Intelligent regulation and control method of garden high-density green plant irrigation water controller

Abstract

The invention relates to the technical field of green plant irrigation control, and discloses an intelligent regulation and control method for a park high-density green plant irrigation water controller, and the method comprises the steps: carrying out the preprocessing of interpolation completion based on sensor confidence score on depth sensor data; hydrological information evolution is carried out by using a layered hydrological prior information evolution model, and the water content is corrected layer by layer by using a residual neural network model; utilizing an improved AquaCrop model to generate irrigation strategies of different vertical depth layers; and performing weighted correction on the irrigation strategies of different vertical depth layers based on the observation result of the short-term rainfall. According to the method, accurate reconstruction of the moisture content of multiple vertical depth layers is realized through interpolation completion, hydrological prior evolution and residual network correction, a layered irrigation strategy is generated by using an improved AquaCrop model, and weighted correction is performed in combination with short-term rainfall, so that intelligent decision-making type irrigation control of a high-density green plant area is realized, and the method is suitable for large-scale popularization and application. The water utilization efficiency is improved; and the vegetation growth water environment is stabilized.

Description

Technical Field

[0001] This invention relates to the field of green plant irrigation control, and more particularly to a smart control method for a high-density green plant irrigation water controller in a park. Background Technology

[0002] High-density green areas in parks typically consist of a mix of trees, shrubs, herbs, and ground cover plants. Their varying root depths and complex canopy structures result in significant spatial heterogeneity in soil moisture distribution at different vertical depths. Furthermore, the water content at each depth exhibits strong temporal dynamism, influenced by factors such as rainfall intensity variations, evapotranspiration rate fluctuations, dynamic root water absorption behavior, soil texture differences, and seasonal environmental conditions. Traditional irrigation methods often rely on single-point or single-layer water content information, or even on manual experience, to determine irrigation volume and timing. This fails to reflect the true water evolution process within the multi-layered soil-vegetation system, easily leading to over-irrigation, under-irrigation, or delayed response, hindering truly precise management.

[0003] In high-density vegetation environments, shallow soils are prone to rapid wet-dry cycles, while moisture changes in deeper soils lag significantly. Traditional irrigation methods cannot effectively identify and regulate this phenomenon, resulting not only in water waste but also potential impacts on plant root vitality and growth health. Meanwhile, with the widespread adoption of IoT sensors and reduced deployment costs, multiple sensors can provide richer, more layered data.

[0004] Existing research, application number CN118020614A, discloses a vegetation irrigation method that acquires target images of vegetated areas using image acquisition equipment and uses a pre-trained judgment model on a GPU server to identify whether vegetation needs irrigation. This eliminates the need for additional data collection, reducing the server's computational burden and improving judgment efficiency. This approach achieves contactless irrigation demand identification through visual information, offering advantages in low cost and high flexibility. However, this method still has the following technical problems: images can only reflect surface vegetation characteristics and cannot observe the moisture status of different vertical depths underground; in high-density vegetation scenarios, visual information is easily affected by occlusion, lighting, and shadows, leading to unstable irrigation demand judgments.

[0005] To address this issue, this invention proposes a smart control method for high-density green plant irrigation water controllers in parks, which improves the accuracy of water monitoring and irrigation response efficiency in different areas and at different levels, thereby realizing intelligent, precise and sustainable operation and maintenance management of park-level green plant systems. Summary of the Invention

[0006] This invention proposes a smart control method for irrigation water controllers for high-density green plants in parks. Step S1 employs an interpolation completion mechanism based on sensor confidence scores to adaptively correct data differences at different vertical depths, achieving data integrity and reliability, and effectively improving sensor data quality. Step S2 utilizes a layered hydrological prior information evolution model to perform layered evolution of water content at different vertical depths, correcting the water content of each layer based on vegetation density and light intensity, achieving accurate characterization of water content at multiple vertical depths, and solving the problem of the inability to directly observe the hydrological state at multiple levels. Step S3 introduces shading correction factors and canopy humidity factors to improve the AquaCrop model, calculates the proportion of stratified irrigation and the daily irrigation amount at the grid level, and generates a precise stratified irrigation strategy to solve the problems of water waste and root under-irrigation caused by uniform irrigation. Step S4 introduces a weighted correction formula based on short-term rainfall and rainfall time to dynamically decay and correct the irrigation strategy over-irrigation, solving the problems of over-irrigation, resource waste, and root zone water accumulation caused by short-term rainfall.

[0007] To achieve the above objectives, the present invention provides a smart control method for a high-density green plant irrigation water controller in a park, comprising the following steps: S1: Collect depth sensor data of the green areas in the park at different vertical depths, and perform preprocessing on the depth sensor data based on interpolation completion of sensor confidence scores to obtain preprocessed depth sensor data. S2: The hydrological information evolution model of the layered hydrological prior information is used to perform hydrological information evolution on the preprocessed depth sensor data to obtain the water content of the green area in the park at different vertical depth layers. The residual neural network model based on vegetation density and light intensity is used to correct the water content layer by layer to obtain the corrected water content of the green area in the park at different vertical depth layers. S3: Convert the layer-by-layer corrected moisture content into effective moisture content and available water content for different vertical depth layers, and use the improved AquaCrop model to generate irrigation strategies for different vertical depth layers. S4: Based on the observation results of short-term rainfall, the irrigation strategy for different vertical depth layers is weighted and corrected, and the weighted and corrected irrigation strategy is sent to the green plant irrigation water controller, which executes the received weighted and corrected irrigation strategy.

[0008] As a further improvement of the present invention: Furthermore, depth sensor data was collected from the green areas of the park at different vertical depths, including: Based on the height of the green plants, the green areas of the park are divided into multiple vertical layers with different heights and depths, including the ground layer, the lower vegetation layer, the middle vegetation layer, and the upper vegetation layer. The horizontal plane corresponding to the green area of ​​the park is divided into multiple non-overlapping grids. The center of the grid is used as the sensor deployment location, and a depth sensor is placed at each sensor deployment location. The depth sensor is a combination sensor deployed at different vertical depth layers, including ground layer sensor, lower vegetation layer sensor, middle vegetation layer sensor, and upper vegetation layer sensor. The ground layer sensor, lower vegetation layer sensor, middle vegetation layer sensor, and upper vegetation layer sensor are deployed in sequence at the ground layer, lower vegetation layer, middle vegetation layer, and upper vegetation layer at the sensor deployment location, respectively, to sense the humidity data of the ground layer, the humidity data of the lower vegetation layer, the humidity data of the middle vegetation layer, as well as the vegetation density, the humidity data of the upper vegetation layer, and the light intensity data, respectively. The humidity data of the ground layer, the humidity data of the lower vegetation layer, the humidity data of the middle vegetation layer, as well as the vegetation density, the humidity data of the upper vegetation layer, and the light intensity data are used as depth sensor data at the sensor deployment location. The humidity data and vegetation data are in the form of sequence data. The depth sensor periodically uploads its data to the edge node. The edge node then calculates the sensor confidence score of the sensor corresponding to the depth sensor data and performs preprocessing to fill in missing data in the depth sensor data by interpolation based on the sensor confidence score.

[0009] Specifically, the formula for calculating the sensor confidence score is as follows: ; ; in, Indicates sensor Data collected at time t Sensor confidence score, Indicates sensor Static reliability score, Indicates sensor The percentage of missing depth sensor data collected in the previous cycle. Indicates sensor The collected data sequence is composed of data The mean of a window sequence centered at L and of length L. Representing data Data deviation, Indicates sensor The maximum allowable data deviation is set by the sensor. The absolute value of the maximum data difference between any two data points in the collected historical depth sensor data. All of these represent the scoring weight coefficients.

[0010] Furthermore, the depth sensor data undergoes preprocessing including interpolation completion based on sensor confidence scores, which includes: The system extracts the time and corresponding sensor of missing values ​​from the depth sensor data, and extracts the data collected by adjacent sensors and the sensor confidence score. It then uses weighted interpolation to fill in the missing values. Two adjacent sensors are defined as two sensors deployed in the same vertical depth layer, of the same sensor type, and whose deployment locations are less than a preset distance threshold. The edge node uploads the preprocessed depth sensor data to the campus server.

[0011] Furthermore, a layered hydrological prior information evolution model was used to perform hydrological information evolution on the preprocessed depth sensor data, obtaining the water content of the park's green areas at different vertical depths, including: The hierarchical hydrological prior information evolution model includes an input layer, a prior information extraction layer, an effective water saturation calculation layer, a water potential calculation layer, and a water content calculation layer. A layered hydrological prior information evolution model is used to perform hydrological information evolution on preprocessed depth sensor data, obtaining the water content at different vertical depth layers of the sensor deployment locations associated with the preprocessed depth sensor data. The hydrological information evolution process is as follows: S201: The input layer receives the preprocessed depth sensor data and the vegetation type at the sensor deployment location associated with the preprocessed depth sensor data, and extracts the average humidity data of each vertical depth layer. S202: The prior information extraction layer obtains the empirical transpiration of vegetation at the sensor deployment location in one day based on the vegetation type, and calculates the potential transpiration of vegetation at the sensor deployment location associated with the preprocessed depth sensor data. S203: Effective water saturation calculation layer: The effective water saturation of each vertical depth layer is calculated based on the average humidity data of each vertical depth layer. S204: The water potential calculation layer is based on the potential transpiration of vegetation, effective water saturation, and the height of the vertical depth layer. The hydraulic conductivity of each vertical depth layer is calculated based on Darcy's law, and the hydraulic conductivity is converted into water potential. Specifically, the formula for calculating the hydraulic conductivity of each vertical depth layer is as follows: ; in, This represents the water conductivity of the j-th vertical depth layer. This represents the saturated hydraulic conductivity when the effective water saturation reaches 100% under laboratory conditions. This represents the effective water saturation at the j-th vertical depth layer. This represents the hydraulic conductivity control coefficient; The water conductivity The formula for converting to water potential is: ; in, This represents the water potential at the j-th vertical depth. This represents the height of the j-th vertical depth layer. This represents the fitting parameters in the van Genuchten model that characterize the fitting of the permeation curve; S205: The water content calculation layer uses the water potential of each vertical depth layer to vertically evolve the historical water content calculation results to obtain the current water content of different vertical depth layers.

[0012] Specifically, the vertical evolution formula is: ; in, This represents the water content of the j-th vertical depth layer. This represents the water content of the j-th vertical depth layer observed based on the depth sensor data after the last preprocessing. This represents the water potential of the j-th vertical depth layer calculated based on the depth sensor data after the previous preprocessing.

[0013] Furthermore, a residual neural network model based on vegetation density and light intensity is used to correct the water content layer by layer, including: The residual neural network model includes an input layer, a residual block, and a residual connection unit. The input layer is used to receive the water content of different vertical depth layers, the average vegetation density and the average light intensity data in the preprocessed depth sensor data corresponding to the water content. The residual block consists of 8 fully connected layers and a ReLU activation function. Each pair of fully connected layers is associated with a vertical depth layer. The residual of the water content of the associated vertical depth layer is calculated based on the mean vegetation density and the mean light intensity data, resulting in the residual output of the water content. The residual calculation formula is as follows: ; in, Represents the water content of the j-th vertical depth layer. The residual output shows that the first to fourth vertical depth layers are, respectively, the ground layer, the lower vegetation layer, the middle vegetation layer, and the upper vegetation layer. The values ​​represent the mean vegetation density and mean light intensity data in the preprocessed depth sensor data corresponding to the water content, respectively. These represent the parameters of the trainable fully connected weight matrix associated with the j-th vertical depth layer. These represent the trainable fully connected bias parameters associated with the j-th vertical depth layer. Represents the ReLU activation function; The residual connection unit is used to add the water content of the vertical depth layer to the residual output as the correction result for the water content.

[0014] Furthermore, the layer-by-layer corrected moisture content is converted into the effective moisture content and usable water content of different vertical depth layers, including: Obtain the wilting water content of vegetation at various vertical depths in the green areas of the park; Based on the height of the vertical depth layer, the corrected water content is converted into volumetric water content as the effective water content of the vertical depth layer. Based on the height of the vertical depth layer, the wilting water content is converted into volumetric water content as the lower limit water content of the vertical depth layer, and the difference between the effective water content and the lower limit water content is taken as the usable water content of the vertical depth layer.

[0015] Furthermore, using the improved AquaCrop model to receive the effective water content and available water volume at different vertical depths, irrigation strategies for different vertical depths are generated, including: The strategy generation process for irrigation strategies at different vertical depths within the same grid is as follows: The water loss degree D of the grid is calculated based on the effective water content and available water volume of different vertical depth layers. If the water loss degree D is greater than 0, the irrigation strategy generation condition of the grid is triggered, and the irrigation allocation ratio of each vertical depth layer is calculated. Otherwise, the irrigation strategy of the grid is not generated in the current cycle, and the depth sensor data sent in the next cycle is continuously received. The AquaCrop model was improved by introducing a shading correction factor and a canopy humidity factor to generate the daily irrigation amount required for the grid. The irrigation amount was then split based on the irrigation allocation ratio to obtain the stratified irrigation amount for different vertical depth layers within the grid, which serves as the irrigation strategy for each vertical depth layer.

[0016] Furthermore, based on short-term rainfall observations, irrigation strategies at different vertical depths are weighted and adjusted, including: The weighted correction formula for the irrigation strategy at vertical depth is: ; in, This represents the stratified irrigation amount for the j-th vertical depth layer. The 1st to 4th vertical depth layers are, respectively, the ground layer, the lower vegetation layer, the middle vegetation layer, and the upper vegetation layer. Indicates stratified irrigation volume The weighted correction results This represents the observed short-term rainfall. This represents the control coefficient. This indicates a weighted average based on rainfall duration. This represents the weighted control coefficient. This represents the time interval between the observed short-term rainfall event and the current moment. This represents an exponential function with the natural constant as its base. Indicates selection The maximum value in, Indicates selection The maximum value in.

[0017] Compared with existing technologies, this invention proposes a smart control method for irrigation water controllers for high-density green plants in parks. This technology has the following beneficial effects: First, the layered hydrological prior information evolution model of this invention realizes continuous reasoning about vegetation moisture status by constructing a multi-layered layered calculation dynamic link. First, it uses vegetation type to obtain the potential transpiration of vegetation, and incorporates the differences in water absorption capacity of different vegetation into vegetation hydrological constraints in a priori form, so that the humidity response between different vertical depth layers is more in line with natural laws. Then, it calculates the effective water saturation based on the average humidity of multiple depth layers, obtains the hydraulic conductivity of each layer through Darcy's law, and further converts it into water potential, so as to accurately characterize the vertical water movement of each vertical depth layer in vegetation. The moisture content calculation layer adopts the evolution estimation of historical moisture content driven by water potential, so as to obtain the current moisture content distribution of multiple vertical depth layers.

[0018] Meanwhile, this invention introduces the average vegetation density and average light intensity as environmental adjustment factors to specifically compensate for the water content of each vertical depth layer, achieving refined correction of the water content of each vertical depth layer. Specifically, the residual block uses a two-layer fully connected network as a functional unit, which can learn the nonlinear disturbances to water content caused by environmental factors such as vegetation shading and light changes, and superimpose them on the water content in the form of residuals, so that the correction process maintains a structural dependence on physical priors and avoids non-physical deviations. Through independent modeling layer by layer, it can learn the water content correction law of inter-layer differences for the differentiated light transmittance and vegetation coverage of the ground layer, lower vegetation layer, middle layer and upper layer, significantly improving the accuracy and robustness of water content inversion under different vegetation coverage scenarios, and providing more reliable data for irrigation regulation. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a smart control method for a high-density green plant irrigation water controller in a park, as provided in an embodiment of the present invention.

[0020] Figure 2 This is a flowchart of hydrological information evolution provided in one embodiment of the present invention.

[0021] Figure 3 This is a schematic diagram of a residual neural network model structure provided in an embodiment of the present invention.

[0022] Figure 4 This is a schematic diagram of the structure of a green plant irrigation water controller provided in an embodiment of the present invention. Detailed Implementation

[0023] The realization of the objectives, functional characteristics, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0024] This invention provides a smart control method for a high-density green plant irrigation water controller in a park. The executing entity of this smart control method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this invention: a server, a terminal, etc. In other words, the smart control method for a high-density green plant irrigation water controller in a park can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0025] Reference Figure 1 , Figure 2 as well as Figure 3 Embodiment 1 of the present invention is as follows: A smart control method for irrigation water controllers for high-density green plants in a park, the method comprising: S1: Collect depth sensor data of the green areas in the park at different vertical depths, and perform preprocessing on the depth sensor data based on interpolation completion of sensor confidence scores to obtain preprocessed depth sensor data.

[0026] Depth sensor data was collected from different vertical depths of the park's green areas, including: Based on the height of the green plants, the green areas of the park are divided into multiple vertical layers with different heights and depths, including the ground layer, the lower vegetation layer, the middle vegetation layer, and the upper vegetation layer. As an embodiment of the present invention, the height range of the ground layer is set to 0 meters to 0.2 meters (inclusive), the height range of the lower vegetation layer is set to 0.2 meters to 1 meter (inclusive), the height range of the middle vegetation layer is set to 1 meter to 2 meters (inclusive), and the height range of the upper vegetation layer is set to more than 2 meters, and adjacent vertical depth layers do not overlap. Specifically, the ground layer mainly reflects soil moisture and microclimate changes in the lower layer; the lower vegetation layer reflects the humidity environment of the lower vegetation and affects the microbial activity of the lower vegetation; the middle vegetation layer is used to monitor the density of the middle vegetation layer; the upper vegetation layer reflects changes in light, humidity and wind speed in the canopy layer and is used for overall greening management and energy balance analysis; the vertical depth layer uses the typical height of vegetation as a reference, and takes into account the installability of sensors and the integrity of data coverage. By setting vertical depth layers of different heights, it is ensured that the data do not overlap with each other and can completely reflect the vertical microenvironment gradient. The horizontal plane corresponding to the green area of ​​the park is divided into multiple non-overlapping grids. The center of the grid is used as the sensor deployment location, and a depth sensor is placed at each sensor deployment location. The depth sensor is a combination sensor deployed at different vertical depth layers, including ground layer sensor, lower vegetation layer sensor, middle vegetation layer sensor, and upper vegetation layer sensor. The ground layer sensor, lower vegetation layer sensor, middle vegetation layer sensor, and upper vegetation layer sensor are deployed in sequence at the ground layer, lower vegetation layer, middle vegetation layer, and upper vegetation layer at the sensor deployment location, respectively, to sense the humidity data of the ground layer, the humidity data of the lower vegetation layer, the humidity data of the middle vegetation layer, as well as the vegetation density, the humidity data of the upper vegetation layer, and the light intensity data, respectively. Optionally, the grid is square in shape in the horizontal plane, and the grid width can be set to 5 meters; Specifically, the ground layer sensor is a soil moisture meter, which collects ground layer humidity data in real time by being directly buried or fixed to the ground layer; the lower vegetation layer sensor is an air humidity sensor, which collects lower vegetation layer humidity data in real time by being suspended among the branches and leaves of the lower vegetation layer; the middle vegetation layer sensor includes an air humidity sensor and a vegetation density sensor, which are suspended in the middle vegetation layer by a bracket to collect humidity data and vegetation density data of the middle vegetation layer in turn; the upper vegetation layer sensor includes an air humidity sensor and a light meter, which are fixed in the upper vegetation layer by a support rod to collect humidity data and light intensity data of the upper vegetation layer in turn. It should be noted that the vegetation density sensor indirectly measures the vegetation density per unit area of ​​the grid corresponding to the sensor's deployment location through optical principles. By placing a light source under the vegetation, the intensity of transmitted light reaching the vegetation density sensor is measured. The denser the vegetation, the weaker the transmitted light intensity, thus calculating the vegetation density. The formula for measuring vegetation density is as follows: ; in, Indicates vegetation density. This represents the intensity of transmitted light measured by the vegetation density sensor. This represents the transmitted reference light intensity when there is no vegetation at the sensor deployment location. This represents the transmitted light intensity attenuation coefficient, which is measured under laboratory conditions. Represents a logarithmic function with the natural constant as its base; The humidity data of the ground layer, the humidity data of the lower vegetation layer, the humidity data of the middle vegetation layer, as well as the vegetation density, the humidity data of the upper vegetation layer, and the light intensity data are used as depth sensor data at the sensor deployment location. The humidity data and vegetation data are in the form of sequence data. The depth sensor periodically uploads its data to the edge node. The edge node then calculates the sensor confidence score of the sensor corresponding to the depth sensor data and performs preprocessing to fill in missing data in the depth sensor data by interpolation based on the sensor confidence score.

[0027] As an embodiment of the present invention, the formula for calculating the sensor confidence score is as follows: ; ; in, Indicates sensor Data collected at time t Sensor confidence score, Indicates sensor Static reliability score, Indicates sensor The percentage of missing depth sensor data collected in the previous cycle. Indicates sensor The collected data sequence is composed of data The mean of a window sequence centered at L and of length L. Representing data Data deviation, Indicates sensor The maximum allowable data deviation is set by the sensor. The absolute value of the maximum data difference between any two data points in the collected historical depth sensor data. All represent the scoring weight coefficients, set They are 0.6 and 0.4 respectively; It should be noted that the sensor confidence score calculation formula combines the sensor's static reliability score with dynamic data deviation assessment to reflect the reliability of sensor data in real time. Specifically, this invention uses the proportion of historical missing data to correct static reliability, which can effectively identify data acquisition failures or abnormal sensors. At the same time, it calculates data deviation through window mean and combines it with the allowable deviation threshold to achieve adaptive detection of instantaneous abnormal or sudden data, thereby calculating the sensor confidence score. This takes into account both the continuity of time series and the characteristics of the sensor itself, making the subsequent interpolation and completion results more accurate and reliable. This improves the integrity and usability of multi-layer vertical depth sensor data in the monitoring of green plants in parks, and provides a stable and reliable data foundation.

[0028] Preprocessing of depth sensor data using interpolation completion based on sensor confidence scores includes: The system extracts the timestamps and corresponding sensors with missing values ​​from the depth sensor data, and also extracts data collected by adjacent sensors and their confidence scores. Weighted interpolation is used to fill in the missing values. Two adjacent sensors are defined as those deployed at the same vertical depth layer, of the same type, and with a distance between their deployment locations below a preset distance threshold. The edge nodes upload the preprocessed depth sensor data to the campus server. Specifically, the campus server executes subsequent steps S2 to S4.

[0029] It should be noted that the sensor types include soil moisture meters, air humidity sensors, vegetation density sensors, and light meters; Optionally, a preset distance threshold can be set to 2.5 times or 2 times the grid side length; As an embodiment of the present invention, if the sensor If no data is collected at time t+1, the formula for interpolating and completing the missing values ​​at time t+1 is: ; ; in, Indicates sensor Missing value completion result at time t+1 Indicates sensor The set of adjacent sensors, , Represents the set of adjacent sensors Any sensor in it, Indicates sensor Data collected at time t+1 Indicates sensor Data collected at time t+1 Sensor confidence score, Indicates sensor With sensors The positional weighting coefficients between them Indicates sensor With sensors The distance between deployment locations.

[0030] It should be noted that the interpolation completion formula proposed in this invention simultaneously integrates spatial neighborhood correlation, sensor confidence score, and time series continuity to achieve highly reliable completion of missing depth sensing data. Specifically, this invention introduces a distance weighting coefficient. This approach allows data collected by neighboring sensors that are closer to the main sensor to contribute more significantly, aligning with the physical characteristics of continuous variations in humidity, light, and density within a localized area of ​​vegetation. Furthermore, by combining the sensor confidence scores of data collected by neighboring sensors at the current moment, the impact of abnormal or low-quality data on the completion results can be dynamically suppressed, while also utilizing historical data from the sensors at the previous moment. Applying time smoothing constraints can effectively reduce abrupt completion errors and significantly improve the data integrity, stability, and reliability of depth sensor data in the monitoring of green plants in parks under multiple vertical depth layers.

[0031] S2: The hydrological information evolution model of the layered hydrological prior information is used to perform hydrological information evolution on the preprocessed depth sensor data to obtain the water content of the green area in the park at different vertical depth layers. The residual neural network model based on vegetation density and light intensity is used to correct the water content layer by layer to obtain the corrected water content of the green area in the park at different vertical depth layers.

[0032] A layered hydrological prior information evolution model was used to analyze the hydrological information evolution of preprocessed depth sensor data, obtaining the water content of the park's green areas at different vertical depths, including: The hierarchical hydrological prior information evolution model includes an input layer, a prior information extraction layer, an effective water saturation calculation layer, a water potential calculation layer, and a water content calculation layer. For reference Figure 2 The diagram illustrates a hydrological information evolution process. A layered hydrological prior information evolution model is used to perform hydrological information evolution on preprocessed depth sensor data, obtaining the water content at different vertical depth layers corresponding to the sensor deployment locations associated with the preprocessed depth sensor data. The hydrological information evolution process is as follows: S201: The input layer receives the preprocessed depth sensor data and the vegetation type at the sensor deployment location associated with the preprocessed depth sensor data, and extracts the average humidity data of each vertical depth layer. Specifically, the vegetation types include turfgrass, shrubs, and trees. Turfgrass includes cool-season turfgrass (such as Kentucky bluegrass and ryegrass) and warm-season turfgrass (such as bermudagrass and zoysiagrass), with an empirical transpiration rate of 5 mm per day, where empirical transpiration is the change in water depth per unit area per day. Shrubgrass includes flowering shrubs (such as rhododendrons and roses) and evergreen shrubs (such as boxwood and pittosporum), with an empirical transpiration rate of 4 mm per day. Treegrass includes small trees (such as cherry blossoms and crabapple trees), medium-sized trees (such as London plane trees and zelkova trees), and large trees (such as camphor trees and sycamore trees), with an empirical transpiration rate of 7 mm per day. S202: The prior information extraction layer obtains the empirical transpiration of vegetation at the sensor deployment location in one day based on the vegetation type, and calculates the potential transpiration of vegetation at the sensor deployment location associated with the preprocessed depth sensor data. As one embodiment of the present invention, the empirical transpiration of vegetation at the sensor deployment location within one day is obtained based on vegetation type, and based on the time interval between the last hydrological information evolution and the current time. The empirical transpiration rate is converted into the potential transpiration rate of vegetation over the time interval, where the conversion formula is: ; in, This represents the potential transpiration of vegetation during the time interval. This represents the empirical transpiration of vegetation at the sensor deployment location over one day. S203: Effective water saturation calculation layer: The effective water saturation of each vertical depth layer is calculated based on the average humidity data of each vertical depth layer. Specifically, the effective water saturation of each vertical depth layer is calculated as follows: ; in, This represents the effective water saturation at the j-th vertical depth layer. This represents the mean humidity data of the j-th vertical depth layer. This represents the maximum humidity data for vegetation determined in the laboratory environment. This represents the minimum humidity data for vegetation determined by the laboratory environment, where the 1st to 4th vertical depth layers are, respectively, the ground layer, the lower vegetation layer, the middle vegetation layer, and the upper vegetation layer. Optionally, the maximum and minimum humidity data for each vertical depth layer can be determined in a laboratory environment, and the effective water saturation of the corresponding vertical depth layer can be calculated. S204: The water potential calculation layer is based on the potential transpiration of vegetation, effective water saturation, and the height of the vertical depth layer. The hydraulic conductivity of each vertical depth layer is calculated based on Darcy's law, and the hydraulic conductivity (the seepage velocity of water flow) is converted into water potential. As an embodiment of the present invention, the formula for calculating the water conductivity of each vertical depth layer is as follows: ; in, This represents the water conductivity of the j-th vertical depth layer. This represents the saturated hydraulic conductivity when the effective water saturation reaches 100% under laboratory conditions. This represents the effective water saturation at the j-th vertical depth layer. Indicates the hydraulic conductivity control coefficient, set It is 0.5; The water conductivity The formula for converting to water potential is: ; in, This represents the water potential at the j-th vertical depth. This represents the height of the j-th vertical depth layer. This represents the fitting parameters in the van Genuchten model that characterize the fitting of the permeation curve; S205: The water content calculation layer uses the water potential of each vertical depth layer to vertically evolve the historical water content calculation results to obtain the current water content of different vertical depth layers.

[0033] As an embodiment of the present invention, the vertical evolution formula is: ; in, This represents the water content of the j-th vertical depth layer. This represents the water content of the j-th vertical depth layer observed based on the depth sensor data after the last preprocessing. This represents the water potential at the j-th vertical depth layer, calculated based on the depth sensor data after the previous preprocessing. Specifically, in the water potential calculation layer and the water content calculation layer, effective water saturation and Darcy flux (i.e., hydraulic conductivity) are introduced in the vertical direction. The calculation process (and van Genuchten infiltration curves) allows the water migration process at different vertical depths to be constrained by physical laws, rather than directly relying on instantaneous humidity values ​​from sensors, thereby constructing a continuous, smooth, and stable groundwater profile; and through... The nonlinear expression maps the effective water saturation to permeability (hydraulic conductivity), which can reflect the significantly different permeability changes between layers at different vertical depths. Furthermore, the driving force of water between layers at vertical depths is captured by calculating the water potential. During the water cut evolution process, the water potential gradient is introduced into the historical water cut update, so that the upward and downward movement trends of water between layers can be quantified, thereby avoiding non-physical water cut jumps caused by sensor noise, and finally obtaining a more stable and reliable water cut. Among them, water content represents the proportion of soil pores filled with water, which is used to characterize the water storage of each vertical depth layer; water potential represents the driving force for water migration into the soil, which is the essential indicator for determining the flow of water from high potential to low potential; hydraulic conductivity represents the seepage capacity of water in the soil under a unit water potential gradient, reflecting the water permeability of different vertical depth layers at the current saturation level.

[0034] A residual neural network model based on vegetation density and light intensity was used to correct the water content layer by layer, including: For reference Figure 3 The diagram shows the structure of the residual neural network model. The residual neural network model includes an input layer, a residual block, and a residual connection unit. The input layer is used to receive the water content of different vertical depth layers, the average vegetation density and the average light intensity data in the preprocessed depth sensor data corresponding to the water content. The residual block consists of 8 fully connected layers and a ReLU activation function. Each pair of fully connected layers is associated with a vertical depth layer. The residual of the water content of the associated vertical depth layer is calculated based on the mean vegetation density and the mean light intensity data, resulting in the residual output of the water content. The residual calculation formula is as follows: ; in, Represents the water content of the j-th vertical depth layer. The residual output shows that the first to fourth vertical depth layers are, respectively, the ground layer, the lower vegetation layer, the middle vegetation layer, and the upper vegetation layer. The values ​​represent the mean vegetation density and mean light intensity data in the preprocessed depth sensor data corresponding to the water content, respectively. These represent the parameters of the trainable fully connected weight matrix associated with the j-th vertical depth layer. These represent the trainable fully connected bias parameters associated with the j-th vertical depth layer. Represents the ReLU activation function; The residual connection unit is used to add the water content of the vertical depth layer to the residual output as the correction result for the water content.

[0035] Specifically, by collecting the true water content of multiple vegetation areas at each vertical depth layer and the pre-processed depth sensor data, the water content of each vertical depth layer in the pre-processed depth sensor data is extracted as described in steps S1 and S2, and then corrected. The training loss function is constructed with the goal of minimizing the absolute value of the difference between the true water content and the corrected water content. The trainable parameters in the residual neural network model are trained and optimized using the gradient descent algorithm or the Adam optimizer.

[0036] S3: Convert the layer-by-layer corrected moisture content into effective moisture content and available water volume for different vertical depth layers, and use the improved AquaCrop model to generate irrigation strategies for different vertical depth layers.

[0037] The layer-by-layer corrected moisture content is converted into effective moisture content and usable water content for different vertical depth layers, including: Obtain the wilting water content of vegetation at various vertical depths in the green areas of the park; Specifically, the wilting moisture content is the soil moisture content when the vegetation is permanently wilted due to its inability to absorb water from the soil. The wilting moisture content of each vertical depth layer is measured when the vegetation roots are unable to absorb water. Based on the height of the vertical depth layer, the corrected water content is converted into volumetric water content as the effective water content of the vertical depth layer. Specifically, the volumetric water content is the water content per unit layer volume, and the conversion formula for the corrected water content is: ; in, This represents the corrected water content of the j-th vertical depth layer. This represents the height of the j-th vertical depth layer. This represents the effective water content of the j-th vertical depth layer; Based on the height of the vertical depth layer, the wilting water content is converted into volumetric water content as the lower limit water content of the vertical depth layer, and the difference between the effective water content and the lower limit water content is taken as the usable water content of the vertical depth layer.

[0038] Specifically, the conversion formula for the wilting moisture content is: ; in, This represents the wilting water content of the j-th vertical depth layer. This represents the lower limit of water content at the j-th vertical depth layer; The formula for calculating the available water volume is as follows: ; in, This represents the amount of usable water in the j-th vertical depth layer. Indicates selection The maximum value in.

[0039] An improved AquaCrop model is used to receive the effective water content and available water volume at different vertical depths, generating irrigation strategies for these depths, including: The strategy generation process for irrigation strategies at different vertical depths within the same grid is as follows: The water loss degree D of the grid is calculated based on the effective water content and available water volume of different vertical depth layers. If the water loss degree D is greater than 0, the irrigation strategy generation condition of the grid is triggered, and the irrigation allocation ratio of each vertical depth layer is calculated. Otherwise, the irrigation strategy of the grid is not generated in the current cycle, and the depth sensor data sent in the next cycle is continuously received. As an embodiment of the present invention, the formula for calculating the degree of water loss D is: ; in, Represents the scaling factor, set It is 0.85. Indicates selection The maximum value in; The formula for calculating the irrigation allocation ratio for each vertical depth layer is as follows: ; ; in, This represents the irrigation allocation ratio for the j-th vertical depth layer. Indicates the corrected moisture content The spurious factor is identified when the corrected moisture content of the lower vertical depth layer is lower than that of the upper vertical depth layer. In this case, the corrected moisture content of the lower vertical depth layer is considered potentially spurious, and the corresponding irrigation allocation ratio is reduced. Indicates the spurious weighting coefficient, set It is 0.4. This represents the height ratio of the j-th vertical depth layer; Optionally, when j is 4, set =1; The AquaCrop model was improved by introducing a shading correction factor and a canopy humidity factor to generate the daily irrigation amount required for the grid. The irrigation amount was then split based on the irrigation allocation ratio to obtain the stratified irrigation amount for different vertical depth layers within the grid, which serves as the irrigation strategy for each vertical depth layer.

[0040] Specifically, the formula for generating the daily irrigation amount required by the grid is as follows: ; in, This indicates the daily irrigation requirement for the grid. This represents the empirical transpiration of vegetation at the sensor deployment location over one day. This represents the normalization function, which can be set to the Sigmoid function. Indicates the shading correction factor. This represents the mean vegetation density in the preprocessed depth sensor data. Indicates crown humidity factor, All represent the irrigation volume generation coefficient, set They are 0.6 and 0.4 respectively. Indicates selection The maximum value in.

[0041] It should be noted that this invention, by introducing effective and available water content at different vertical depths, a spurious factor weighting mechanism, and a dual correction factor, enables the irrigation strategy to be adaptive at both the vertical and vegetation environment scales, thereby significantly improving the reliability of irrigation decisions and resource utilization efficiency. Specifically, the water loss level is constructed based on the overall difference between the effective and available water content of the four layers. The irrigation strategy is only triggered when D>0, avoiding unnecessary irrigation and improving the water-saving performance of the park's greenery. Based on the spurious factor, when the water content of the lower vertical depth layer is abnormally lower than that of the upper layer, potential calculation anomalies can be automatically identified, and the corresponding... The irrigation allocation ratio is suppressed, thus ensuring that irrigation resources are not misallocated due to mismeasurement, and improving the robustness of the stratified irrigation strategy. In addition, by introducing shading correction factors and canopy humidity factors, the transpiration in the AuaCrop model is dynamically adjusted under vegetation density and canopy humidity conditions, so that the daily irrigation demand can reflect the actual microenvironment evapotranspiration characteristics, significantly improving the environmental adaptability of irrigation estimation. Thus, a highly refined stratified irrigation strategy is realized based on the irrigation allocation ratio, so that the precise water replenishment capacity can still be maintained in heterogeneous soil structures and complex vegetation cover scenarios, improving the overall irrigation efficiency and vegetation health level. Furthermore, the shading correction factor is used to characterize the weakening effect of vegetation density on surface evapotranspiration demand, and the shading correction factor is used to characterize the amplifying effect of the moisture gradient between the upper and lower parts of the canopy on evapotranspiration intensity, reflecting the physical mechanism between canopy humidity structure and stomatal conductance. It should be explained that by dividing the horizontal plane corresponding to the green area of ​​the park into multiple non-overlapping grids, the calculation results of the grids are independent of each other in the subsequent calculation process, thus obtaining the irrigation strategies of different grids at each vertical depth layer.

[0042] S4: Based on the observation results of short-term rainfall, the irrigation strategy for different vertical depth layers is weighted and corrected, and the weighted and corrected irrigation strategy is sent to the green plant irrigation water controller, which executes the received weighted and corrected irrigation strategy.

[0043] Based on short-term rainfall observations, irrigation strategies at different vertical depths were weighted and adjusted, including: The weighted correction formula for the irrigation strategy at vertical depth is: ; in, This represents the stratified irrigation amount (in millimeters) for the j-th vertical depth layer. Indicates stratified irrigation volume The weighted correction results This represents the observed short-term rainfall (in millimeters). Indicates the control coefficient, set It is 0.1. This indicates a weighted average based on rainfall duration. Indicates the weighted control coefficient, set It is 0.2. This represents the time interval between the observed short-term rainfall event and the current moment. This represents an exponential function with the natural constant as its base. Indicates selection The maximum value in, Indicates selection The maximum value in.

[0044] It should be noted that the weighted correction formula, by introducing a short-term rainfall and rainfall time decay mechanism (weighting based on rainfall time), can enhance the real-time adaptability of the irrigation strategy. Specifically, through... It enables immediate response to sudden or short-duration high-intensity rainfall. When rainfall significantly offsets a water deficit at a certain depth, it can automatically reduce or even eliminate the corresponding irrigation amount, avoiding deep leaching and root zone hypoxia caused by over-irrigation; through Encoding the natural decay of rainfall effects over time, the term tends to 1 the further away the rainfall time is from the current time, thus effectively capturing the temporal differences such as rapid evaporation of shallow water and slow release of deep water, further improving the precision of vertical water regulation, and ultimately achieving more water-saving and more robust stratified irrigation control.

[0045] Example 2: For reference Figure 4The diagram shows the structure of a green plant irrigation water controller 100. The green plant irrigation water controller 100 includes a main water inlet pipe 1, a PLC controller 2, four independent drive circuits 3, four solenoid valve groups 4, a stratified water distributor 5, a ground layer water supply branch pipe 6, a lower vegetation layer water supply branch pipe 7, a middle vegetation layer water supply branch pipe 8, and an upper vegetation layer water supply branch pipe 9. The dashed lines in the structural diagram represent signal transmission circuits, which are used to transmit signals. The main water inlet pipe 1 serves as the water input end of the entire green plant irrigation water controller 100. One end is connected to a water source (such as a water storage tank or water pipe), and the other end is connected to the subsequent control components. It is responsible for introducing water from the water source into the green plant irrigation water controller. At the same time, a simple filter can be installed on the pipeline to prevent impurities from clogging the downstream valves and drip irrigation heads. The PLC controller 2 has a built-in storage unit and programming module, which is responsible for receiving the weighted and corrected irrigation strategy and issuing precise irrigation instructions to multiple components. The precise irrigation instructions are in the form of low-voltage control signals. The four independent drive circuits 3 serve as a transfer station between the PLC controller 2 and the execution components (four-way solenoid valve group 4, stratified water distributor 5, ground layer water supply branch pipe 6, lower vegetation layer water supply branch pipe 7, middle vegetation layer water supply branch pipe 8, and upper vegetation layer water supply branch pipe 9), converting the weak current control signals output by the PLC controller 2 into strong current drive signals, which respectively drive the start / stop and opening / closing degree adjustment of the four solenoid valves in the four-way solenoid valve group 4. The four-way solenoid valve group 4 consists of four independent solenoid valves, each corresponding to a water supply branch pipe. After receiving the drive circuit signal, the water flow rate and irrigation time of the corresponding water supply branch pipe are controlled by adjusting the valve opening degree. The stratified water distributor 5 has 4 independent water distribution channels inside, which can smoothly distribute the water flow regulated by the solenoid valve to 4 different water delivery branch pipes, while also playing a role in stabilizing pressure and avoiding water flow fluctuations from affecting the accuracy of irrigation volume. One end of the ground floor water supply branch pipe 6 is connected to the stratified water distributor 5, and the other end is connected to the ground floor drip irrigation head, which is responsible for delivering the diverted water flow to the ground floor. One end of the water supply branch pipe 7 under the vegetation is connected to the stratified water distributor 5, and the other end is connected to the drip irrigation head under the vegetation, which is responsible for delivering the diverted water flow to the under vegetation layer. One end of the vegetation middle layer water supply branch pipe 8 is connected to the stratified water distributor 5, and the other end is connected to the vegetation middle layer drip irrigation head, which is responsible for delivering the diverted water flow to the vegetation middle layer; One end of the upper vegetation water supply branch pipe 9 is connected to the layered water distributor 5, and the other end is connected to the upper vegetation drip irrigation head, which is responsible for delivering the diverted water flow to the upper vegetation layer.

[0046] It should be noted that the sequence numbers of the above embodiments of the present invention are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0047] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0048] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A smart control method for irrigation water controllers for high-density green plants in a park, characterized in that, The method includes: S1: Collect depth sensor data of the green areas in the park at different vertical depths, and perform preprocessing on the depth sensor data based on interpolation completion of sensor confidence scores to obtain preprocessed depth sensor data. S2: The hydrological information evolution model of the layered hydrological prior information is used to perform hydrological information evolution on the preprocessed depth sensor data to obtain the water content of the green area in the park at different vertical depth layers. The residual neural network model based on vegetation density and light intensity is used to correct the water content layer by layer to obtain the corrected water content of the green area in the park at different vertical depth layers. S3: Convert the layer-by-layer corrected moisture content into effective moisture content and available water content for different vertical depth layers, and use the improved AquaCrop model to generate irrigation strategies for different vertical depth layers. S4: Based on the observation results of short-term rainfall, the irrigation strategy for different vertical depth layers is weighted and corrected, and the weighted and corrected irrigation strategy is sent to the green plant irrigation water controller, which executes the received weighted and corrected irrigation strategy.

2. The intelligent control method for a high-density green plant irrigation water controller in a park as described in claim 1, characterized in that, Depth sensor data was collected from different vertical depths of the park's green areas, including: Based on the height of the green plants, the green areas of the park are divided into multiple vertical layers with different heights and depths, including the ground layer, the lower vegetation layer, the middle vegetation layer, and the upper vegetation layer. The horizontal plane corresponding to the green area of ​​the park is divided into multiple non-overlapping grids. The center of the grid is used as the sensor deployment location, and a depth sensor is placed at each sensor deployment location. The depth sensor is a combination sensor deployed at different vertical depth layers, including ground layer sensor, lower vegetation layer sensor, middle vegetation layer sensor, and upper vegetation layer sensor. The ground layer sensor, lower vegetation layer sensor, middle vegetation layer sensor, and upper vegetation layer sensor are deployed in sequence at the ground layer, lower vegetation layer, middle vegetation layer, and upper vegetation layer at the sensor deployment location, respectively, to sense the humidity data of the ground layer, the humidity data of the lower vegetation layer, the humidity data of the middle vegetation layer, as well as the vegetation density, the humidity data of the upper vegetation layer, and the light intensity data, respectively. The humidity data of the ground layer, the humidity data of the lower vegetation layer, the humidity data of the middle vegetation layer, as well as the vegetation density, the humidity data of the upper vegetation layer, and the light intensity data are used as depth sensor data at the sensor deployment location. The humidity data and vegetation data are in the form of sequence data. The depth sensor periodically uploads its data to the edge node. The edge node then calculates the sensor confidence score of the sensor corresponding to the depth sensor data and performs preprocessing to fill in missing data in the depth sensor data by interpolation based on the sensor confidence score.

3. The intelligent control method for a high-density green plant irrigation water controller in a park as described in claim 2, characterized in that, Preprocessing of depth sensor data using interpolation completion based on sensor confidence scores includes: The system extracts the time and corresponding sensor of missing values ​​from the depth sensor data, and extracts the data collected by adjacent sensors and the sensor confidence score. It then uses weighted interpolation to fill in the missing values. Two adjacent sensors are defined as two sensors deployed in the same vertical depth layer, of the same sensor type, and whose deployment locations are less than a preset distance threshold. The edge node uploads the preprocessed depth sensor data to the campus server.

4. The intelligent control method for a high-density green plant irrigation water controller in a park as described in claim 1, characterized in that, A layered hydrological prior information evolution model was used to analyze the hydrological information evolution of preprocessed depth sensor data, obtaining the water content of the park's green areas at different vertical depths, including: The hierarchical hydrological prior information evolution model includes an input layer, a prior information extraction layer, an effective water saturation calculation layer, a water potential calculation layer, and a water content calculation layer. A layered hydrological prior information evolution model is used to perform hydrological information evolution on preprocessed depth sensor data, obtaining the water content at different vertical depth layers of the sensor deployment locations associated with the preprocessed depth sensor data. The hydrological information evolution process is as follows: S201: The input layer receives the preprocessed depth sensor data and the vegetation type at the sensor deployment location associated with the preprocessed depth sensor data, and extracts the average humidity data of each vertical depth layer. S202: The prior information extraction layer obtains the empirical transpiration of vegetation at the sensor deployment location in one day based on the vegetation type, and calculates the potential transpiration of vegetation at the sensor deployment location associated with the preprocessed depth sensor data. S203: Effective water saturation calculation layer: The effective water saturation of each vertical depth layer is calculated based on the average humidity data of each vertical depth layer. S204: The water potential calculation layer is based on the potential transpiration of vegetation, effective water saturation, and the height of the vertical depth layer. The hydraulic conductivity of each vertical depth layer is calculated based on Darcy's law, and the hydraulic conductivity is converted into water potential. S205: The water content calculation layer uses the water potential of each vertical depth layer to vertically evolve the historical water content calculation results to obtain the current water content of different vertical depth layers.

5. The intelligent control method for a high-density green plant irrigation water controller in a park as described in claim 4, characterized in that, A residual neural network model based on vegetation density and light intensity was used to correct the water content layer by layer, including: The residual neural network model includes an input layer, a residual block, and a residual connection unit. The input layer is used to receive the water content of different vertical depth layers, the average vegetation density and the average light intensity data in the preprocessed depth sensor data corresponding to the water content. The residual block consists of 8 fully connected layers and a ReLU activation function. Each pair of fully connected layers is associated with a vertical depth layer. The residual of the water content of the associated vertical depth layer is calculated based on the mean vegetation density and the mean light intensity data to obtain the residual output of water content. The residual connection unit is used to add the water content of the vertical depth layer to the residual output as the correction result for the water content.

6. The intelligent control method for a high-density green plant irrigation water controller in a park as described in claim 5, characterized in that, The layer-by-layer corrected moisture content is converted into effective moisture content and usable water content for different vertical depth layers, including: Obtain the wilting water content of vegetation at various vertical depths in the green areas of the park; Based on the height of the vertical depth layer, the corrected water content is converted into volumetric water content as the effective water content of the vertical depth layer. Based on the height of the vertical depth layer, the wilting water content is converted into volumetric water content as the lower limit water content of the vertical depth layer, and the difference between the effective water content and the lower limit water content is taken as the usable water content of the vertical depth layer.

7. The intelligent control method for a high-density green plant irrigation water controller in a park as described in claim 6, characterized in that, An improved AquaCrop model is used to receive the effective water content and available water volume at different vertical depths, generating irrigation strategies for these depths, including: The strategy generation process for irrigation strategies at different vertical depths within the same grid is as follows: The water loss degree D of the grid is calculated based on the effective water content and available water volume of different vertical depth layers. If the water loss degree D is greater than 0, the irrigation strategy generation condition of the grid is triggered, and the irrigation allocation ratio of each vertical depth layer is calculated. Otherwise, the irrigation strategy of the grid is not generated in the current cycle, and the depth sensor data sent in the next cycle is continuously received. The AquaCrop model was improved by introducing a shading correction factor and a canopy humidity factor to generate the daily irrigation amount required for the grid. The irrigation amount was then split based on the irrigation allocation ratio to obtain the stratified irrigation amount for different vertical depth layers within the grid, which serves as the irrigation strategy for each vertical depth layer.

8. The intelligent control method for a high-density green plant irrigation water controller in a park as described in claim 7, characterized in that, Based on short-term rainfall observations, irrigation strategies at different vertical depths were weighted and adjusted, including: The weighted correction formula for the irrigation strategy at vertical depth is: ； in, This represents the stratified irrigation amount for the j-th vertical depth layer. The 1st to 4th vertical depth layers are, respectively, the ground layer, the lower vegetation layer, the middle vegetation layer, and the upper vegetation layer. Indicates stratified irrigation volume The weighted correction results This represents the observed short-term rainfall. Indicates the control coefficient. This indicates a weighted average based on rainfall duration. This represents the weighted control coefficient. This represents the time interval between the observed short-term rainfall event and the current moment. This represents an exponential function with the natural constant as its base. Indicates selection The maximum value in, Indicates selection The maximum value in.

Images