the function V(h), thereby obtaining the available water at time t j . For example, if the reservoir water level is measured to be 121.5m at the current time, V(121.5)=8000m 3 ; if the river inflow increases to 122m in the next hour, V(121.5)=8500m 3 . After obtaining this set of available water resources data, the upper limit guarantee for the scheduling constraints of each period can be provided, ensuring that the total irrigation amount does not exceed the available water.

[0056] After determining the water demand of each plot in each period and the available water in each period, a quadratic weighted optimization objective function that takes into account both "plot differences" and "period urgency" is constructed. First, the weight coefficient w i of each plot needs to be pre-set by regional management or agricultural experts, reflecting crop value, drought resistance importance, and plot position in the network, etc. Second, a period weight coefficient a j needs to be pre-assigned to each period to reflect "the current period is more urgent than the future period" or "the weight of a future period needs to be appropriately raised if the inflow is unstable". On this basis, if W i (t j ) represents the predicted water demand of plot i in period t j , and x i (t j ) represents the actual water amount scheduled to plot i in that period, then the "supply-demand deviation" can be represented as W i (t j )-x i (t j ). In order to let the deviation affect the objective function considering both the size of the deviation itself and the importance of the plot and period, the quadratic weighted optimization objective function can be represented by the following formula: The first term in the quadratic weighted optimization objective function represents the sum of "square of supply-demand difference" on each plot in each time period, where a j is the time period weight, w i is the plot weight. The time period weight determines "whether the current time period is more important in the overall optimization"; the plot weight determines "whether a plot is more prioritized". For example, if the current time period weight a0= 0.6 and the weight of the next hour a1= 0.3, the current supply-demand deviation is amplified to twice that of the future. The second term introduces a smoothing constraint between adjacent time period nodes by means of λ, preventing the scheduling scheme from repeatedly rising and falling in a short time, and ensuring smooth operation of the irrigation system; where λ is the smoothing penalty coefficient, which can make the scheduling result more consistent with the physical characteristics of the irrigation equipment start-stop. For example, if λ takes a large value, the water pump and valve will not be frequently turned on and off in a short time, so as to avoid mechanical wear and tear.

[0057] After the objective function is constructed, the water demand data and available water resource data need to be included in the constraint conditions to ensure that the water allocated to each plot meets the physical feasibility. The most basic constraint is that for each time period t j , the total water obtained by all plots cannot exceed the available water resources V(h(t j )) in that time period. The significance of this constraint is to prevent unrealistic situations such as "scheduling demand exceeding current or future available water storage". For example, if the reservoir can only provide 8000m 3 in the current time period, and if the sum of the demands of all plots is 9000m 3 , then part of the demand must be reduced to within 8000m 3 , otherwise it will cause the system scheduling to be unbalanced.

[0058] Another type of constraint is the upper limit for each plot in each time period, i.e. each plot cannot obtain more water than its demand in any time period, which can avoid wasting water resources and ensure that the plot is not over-irrigated. For example, if the current water demand of a plot is only 100m 3 , allocating 150m 3 to it is both redundant and can cause runoff loss. Finally, the pump group or valve flow limit can also be included in the constraints, such as if the maximum allowable flow of a branch canal is 200m 3 / h, then the total water allocated to all plots downstream of the valve in the current time period cannot exceed 200, which ensures that the pipe network will not have the risk of sudden pressure increase or pipe rupture due to overload.

[0059] With the objective function and constraint conditions, a numerical solver can be called to solve the quadratic programming problem to obtain a series of optimal irrigation allocation quantities {x i (t jThe solver continuously adjusts the allocation of each plot at each time period by iteration to minimize the objective function while strictly meeting the constraint conditions. For example, if water is tight at the current time period, the solver may prioritize plots with higher weights and the most urgent water needs; if the available water increases in the future period, more water can be allocated to the secondary water demand plot to balance the supply and demand rhythm throughout the irrigation period. Through this combined solution, the system not only takes into account the economic value and drought resistance of each plot, but also takes into account the smoothness of the irrigation equipment operation, so as to obtain the optimal irrigation scheduling decision set in actual engineering.

[0060] In order to convert the irrigation scheduling decision set into the control signal of the automatic irrigation equipment, the embodiments of the present application further comprise: After generating the multi-objective irrigation scheduling decision set, the irrigation amount required by each plot at each time period needs to be converted into actual execution instructions. This process first reads the allocated water amount at each time period in the decision set, and then calculates the target flow under each irrigation branch or valve for each time period to determine the opening of the corresponding valve or the speed of the pump. Specifically, if the decision set indicates that 120 m 3 / h is to be supplied to the first area valve V1 at the current time period, and there is a nonlinear mapping relationship between the opening of the valve and the flow, the target flow is interpolated through the known calibration curve to obtain the corresponding opening percentage. Then, the controller sends a setting signal to the field execution unit, such as rotating the valve actuator to a certain angle, and dynamically correcting between the angle and the flow. In some cases, the valve and the pump are operated together: if the plots are distributed at the end of the multi-stage pipe network, in order to ensure the back-end pressure, the water pump will also be operated at a certain speed to overcome the flow resistance, at this time the valve opening and the pump frequency need to be calculated together to obtain the accurate execution parameters.

[0061] Subsequently, PID closed-loop controllers are enabled on each valve and pumping execution unit to ensure that the actual flow rate matches the target flow rate. The core idea of closed-loop control is to collect the feedback value of the flow rate sensor in real time and perform error calculation with the preset flow rate target. The PID controller calculates an adjustment amount according to the current error, the integral of the past error, and the rate of change of the current error. This means that if the on-site flow rate is lower than the target value, the controller will issue a “increase the opening” or “speed up the pump” instruction to the execution mechanism; otherwise, it will “reduce the opening” or “slow down the pump”. In the setting of PID parameters, the proportional gain determines the direct influence of the error on the control amount, the integral gain is responsible for eliminating the steady-state error, and the differential gain suppresses system overshoot. Through this feedback mechanism, the system can quickly correct itself in the case of fluctuations in pipe network pressure, leakage, or sudden increase in water consumption. For example, when a blockage occurs in the upstream pipe network at a certain time, causing the flow rate to drop sharply, the flow rate sensor immediately feeds back a low value, and the PID controller quickly increases the pump frequency until it returns to the target flow rate or approaches the maximum allowed value.

[0062] The timing switching logic is also pre-set at each time period node. For example, if the decision set specifies that the flow rate of a certain pump group should be switched from 120 m 3 / h to 90 m 3 / h after one hour (i.e., the next time period), the control system will issue a “pre-switching signal” a few minutes before the end of the current time period, causing the pump to gradually slow down instead of being directly powered off, in order to avoid water hammer effects. The pre-switching logic is intended to protect the pipe network from sudden impacts and extend the service life of the equipment. If there are multiple valves sharing a pump in the pipe network architecture, the system will analyze the pressure and flow rate values detected at the boundary of each valve and achieve overall balance in a cascading control manner: for example, when reducing the flow rate of a downstream branch, the pump output will also be reduced accordingly to prevent overall overpressure or pump idling.

[0063] At the same time, in order to further improve the reliability of equipment operation, the response state of each execution unit is recorded in the background after the instruction is sent. For example, after the controller issues an instruction to set the valve opening to 45%, the real-time output of the angle sensor or encoder feedback by the valve is read. If the position deviation exceeds the set threshold (such as ±2%), compensation will be performed again through closed-loop control, and this process will be recorded in the log. The significance of the log is that maintenance personnel can track the response speed, overshoot amplitude, or hysteresis phenomenon of each valve or pump during the control process in the future, which facilitates the diagnosis of whether there is an actuator aging, pipe blockage, or sensor failure.

[0064] As time enters the next period, the control system will automatically load the next period of scheduling targets, continue to perform the same valve opening calculation and PID adjustment process, so as to form a ring-shaped automatic control link in the whole irrigation period. For example, at the end of the current period, the system will first compare the "scheduled water volume remaining in the current period" with the "predicted demand in the next period", if it is found that the demand in the next period is greater than the predicted available water volume, the pump speed will be reduced or some valves will be closed in advance, so as to leave the available water volume for future periods to prevent future water shortage; if it is found that the available water volume is sufficient, the valves can be opened early in the next period to irrigate the high water demand block early. Through such continuous dynamic adjustment, the system eventually generates a complete set of "time series" control instructions for the valves and pumps, including the opening percentage, pump speed, pre-switching signal and sensor feedback monitoring threshold every minute or second, so as to form a complete and adaptive linkage irrigation control instruction set.

[0065] The application belongs to the technical field of water conservancy monitoring, and obtains an initial water area boundary data set by extracting a water body boundary of a target irrigation area from remote sensing image data, and obtains a water body change trend data set by time series interpolation on the initial water area boundary data set. The soil moisture content prediction data set is generated by combining satellite cloud images to deduce and generate soil moisture content prediction data sets, and the water storage capacity model is constructed by fitting the water level and area of the regional DEM according to the water body change trend data set. The crop water demand matrix data set is obtained by ground sensing and sensing the soil moisture content and crop growth state of the target irrigation area. The multi-objective irrigation scheduling decision set is obtained by constraint programming according to the soil moisture content prediction data set, the crop water demand matrix data set and the water storage capacity model. Finally, the linkage irrigation control instruction set is generated according to the multi-objective irrigation scheduling decision set. The application itself realizes dynamic, accurate and automatic monitoring and scheduling of water resources and crop water demand in the irrigation area by fusing remote sensing data, time series analysis, ground sensing and intelligent scheduling.

[0066] As shown in Figure 2 , it is a functional module diagram of an irrigation area intelligent water conservancy monitoring device provided by an embodiment of the application.

[0067] In some embodiments, the target irrigation area intelligent water conservancy monitoring device 2 can include a plurality of functional modules composed of computer program segments. The computer programs of each program segment in the target irrigation area intelligent water conservancy monitoring device 2 can be stored in the memory of the server and executed by at least one processor to perform the functions of the irrigation area intelligent water conservancy monitoring method (see Figure 1 Description).

[0068] In the embodiment, the target irrigation area intelligent water conservancy monitoring device 2 can be divided into multiple functional modules according to the functions performed by the target irrigation area intelligent water conservancy monitoring device 2. The functional modules can include a water area analysis module 21, a soil moisture condition prediction module 22, a water storage model module 23, a water demand measurement module 24, and an irrigation decision module 25. The module referred to in the present application refers to a series of computer program segments that can be executed by at least one processor and can complete a fixed function, which is stored in a memory. In the embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0069] The water area analysis module 21 is configured to extract the water body boundary of the target irrigation area from the collected remote sensing image data to obtain an initial water area boundary data set, and perform time series interpolation processing on the initial water area boundary data set to obtain a water body change trend data set.

[0070] In an optional implementation, the water area analysis module 21 is configured to: fuse and correct the collected remote sensing image data to obtain multispectral images; extract green light band pixel values and near-infrared band pixel values from the multispectral images, and perform normalization calculation on the green light band pixel values and the near-infrared band pixel values to obtain an NDWI image; perform binaryzation segmentation on the NDWI image according to a preset empirical threshold to obtain a water body candidate area, and perform morphological opening operation and morphological closing operation on the water body candidate area to obtain a binary water body image; perform connected component analysis and contour coordinate extraction conversion on the binary water body image to obtain map coordinate sequences and pixel areas of each connected water body, and convert the map coordinate sequences into vector element data through a preset vertex simplification algorithm.

[0071] In an optional implementation, the water area analysis module 21 is further configured to: group and sort the initial water area boundary data set according to a preset water body ID and a shooting time to obtain a polygon of each water body; perform polar coordinate conversion on each vertex in the polygon to obtain a polar coordinate sequence of each polygon, and construct an angle and radius association data set of each water body at each shooting time according to the polar coordinate sequence; perform linear sample interpolation processing on the radius corresponding to each sampling angle at each time according to a preset sampling angle set and the angle and radius association data set to obtain the radius corresponding to each sampling angle at each time for each water body; updating the polar coordinate sequence according to the radius corresponding to each sampling angle at each time, and performing Cartesian coordinate system conversion on the updated polar coordinate sequence to obtain a water body contour polygon of each water body at each time; arranging the water body contour polygon according to time sequence to obtain a water body change trend data set corresponding to each water body.

[0072] The soil moisture prediction module 22 is configured to perform soil moisture prediction deduction based on the satellite cloud image acquired in real time and the water body change trend data set to generate a soil moisture prediction data set.

[0073] In an optional embodiment, the soil moisture prediction module 22 is specifically configured to: projecting and cloud feature recognizing and extracting the satellite cloud image acquired in real time according to a preset irrigation area coordinate system to obtain continuous regional cloud data; predicting the cloud layer moving speed and direction of the continuous regional cloud data by a preset cloud field tracking algorithm to obtain cloud band prediction data of each grid in the irrigation area coordinate system; estimating the precipitation of each plot in the target irrigation area in a plurality of future time periods according to preset historical precipitation and cloud top temperature statistical data and the cloud band prediction data to obtain precipitation prediction data of each plot; performing irrigation area surface recharge prediction according to the water body change trend data set to obtain surface recharge prediction data; deducing the soil layer moisture content of the target irrigation area according to the precipitation prediction data and the surface recharge prediction data to obtain predicted moisture content; obtaining the corresponding field water holding capacity according to the current crop growth stage, and calculating water supplement state data according to the field water holding capacity and the predicted moisture content, to generate the soil moisture prediction data set according to the water supplement state data and the predicted moisture content.

[0074] The water storage model module 23 is configured to construct a water storage capacity model by fitting the water level and area of a preset regional DEM according to the water body change trend data set.

[0075] In an optional embodiment, the water storage model module 23 is specifically configured to: performing raster processing on the water body contour polygon in the water body change trend data set to obtain a binary mask at each time; performing per-pixel superposition and elevation value mode statistics on the binary mask and a preset regional DEM to obtain an average water level at each time; calculating the horizontal water area corresponding to each time according to the pixel number of all water bodies in the binary mask and the single-pixel area in the regional DEM; construct an area-height correlation function according to the average water level and the horizontal water area by a preset polynomial fitting algorithm, and perform numerical integral operation on the area-height correlation function according to a preset water level threshold to obtain a volume-height correlation function; encapsulate the area-height correlation function and the volume-height correlation function to construct the water storage capacity model.

[0076] The water demand measurement module 24 is configured to calculate crop water demand according to the real-time collected regional data and crop data of the target irrigation area to obtain a crop water demand matrix dataset.

[0077] In an optional embodiment, the water demand measurement module 24 is specifically configured to: correct the original soil moisture content data according to the soil temperature by a preset temperature correction formula to obtain standardized soil moisture content data; calculate the reference evapotranspiration according to the standardized soil moisture content data, the environmental humidity, and the weather element data by a preset Penman-Monteith algorithm; obtain original crop coefficients according to the crop type and the current crop growth stage of the target irrigation area, and dynamically correct the original crop coefficients according to the real-time vegetation index to obtain real-time crop coefficients; calculate the unit area water demand depth according to the reference evapotranspiration and the real-time crop coefficients, and calculate the current water supplement demand according to a preset field water holding rate, the area of each plot in the target irrigation area, the standardized soil moisture content data, and the unit area water demand depth to obtain the crop water demand matrix dataset.

[0078] The irrigation decision module 25 is configured to perform constraint programming according to the soil moisture prediction dataset, the crop water demand matrix dataset, and the water storage capacity model to obtain a multi-objective irrigation scheduling decision set.

[0079] In an optional embodiment, the irrigation decision module 25 is specifically configured to: perform water demand prediction and extraction according to the soil moisture prediction dataset and the crop water demand matrix dataset to obtain water demand data of the current period and a plurality of future periods; perform water storage amount mapping value calculation on the water storage capacity model to obtain available water resource data of the current period and a plurality of future periods; construct a quadratic weighted optimization objective function of each plot in each period according to a preset plot weight coefficient and the water demand data; The secondary weighted optimization objective function is solved with multi-objective constraints according to the water demand data and the available water resource data, to obtain a multi-objective irrigation scheduling decision set.

[0080] It should be understood that various changes and modifications to the method provided by the above embodiments are equally applicable to the irrigation area intelligent water conservancy monitoring device of the present embodiment. Through the foregoing detailed description of the irrigation area intelligent water conservancy monitoring method, those skilled in the art can clearly understand the implementation method of the irrigation area intelligent water conservancy monitoring device in the present embodiment. For the sake of brevity of the description, the implementation method of the irrigation area intelligent water conservancy monitoring device in the present embodiment will not be described in detail here.

[0081] As shown in Figure 3 FIG. 1 is a structural schematic diagram of an electronic device according to an embodiment of the present application.

[0082] In the preferred embodiment of the present application, the electronic device 3 can include, but is not limited to, a memory 31, at least one processor 32, and at least one communication bus 33.

[0083] Those skilled in the art should understand that Figure 3 The structure of the electronic device 3 shown is not a limitation of the embodiments of the present application. The electronic device 3 can also include more or fewer other hardware or software, or different component arrangements than those shown.

[0084] In some embodiments, the electronic device 3 is a device capable of automatically performing numerical calculation and / or information processing according to pre-set or stored instructions. The hardware thereof includes, but is not limited to, a microprocessor, an application specific integrated circuit, a programmable gate array, a digital processor, and an embedded device, etc.

[0085] It should be noted that the electronic device 3 is only an example. Other existing or future electronic products, such as those that can be adapted to the present application, should also be included within the scope of protection of the present application and are hereby incorporated by reference.

[0086] In some embodiments, the memory 31 stores a computer program which, when executed by the at least one processor 32, implements all or part of the steps of the irrigation area intelligent water conservancy monitoring method as described. The memory 31 includes Read-Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), One-time Programmable Read-Only Memory (OTPROM), Electrically-Erasable Programmable Read-Only Memory (EEPROM), Compact Disc Read-Only Memory (CD-ROM) or other optical disk memory, magnetic disk memory, magnetic tape memory, or any other computer readable medium capable of carrying or storing data. Further, the computer readable storage medium can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function, etc.

[0087] In some embodiments, the at least one processor 32 is a control core of the electronic device 3, which connects various components of the entire electronic device 3 through various interfaces and lines, and performs various functions of the electronic device 3 and processes data by running or executing programs or modules stored in the memory 31 and calling data stored in the memory 31. For example, the at least one processor 32 implements all or part of the steps of the irrigation area intelligent water conservancy monitoring method as described in the embodiments of the present application when executing the computer program stored in the memory 31, or implements all or part of the functions of the irrigation area intelligent water conservancy monitoring device. The at least one processor 32 can be composed of integrated circuits, for example, can be composed of a single packaged integrated circuit, or can be composed of multiple packaged integrated circuits with the same function or different functions, including one or more combinations of central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips.

[0088] In some embodiments, the at least one communication bus 33 is configured to enable connection communication between the memory 31 and the at least one processor 32. Although not shown, the electronic device 3 can further include a power supply (such as a battery) for powering the various components of the electronic device 3. Preferably, the power supply is logically connected to the at least one processor 32 via a power management device, thereby enabling management of charging, discharging, and power consumption management, etc. by the power management device. The power supply can also include one or more direct current or alternating current power sources, recharging circuits, power failure detection circuitry, power converters or inverters, power status indicators, and the like. The electronic device 3 can further include a variety of sensors, a Bluetooth module, a Wi-Fi module, and the like, which are not described herein.

[0089] The integrated units in the form of software function modules described above can be stored in a computer readable storage medium. The software function modules described above are stored in a storage medium, and include a plurality of instructions for causing an electronic device (which can be a personal computer, an electronic device, or a network device, etc.) or a processor to execute part of the method described in various embodiments of the present application.

[0090] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. For example, the described apparatus embodiments are merely illustrative. For example, the division of the modules is merely a logical function division. There can be another division manner in actual implementation.

[0091] The modules illustrated as separated components can or can not be physically separated, and the components illustrated as modules can or can not be physical units. They can be located in one place or distributed on a plurality of network units. Some or all of the modules can be selected according to actual needs to achieve the purposes of the embodiments of the present application.

[0092] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Therefore, any equivalent changes made on the structure, shape, and principle of the present application should be covered within the protection scope of the present application.

Claims

1. A method for intelligent water conservancy monitoring in an irrigation area, characterized in that, The method includes: The water body boundaries of the target irrigation area are extracted from the collected remote sensing image data to obtain an initial water body boundary dataset, and the initial water body boundary dataset is subjected to time-series interpolation to obtain a water body change trend dataset. Based on real-time acquired satellite cloud images and the water body change trend dataset, a soil moisture prediction dataset is generated through soil moisture prediction and deduction. Based on the water body change trend dataset, a water storage capacity model is constructed by fitting the water level and area of ​​the preset regional DEM. Crop water requirement is calculated based on the real-time collected regional data and crop data of the target irrigation area to obtain a crop water requirement matrix dataset. Constrained planning is performed based on the soil moisture prediction dataset, the crop water requirement matrix dataset, and the water storage capacity model to obtain a multi-objective irrigation scheduling decision set.

2. The intelligent water conservancy monitoring method for irrigation areas according to claim 1, characterized in that, The initial water boundary dataset includes pixel area and vector feature data; the extraction of water body boundaries of the target irrigation area from the acquired remote sensing image data to obtain the initial water boundary dataset includes: The acquired remote sensing image data is fused and corrected to obtain multispectral images; The pixel values ​​of the green band and the near-infrared band are extracted from the multispectral image, and the pixel values ​​of the green band and the near-infrared band are normalized to obtain an NDWI image. The NDWI image is binarized and segmented according to a preset empirical threshold to obtain water body candidate regions, and morphological opening and closing operations are performed on the water body candidate regions to obtain a binary water body image. The binary water body image is subjected to connected component analysis and contour coordinate extraction and transformation to obtain the map coordinate sequence and pixel area of ​​each connected water body, and the map coordinate sequence is converted into vector feature data through a preset vertex simplification algorithm.

3. The intelligent water conservancy monitoring method for irrigation areas according to claim 1, characterized in that, The step of performing time-series interpolation on the initial water boundary dataset to obtain a water change trend dataset includes: The initial water body boundary dataset is grouped and sorted into vector elements based on the preset water body ID and shooting time to obtain the polygon of each water body. Perform polar coordinate transformation on each vertex of the polygon to obtain the polar coordinate sequence of each polygon, and construct an angle and radius association dataset for each water body at each shooting time based on the polar coordinate sequence; Based on the preset set of sampling angles and the associated dataset of angles and radii, linear sample interpolation is performed on the radii corresponding to each sampling angle at each time moment to obtain the radii corresponding to each sampling angle at each time moment for each water body. The polar coordinate sequence is updated based on the radius corresponding to each sampling angle at each time, and the updated polar coordinate sequence is transformed into Cartesian coordinates to obtain the water body contour polygon at each time for each water body. The water body contour polygons are arranged in chronological order to obtain a continuous water body change trend dataset for each water body.

4. The intelligent water conservancy monitoring method for irrigation areas according to claim 1, characterized in that, The process of generating a soil moisture prediction dataset based on real-time acquired satellite cloud images and the water body change trend dataset includes: Based on the preset irrigation area coordinate system, the real-time acquired satellite cloud images are projected and cloud features are identified and extracted to obtain continuous regionalized cloud data. The cloud movement speed and direction are predicted by a preset cloud field tracking algorithm to obtain cloud band prediction data for each grid in the irrigation area coordinate system. Based on preset historical precipitation and cloud top temperature statistics and cloud band prediction data, precipitation estimates for each plot in the target irrigation area over several future periods are obtained to obtain precipitation prediction data for each plot. Based on the water body change trend dataset, surface recharge prediction for the irrigation area is performed to obtain surface recharge prediction data. Based on the precipitation forecast data and the surface recharge forecast data, the soil moisture content of the target irrigation area is extrapolated by layering to obtain the predicted moisture content. The corresponding field water holding capacity is obtained based on the current crop growth stage, and the water replenishment status data is calculated based on the field water holding capacity and the predicted moisture content, so as to generate the soil moisture prediction dataset based on the water replenishment status data and the predicted moisture content.

5. The intelligent water conservancy monitoring method for irrigation areas according to claim 1, characterized in that, The step of fitting water level and area to a preset regional DEM based on the water body change trend dataset to construct a water storage capacity model includes: The water body contour polygons in the water body change trend dataset are rasterized to obtain binary masks at each time point. The binary mask and the preset regional DEM are overlaid pixel by pixel and the mode of elevation values ​​are statistically analyzed to obtain the average water level at each time. The horizontal water area at each time point is calculated based on the number of pixels of all water bodies in the binary mask and the area of ​​a single pixel in the regional DEM. A pre-defined polynomial fitting algorithm is used to construct an area-height correlation function based on the average water level and the horizontal water area. The area-height correlation function is then numerically integrated based on a pre-defined water level threshold to obtain a volume-height correlation function. The area-height correlation function and the volume-height correlation function are encapsulated to construct the water storage capacity model.

6. The intelligent water conservancy monitoring method for irrigation areas according to claim 1, characterized in that, The regional data includes real-time vegetation index, original soil moisture content, soil temperature, environmental humidity, and weather data; the crop data includes crop type and current crop growth stage. The step of calculating crop water requirement based on real-time collected regional data and crop data of the target irrigation area to obtain a crop water requirement matrix dataset includes: The original soil moisture content is corrected based on the soil temperature using a preset temperature correction formula to obtain standardized soil moisture content data. The reference evapotranspiration is calculated using the preset Penman-Monteith algorithm based on the standardized soil moisture content data, the environmental humidity, and the weather element data. The original crop coefficient is obtained based on the crop type and current crop growth stage of the target irrigation area, and the original crop coefficient is dynamically corrected based on the real-time vegetation index to obtain the real-time crop coefficient. The water requirement depth per unit area is calculated based on the reference evapotranspiration and the real-time crop coefficient. The current water requirement is calculated based on the preset field water holding capacity, the area of ​​each plot in the target irrigation area, the standardized soil moisture content data, and the water requirement depth per unit area to obtain a crop water requirement matrix dataset.

7. The intelligent water conservancy monitoring method for irrigation areas according to claim 1, characterized in that, The step of performing constrained planning based on the soil moisture prediction dataset, the crop water requirement matrix dataset, and the water storage capacity model to obtain a multi-objective irrigation scheduling decision set includes: Based on the soil moisture prediction dataset and the crop water requirement matrix dataset, water requirement for each time period is predicted and extracted to obtain water requirement data for the current time period and several future time periods. The water storage capacity model is used to calculate the water storage mapping value to obtain available water resource data for the current period and several future periods. Based on the preset land parcel weight coefficients and the water demand data, a secondary weighted optimization objective function is constructed for each land parcel at each time period; Based on the water demand data and the available water resource data, the quadratic weighted optimization objective function is solved with multi-objective constraints to obtain a multi-objective irrigation scheduling decision set.

8. A smart water conservancy monitoring device for irrigation areas, characterized in that, The device includes: The water area analysis module is used to extract the water body boundary of the target irrigation area from the collected remote sensing image data to obtain an initial water body boundary dataset, and to perform time-series interpolation processing on the initial water body boundary dataset to obtain a water body change trend dataset. The soil moisture prediction module is used to generate a soil moisture prediction dataset by performing soil moisture prediction and extrapolation based on real-time acquired satellite cloud images and the water body change trend dataset. The water storage model module is used to fit the water level and area of ​​a preset regional DEM based on the water body change trend dataset to construct a water storage capacity model. The water requirement measurement module is used to calculate crop water requirement based on the regional data and crop data of the target irrigation area collected in real time, so as to obtain a crop water requirement matrix dataset. The irrigation decision module is used to perform constrained planning based on the soil moisture prediction dataset, the crop water requirement matrix dataset, and the water storage capacity model to obtain a multi-objective irrigation scheduling decision set.

9. An electronic device, characterized in that, The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the intelligent water conservancy monitoring method for irrigation areas according to any one of claims 1 to 7.

10. A 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 steps of the intelligent water conservancy monitoring method for irrigation areas according to any one of claims 1 to 7.

Images