Water-saving cultivation method for citrus based on green production

Abstract

The present application relates to the field of agricultural planting irrigation technology, in particular to a kind of citrus water-saving cultivation method based on green production.The method obtains root system spatial distribution dielectric constant matrix through multi-channel time domain reflection sensor array, obtains wet body three-dimensional volume data through soil moisture dielectric sensor;Establish the infiltration model with drip irrigation working pressure and pulse duration as independent variables;Root system spatial distribution dielectric constant matrix and wet body three-dimensional volume data are calculated by spatial overlap rate, when the overlap rate of specific depth soil layer is lower than the preset threshold, the working pressure and pulse width of corresponding depth drip irrigation pipe are calculated and adjusted according to the infiltration model, so that the wetting front migration trajectory and high-density root system distribution space coincide.The present application eliminates the spatial misplacement of fixed wet body and irregular root system distribution, reduces the non-root area invalid water infiltration and deep seepage.

Description

Technical Field

[0001] This invention relates to the field of agricultural planting and irrigation technology, specifically to a water-saving cultivation method for citrus based on green production. Background Technology

[0002] Current drip irrigation technology in citrus orchards mainly employs constant operating pressure and fixed flow rate, or pulse irrigation with a set pulse duration ratio. During irrigation, water drips into the soil and migrates under the influence of gravity and capillary force, forming a moist body. Existing control logic relies on a single soil moisture sensor buried at a fixed depth or location to obtain moisture content data. When the moisture content falls below a set threshold, the water pump and solenoid valve are activated; when the moisture content reaches the set upper limit, the equipment is shut off. Throughout the irrigation cycle, the outlet pressure of the drip irrigation pipeline remains constant, and the single outlet time follows preset fixed parameters. This water supply mode determines that the infiltration process of water in the soil profile is only controlled by the initial soil texture and constant outlet boundary conditions. The trajectory of water migration to the surrounding areas and downwards, and the final shape of the moist body, exhibit static and fixed physical characteristics.

[0003] The core problem with existing constant-pressure, fixed-parameter irrigation methods lies in the spatial misalignment between the fixed wetting body formed and the actual three-dimensional distribution of citrus roots. Citrus roots grow unevenly in the soil, exhibiting significant differences in root density across different radial distances and soil depths. Because the outlet pressure and pulse duration are fixed during irrigation in existing technologies, the resulting wetting front migration trajectory remains constant, failing to actively converge towards high-density roots. This leads to water redundancy in localized areas covered by the fixed wetting body, even seeping deeper into the root zone, while lateral or deep high-density root distribution areas remain untouched by the wetting body, resulting in a physical misalignment between the spatial distribution of irrigation water and the space for root absorption. Summary of the Invention

[0004] The purpose of this invention is to provide a water-saving citrus cultivation method based on green production, which can effectively solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A water-saving citrus cultivation method based on green production includes: acquiring the root spatial distribution dielectric constant matrix through a multi-channel time-domain reflectometry sensor array during the citrus planting period;

[0007] During the irrigation period, three-dimensional volume data of the wetted body around the drip irrigation head are obtained through a soil moisture dielectric sensor.

[0008] An infiltration model was established with drip irrigation working pressure and pulse duration as independent variables and soil hydrodynamic parameters as dependent variables.

[0009] The spatial overlap rate of the dielectric constant matrix of the root system spatial distribution and the three-dimensional volume data of the wetted body is calculated.

[0010] When the overlap rate between the root density and the moisture content of the soil at a specific depth is lower than a preset threshold, the working pressure and pulse width of the drip irrigation pipe at the corresponding depth are calculated and adjusted according to the infiltration model, so that the movement trajectory of the wetting front coincides with the spatial distribution of high-density roots.

[0011] Preferably, the step of obtaining the root spatial distribution dielectric constant matrix includes: establishing a three-dimensional rectangular coordinate system with the center of the base of the citrus tree trunk as the origin, and arranging the multi-channel time-domain reflectometry sensor array in the three-dimensional rectangular coordinate system according to the rule of increasing radial distance and increasing depth layer by layer;

[0012] Within a set acquisition period, the dielectric constant measurement value of each sensor probe in the multi-channel time-domain reflectometry sensor array is recorded;

[0013] The dielectric constant measurement value of each sensor probe is fused and mapped with the spatial coordinates in the three-dimensional rectangular coordinate system to generate the root spatial distribution dielectric constant matrix containing spatial coordinate information and dielectric constant information.

[0014] Preferably, the step of obtaining the three-dimensional volume data of the wet body includes: arranging the soil moisture dielectric sensor around the drip irrigation head according to the three-dimensional grid nodes, and collecting the soil volume moisture content of each grid node according to the set sampling frequency during irrigation;

[0015] The soil volumetric water content of each grid node is calculated by three-dimensional spatial interpolation to generate a continuously distributed soil water content spatial field.

[0016] The set of spatial points in the soil moisture content spatial field whose soil volumetric moisture content is equal to the sum of the initial moisture content before irrigation and the set moisture content increment is defined as the moist front, and the spatial region enclosed by the moist front constitutes the three-dimensional volume data of the moist body.

[0017] Preferably, the steps for establishing the infiltration model include: obtaining the initial moisture content, saturated moisture content, and soil hydraulic diffusivity of the citrus root zone soil;

[0018] The drip irrigation working pressure is converted into the drip irrigation head water flow rate, and the water flow rate and the pulse duration are substituted as input variables into the correction equation based on the Green-Ampt model;

[0019] Using the initial moisture content, the saturated moisture content, and the soil hydraulic diffusivity as parameter variables in the modified equation, the modified equation is solved to obtain the functional relationship between the advancing distance of the wetting front and the cumulative infiltration amount, and the functional relationship is confirmed as the infiltration model.

[0020] Preferably, the step of calculating the spatial overlap rate includes: dividing the three-dimensional space where the root spatial distribution dielectric constant matrix is ​​located and the three-dimensional space where the three-dimensional volume data of the wetted body is located into three-dimensional meshes of the same scale according to a preset voxel size;

[0021] For each divided 3D mesh voxel, the root density feature represented by the dielectric constant and the wetted body feature represented by the volume water content are extracted.

[0022] The proportion of grid voxels that simultaneously possess the root density characteristic and the moist body characteristic is determined to be the root density and moist body water overlap rate of the soil layer at the specific depth.

[0023] Preferably, the step of adjusting the working pressure and pulse width of the drip irrigation pipe at the corresponding depth includes: extracting the difference between the root density and the overlap rate of the wetted body of the soil layer at the specific depth at the current moment and the preset threshold;

[0024] The difference is input into the infiltration model as a target approximation variable, and the target outflow rate and target pulse duration corresponding to the target wetting front advance distance are obtained by inverse solution through the infiltration model;

[0025] The adjustment increment of the working pressure of the drip irrigation pipe at the corresponding depth is calculated based on the target outflow rate, and the adjustment increment of the pulse width is calculated based on the target pulse duration. The adjustment increment of the working pressure and the adjustment increment of the pulse width are then output to the pressure regulating valve and the solenoid valve.

[0026] Preferably, after generating the root spatial distribution dielectric constant matrix, the method further includes: obtaining the current phenological stage of the citrus and retrieving the root growth expansion prediction model corresponding to the phenological stage;

[0027] The grid coordinates in the dielectric constant matrix of the root system spatial distribution are input into the root system growth and expansion prediction model, and the root extension direction vector and extension length prediction value corresponding to each grid coordinate in the phenological stage are output.

[0028] The root spatial distribution dielectric constant matrix is ​​coordinately translated based on the root extension direction vector and the predicted extension length, and the translated dielectric constant measurement value is weighted and attenuated based on the radial distance from the origin, thereby generating the dynamically corrected root spatial distribution dielectric constant matrix.

[0029] Preferably, the step of constituting the three-dimensional volume data of the wet body further includes: during the interval between two adjacent pulse durations, switching the sampling frequency of the soil moisture dielectric sensor to the interval sampling frequency;

[0030] Based on the soil volumetric water content of each grid node collected at the intermittent sampling frequency, the three-dimensional spatial interpolation calculation is re-executed to generate the intermittent soil water content spatial field;

[0031] In the intermittent period, spatial points are extracted from the soil moisture content spatial field where the rate of decrease in soil volume moisture content is equal to a preset critical value. These spatial points are connected to form a water redistribution boundary. The water redistribution boundary is then used to replace the wetting front to update the three-dimensional volume data of the wetting body.

[0032] Preferably, after confirming the infiltration model, the method further includes: during the pause phase of the pulse duration, extracting the actual soil volumetric water content decrease curves of multiple spatial nodes at the wetting front;

[0033] The actual soil volumetric moisture content decrease curve is input into the soil hydraulic diffusivity inversion algorithm to calculate the actual soil hydraulic diffusivity corresponding to the resting stage.

[0034] The actual soil hydraulic diffusivity is used to replace the soil hydraulic diffusivity in the modified equation, and the parameter variables in the infiltration model are locally updated and calibrated.

[0035] In subsequent irrigation cycles, the infiltration model, after local updates and calibration, is used for reverse solving.

[0036] Preferably, the step of outputting the adjustment increment of the working pressure and the adjustment increment of the pulse width to the pressure regulating valve and the solenoid valve further includes: dividing the shallow root distribution area and the deep root distribution area, and obtaining the first overlap rate corresponding to the shallow root distribution area and the second overlap rate corresponding to the deep root distribution area respectively.

[0037] When the first overlap rate meets the preset threshold and the second overlap rate is lower than the preset threshold, the adjustment increment of the working pressure of the shallow drip irrigation pipe is reduced and the adjustment increment of the pulse width is shortened, while the adjustment increment of the working pressure of the deep drip irrigation pipe is increased and the adjustment increment of the pulse width is lengthened.

[0038] The shallow drip irrigation tube and the deep drip irrigation tube are controlled to be staggered in timing from the pulse duration.

[0039] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0040] 1. This invention establishes an infiltration model with drip irrigation working pressure and pulse duration as independent variables. It calculates the spatial overlap rate between the root spatial distribution dielectric constant matrix and the three-dimensional volume data of the wetting body within a three-dimensional grid voxel. When the overlap rate is lower than a preset threshold, the working pressure and pulse width of the corresponding depth drip irrigation pipe are adjusted by reverse calculation based on the infiltration model. This method transforms the adjustment of irrigation parameters from a fixed time dimension to a spatial three-dimensional matching dimension. By changing the outlet pressure and pulse duration, it directly reshapes the wetting front migration trajectory, ensuring that the dynamically generated wetting body spatially matches the high-density root distribution area, eliminating the spatial misalignment between the fixed wetting body and the irregular root distribution.

[0041] 2. This invention introduces a root growth and expansion prediction model corresponding to phenological stages, performs coordinate translation and weight attenuation correction on the dielectric constant matrix of the root spatial distribution, and adapts the calculation benchmark of spatial overlap rate to the dynamic extension of the root system. During the pulse interval, spatial points where the rate of decrease in soil volumetric water content equals a preset critical value are connected to form a water redistribution boundary and replace the original wetting front, updating the three-dimensional volumetric data of the wetting body. By extracting the actual soil volumetric water content decrease curve during the interval, the actual soil hydraulic diffusivity is solved by inversion and the infiltration model parameters are updated and calibrated. Combined with the division of shallow and deep root distribution areas to calculate the corresponding overlap rate, the shallow drip irrigation pipes and deep drip irrigation pipes are controlled to stagger the pulse duration in time, realizing dynamic tracking of changes in soil hydrodynamic parameters and stratified asynchronous water supply. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating the overall process of a water-saving citrus cultivation method based on green production according to the present invention.

[0043] Figure 2 This is a flowchart of the process for obtaining the root spatial distribution dielectric constant matrix of the present invention;

[0044] Figure 3 This is a flowchart of the process for acquiring three-dimensional volume data of the wetted body according to the present invention;

[0045] Figure 4 This is a flowchart illustrating the establishment and dynamic calibration of the infiltration model of the present invention.

[0046] Figure 5 This is a flowchart of the spatial overlap rate calculation of the present invention;

[0047] Figure 6This is a flowchart of the working pressure and pulse width adjustment and layered control of the present invention. Detailed Implementation

[0048] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0049] Please refer to Figure 1 This embodiment provides a water-saving citrus cultivation method based on green production. During the citrus planting period, a multi-channel time-domain reflectometry (TDRS) sensor array is deployed to obtain the dielectric constant matrix of the root system's spatial distribution. Specifically, a three-dimensional rectangular coordinate system is established with the center of the citrus trunk base as the origin. The X-axis is set along the extension direction of the citrus planting row, the Y-axis is set perpendicular to the planting row direction, and the Z-axis is set along the vertically downward soil depth direction. The spatial coordinates of the origin are (0,0,0).

[0050] In a three-dimensional Cartesian coordinate system, a multi-channel time-domain reflectometry sensor array is arranged according to the rule of increasing radial distance and increasing depth layer by layer, with each sensor corresponding to a unique spatial coordinate. The absolute values ​​of x and y increase synchronously with the radial distance from the origin, while the value of z increases synchronously with the soil depth. The dual probes of each sensor are parallel to... The probe is inserted into the corresponding soil layer depth on the horizontal plane. The measurement range of the probe completely covers the preset voxel area of ​​the corresponding spatial location, avoiding the matching error between the dielectric constant measurement value and the spatial coordinate caused by the measurement range spanning voxels.

[0051] Within a set acquisition period, the dielectric constant measurement value of each sensor probe in the multi-channel time-domain reflectometry (TDAR) sensor array is recorded. The duration of the acquisition period is matched with the minimum time step of citrus root growth to avoid data redundancy caused by excessively high acquisition frequency, while also avoiding missing root distribution changes due to excessively low acquisition frequency. The dielectric constant measurement value of each sensor probe is fused and mapped with spatial coordinates in a three-dimensional Cartesian coordinate system to generate a root spatial distribution dielectric constant matrix containing both spatial coordinate information and dielectric constant information. The dimension of the root spatial distribution dielectric constant matrix corresponds to the number of sensor array elements. The row and column of each element in the matrix correspond to the spatial coordinate index of the sensor, and the value of the element is the dielectric constant measurement value at the corresponding spatial location.

[0052] Table 1 Spatial Layout Parameters of Multichannel Time Domain Reflectance Sensor Array

[0053]

[0054] This table is used to clarify the spatial position of each sensor in the multi-channel time-domain reflectometry sensor array and its corresponding acquisition channel, providing a coordinate reference for the spatial mapping of the dielectric constant matrix of the root system spatial distribution, ensuring the correspondence between the dielectric constant measurement value and the three-dimensional spatial position, and eliminating matrix mapping errors caused by spatial misalignment.

[0055] During the irrigation period, three-dimensional volume data of the wetting body around the drip irrigation head are acquired using soil moisture dielectric sensors. Soil moisture dielectric sensors are deployed in the three-dimensional space around the drip irrigation head, with each sensor corresponding to a grid node in the three-dimensional space. The grid nodes are arranged in a cylindrical coordinate system with the water outlet position of the drip irrigation head as the center, and are distributed at equal angles along the circumference, at equal intervals in the radial direction, and at equal intervals in the depth. The layout density of the grid nodes matches the spatial resolution of soil water infiltration to ensure that continuous changes in the wetting front can be captured.

[0056] During irrigation, soil volumetric moisture content at each grid node is collected at a set sampling frequency. The sampling frequency is matched to the duration of the drip irrigation pulse, and a sampling frequency matching the infiltration rate is used within the pulse duration to ensure real-time capture of the dynamic expansion process of the wetting body. Three-dimensional spatial interpolation is performed on the collected soil volumetric moisture content of each grid node to generate a continuously distributed soil moisture spatial field. The three-dimensional spatial interpolation calculation uses an interpolation method based on spatial distance weights to ensure that the deviation between the interpolated soil moisture spatial field and the measured node data is within a preset allowable range. The set of spatial points in the soil moisture spatial field whose soil volumetric moisture content equals the sum of the initial moisture content before irrigation and the set moisture content increment is defined as the wetting front. The closed spatial region enclosed by the wetting front constitutes the three-dimensional volumetric data of the wetting body. The three-dimensional volumetric data of the wetting body includes the coordinate information of all spatial points within this closed spatial region and the corresponding soil volumetric moisture content information.

[0057] An infiltration model was established with drip irrigation operating pressure and pulse duration as independent variables and soil hydrodynamic parameters as dependent variables. Specifically, the initial moisture content, saturated moisture content, and soil hydraulic diffusivity of the citrus root zone soil were first obtained. These parameters were determined through laboratory soil sample tests or in-situ field tests. The drip irrigation operating pressure was converted into drip head outflow rate. A positive correlation was found between the drip head outflow rate and the drip irrigation operating pressure, a relationship pre-calibrated through drip head hydraulic performance testing. The outflow rate and pulse duration were substituted as input variables into a modified equation based on the Green-Ampt model. The modified equation was optimized for the three-dimensional infiltration characteristics of point source drip irrigation, considering the coupling effect of radial diffusion and vertical infiltration of the drip head outflow. Initial moisture content, saturated moisture content, and soil hydraulic diffusivity were used as parameter variables in the modified equation. Solving the modified equation yielded a functional relationship between the wetting front advance distance and the cumulative infiltration amount, which was then confirmed as the infiltration model. The output of the infiltration model consists of soil hydrodynamic parameters, including the wetting front advance distance, cumulative infiltration amount, and soil moisture content distribution. The input variables are drip irrigation working pressure and pulse duration, achieving a quantitative mapping between input variables and output parameters.

[0058] The modified Green-Ampt model for point-source infiltration in drip irrigation is shown in equation (1):

[0059] (1)

[0060] in, The cumulative infiltration amount over time t. For soil saturated hydraulic conductivity, Soil saturation water content, This refers to the initial soil moisture content. The radius of the drip irrigation head's outlet hole. For time The distance the moist front advances, Soil hydraulic diffusivity, Soil volumetric water content, The coordinates represent the depth direction of the soil layer.

[0061] The conversion relationship between drip irrigation working pressure and outflow rate is shown in formula (2):

[0062] (2)

[0063] in, This refers to the water flow rate from the drip irrigation head. The flow rate coefficient of the drip irrigation head is obtained through pre-calibration. This reduces the workload of drip irrigation.

[0064] The spatial overlap rate of the root spatial distribution dielectric constant matrix and the three-dimensional volume data of the wetting body is calculated. The three-dimensional space containing the root spatial distribution dielectric constant matrix and the three-dimensional space containing the wetting body volume data are divided into three-dimensional meshes of the same scale according to a preset voxel size. The preset voxel size matches the spacing of the sensor array to ensure that each three-dimensional mesh voxel contains at most the measurement data from one sensor, avoiding calculation errors caused by the same measurement data being assigned to multiple voxels. For each divided three-dimensional mesh voxel, the root density feature represented by the dielectric constant and the wetting body feature represented by the volumetric water content are extracted. The root density feature is determined by comparing the measured dielectric constant value with a preset dielectric constant threshold; when the measured dielectric constant value within a voxel is greater than the preset dielectric constant threshold, the voxel is determined to possess root density characteristics. The wetting body feature is determined by comparing the soil volumetric water content within a voxel with the water content defined by the wetting front; when the soil volumetric water content within a voxel is greater than or equal to the water content defined by the wetting front, the voxel is determined to possess wetting body characteristics. The number of grid voxels that simultaneously possess root density and wet body characteristics is counted, and the proportion of these voxels to the total number of grid voxels possessing root density characteristics is determined. This proportion is then identified as the overlap rate between root density and wet body moisture in a specific soil depth.

[0065] The overlap rate between root density and moisture content in a specific soil depth is calculated as shown in formula (3):

[0066] (3)

[0067] in, This represents the overlap between root density and moisture content in a specific soil depth. This refers to the number of grid voxels within a soil layer at this depth that simultaneously possess both root density and wet volume characteristics. This represents the total number of grid voxels with root density characteristics within the soil layer at this depth.

[0068] When the overlap rate between the root density and the moisture content of the soil at a specific depth is lower than a preset threshold, the working pressure and pulse width of the drip irrigation pipe at the corresponding depth are calculated and adjusted according to the infiltration model so that the movement trajectory of the wetting front coincides with the spatial distribution of high-density roots.

[0069] In a specific example, an experiment was conducted in a citrus orchard with sandy loam soil and citrus trees that were 4 years old. At the planting period, a multi-channel time-domain reflectometry (TDRS) sensor array was deployed with the base of the tree trunk as the origin, covering a radial distance of 0.1m to 1.8m and a depth of 0.1m to 0.7m, with a data collection period of 5 days. Dielectric constant conversion calculations showed that the high-density root system was mainly located in the soil layer with a radial distance of 0.3m to 1.0m and a depth of 0.2m to 0.5m. During irrigation, soil moisture dielectric sensors were deployed around the drip irrigation heads, forming a three-dimensional grid. The initial drip irrigation pressure was 0.07MPa, and the pulse duration was 12 minutes. Soil volumetric moisture content was collected in real time during irrigation, generating a soil moisture spatial field and defining the three-dimensional volumetric data of the wetting body. The root distribution matrix and the wetting body data were divided into a three-dimensional grid using the same voxels, and the overlap rate of the soil layer at a depth of 0.2m to 0.5m was calculated. The initial value was 0.48, lower than the preset threshold of 0.70. The difference was input into the infiltration model for inverse solving, yielding a target outflow rate of 1.6 L / h. This corresponded to an operating pressure of 0.10 MPa and a target pulse duration extended to 22 minutes. Based on this, the operating pressure and pulse width of the drip irrigation pipes at the corresponding depth were adjusted. Data was collected again after irrigation, and the overlap rate of the soil layer at the same depth increased to 0.78. Two consecutive months of irrigation trials showed that, compared to the control area using a constant pressure of 0.07 MPa and a 12-minute pulse, this example reduced citrus irrigation water consumption by approximately 21%, decreased deep infiltration by approximately 31%, and maintained fruit yield and quality indicators roughly the same as the control area, demonstrating a significant water-saving and efficiency-enhancing effect.

[0070] The preset threshold is set in advance according to the water requirement characteristics of citrus at different growth stages. When the overlap rate of a certain depth soil layer is lower than the preset threshold, it is determined that the high-density root distribution area in the soil layer at that depth is not effectively covered by the moist body, and the irrigation parameters of the drip irrigation pipe at the corresponding depth need to be adjusted.

[0071] The difference between the root density and the overlap rate of the moist body at a specific soil depth at the current moment and a preset threshold is extracted. This difference is used as a target approximation variable and input into the infiltration model. The target outflow rate and target pulse duration corresponding to the target moist front advance distance are obtained by inversely solving the infiltration model. The adjustment increment of the working pressure of the drip irrigation pipe at the corresponding depth is calculated based on the target outflow rate, and the adjustment increment of the pulse width is calculated based on the target pulse duration. The adjustment increments of the working pressure and pulse width are output to the pressure regulating valve and the solenoid valve. The working pressure of the drip irrigation pipe is adjusted by the pressure regulating valve, and the pulse width is adjusted by the solenoid valve, thereby realizing the active control of the moist front movement trajectory. This ensures that the adjusted moist front movement trajectory coincides with the high-density root distribution space, ensuring that the moist body can effectively cover the high-density root distribution area.

[0072] The objective function for the infiltration model obtained by reverse solution is shown in Equation (4):

[0073] (4)

[0074] in, The target overlap rate corresponding to the preset threshold. To use the outflow rate Pulse duration When the input variable is , the overlap rate corresponding to the wetted body output by the infiltration model. This refers to the water flow rate from the drip irrigation head.

[0075] In this embodiment, the dielectric constant matrix of the root spatial distribution is obtained through a multi-channel time-domain reflectometry sensor array, and the three-dimensional volume data of the wet body is obtained through a soil moisture dielectric sensor. An infiltration model is established with drip irrigation working pressure and pulse duration as independent variables. The spatial overlap rate is used to quantify the spatial matching degree between the root distribution and the wet body. Irrigation parameters are adjusted by inversely solving the infiltration model to achieve the overlap between the wet front movement trajectory and the high-density root distribution space, thereby eliminating the spatial misalignment between the fixed wet body and the irregular root distribution.

[0076] In an alternative embodiment, refer to Figure 2 In generating the dielectric constant matrix of the root system spatial distribution, a three-dimensional rectangular coordinate system was established with the center of the citrus trunk base as the origin. Within this system, a multi-channel time-domain reflectometry (TDRS) sensor array was deployed according to a rule of increasing radial distance and increasing depth layer by layer. Specifically, the increasing radial distance step size matched the growth spacing of the citrus horizontal roots, and the increasing depth step size matched the layered distribution characteristics of the citrus vertical roots. Within a radial distance of 0.5m around the trunk base, the deployment density was 0.1m radial step and 0.1m depth step; within a radial distance of 0.5m to 1.5m, the density was 0.2m radial step and 0.2m depth step; and within a radial distance of 1.5m to 3.0m, the density was 0.3m radial step and 0.3m depth step. This ensured higher spatial resolution in the near-trunk region where roots were densely distributed, while reducing the number of sensors deployed in the distant trunk region to control data redundancy.

[0077] Within a set acquisition period, the dielectric constant measurement value of each sensor probe in the multi-channel time-domain reflectometry sensor array is recorded. The acquisition period is adjusted according to the phenological stages of citrus. During the vigorous root growth stages such as spring shoots, summer shoots, and autumn shoots, the acquisition period is shortened; during the slow root growth stages such as fruit enlargement and ripening, the acquisition period is extended to ensure that the acquired data matches the dynamic rate of root growth. The dielectric constant measurement value of each sensor probe is fused and mapped with the spatial coordinates in a three-dimensional rectangular coordinate system to generate a root spatial distribution dielectric constant matrix containing spatial coordinate information and dielectric constant information. During the fusion and mapping process, multiple dielectric constant measurements at the same spatial location are subjected to moving average filtering to eliminate the interference of measurement noise on the matrix values. The window width of the moving average filter is matched with the number of acquisition periods to ensure that the filtered values ​​reflect the true state of root distribution.

[0078] Table 2. Parameters of Root Growth and Expansion Model for Citrus at Different Phenological Stages

[0079] Phenological stage Root radial growth rate (m / d) Vertical root growth rate (m / d) Radial growth weighting coefficient Vertical growth weighting coefficient Weight decay coefficient Data collection period (d) budding stage 0.002 0.0015 0.6 0.4 0.8 7 Spring shoots 0.005 0.003 0.7 0.3 0.6 3 Summer shoots 0.006 0.004 0.65 0.35 0.5 3 Autumn shoots 0.004 0.0035 0.55 0.45 0.6 4 Fruit enlargement period 0.0015 0.0025 0.4 0.6 0.7 10 Fruit ripening period 0.0008 0.001 0.3 0.7 0.9 15 Flower bud differentiation period 0.001 0.0012 0.5 0.5 0.85 10

[0080] This table is used to clarify the root growth characteristic parameters and collection control parameters corresponding to different phenological stages of citrus, providing benchmark data for the input of the root growth expansion prediction model, and providing a basis for adjusting the collection cycle of dielectric constant measurement values, so as to ensure that the dynamic correction of the dielectric constant matrix of root spatial distribution can match the root growth pattern of different phenological stages.

[0081] After generating the root spatial distribution dielectric constant matrix, the current phenological stage of the citrus tree is obtained, and the root growth and expansion prediction model corresponding to that phenological stage is retrieved. Specifically, the phenological stages of citrus include budding, spring shoot, summer shoot, autumn shoot, fruit enlargement, fruit ripening, and flower bud differentiation. Each phenological stage corresponds to a unique root growth and expansion prediction model, which is trained using historical citrus root growth data within the corresponding phenological stage. The model's input is the grid coordinates of the root spatial distribution, and its output is the root extension direction vector and predicted extension length within the corresponding grid coordinates. The grid coordinates from the root spatial distribution dielectric constant matrix are input into the root growth and expansion prediction model, which outputs the root extension direction vector and predicted extension length for each grid coordinate within that phenological stage. The root extension direction vector includes a radial horizontal component and a vertical depth component, corresponding to the root growth trend in the horizontal and depth directions, respectively. The predicted extension length is the length of root growth in that extension direction within the corresponding phenological stage.

[0082] coordinate The root extension direction vector at the location is calculated as shown in formula (5):

[0083] (5)

[0084] in, coordinates The root extension direction vector at that location. coordinates Radial vector on the horizontal plane, Let be the magnitude of the radial vector. It is a vertically downward unit vector. This is the radial growth weighting coefficient. This is the vertical growth weighting coefficient. and The values ​​are predetermined based on the root growth characteristics of the corresponding phenological period.

[0085] The predicted root extension length within time t is calculated as shown in formula (6):

[0086] (6)

[0087] in, For time Predicted root extension length within the area This represents the radial growth rate of the root system. This represents the vertical growth rate of the root system. and These are the radial and vertical growth weighting coefficients corresponding to the phenological stages.

[0088] The dielectric constant matrix of the root spatial distribution is coordinately translated based on the root extension direction vector and the predicted extension length. The translation vector and distance are consistent with the root extension direction vector and the predicted extension length, ensuring that the translated matrix coordinates match the actual spatial location of the roots after growth. The translated dielectric constant measurements are weighted and attenuated based on their radial distance from the origin. The weight attenuation coefficient decreases with increasing radial distance and adjusts with increasing soil depth, ensuring that root distribution data near the trunk have higher weights, while the weights of root distribution data further from the trunk decrease appropriately with increasing distance, avoiding matrix errors caused by over-predicting root distribution in distant areas. A dynamically corrected dielectric constant matrix of the root spatial distribution is generated. This dynamically corrected matrix reflects the dynamic growth trend of the roots during the current phenological period, providing a real-time updated root distribution benchmark for calculating spatial overlap.

[0089] The weighted attenuation correction for the dielectric constant measurement is shown in Equation (7):

[0090] (7)

[0091] in, The measured dielectric constant is the value after weight attenuation correction. The dielectric constant is the measured value after coordinate translation. This is the weight decay coefficient. coordinates The radial distance from the origin of the coordinate system.

[0092] In this embodiment, by matching the sensor array layout rules with root growth characteristics and adjusting the acquisition cycle, a high-precision root spatial distribution dielectric constant matrix is ​​generated. Combined with the root growth extension prediction model corresponding to different phenological stages, the matrix is ​​subjected to coordinate translation and weight attenuation correction, so that the calculation benchmark of root distribution can adapt to the dynamic extension of the root system, improve the accuracy of spatial overlap rate calculation, and provide a basis for adjusting irrigation parameters that is more in line with the actual root growth state.

[0093] In an alternative embodiment, refer to Figure 3 and Figure 4 In the process of acquiring three-dimensional volume data of the wetted body, soil moisture dielectric sensors are arranged around the drip irrigation head according to a three-dimensional grid node. The arrangement of the three-dimensional grid nodes is centered on the water outlet position of the drip irrigation head and is evenly distributed along the horizontal radial direction and the vertical depth direction. The horizontal radial arrangement range covers the maximum horizontal extension range of the drip irrigation wetted body, and the vertical depth arrangement range covers the maximum distribution depth of the citrus root system. The spacing of the grid nodes matches the spatial resolution of soil water infiltration to ensure that the soil moisture content gradient between adjacent nodes can be accurately captured.

[0094] During irrigation, soil volumetric moisture content at each grid node was collected at a set sampling frequency. The sampling frequency was matched with the duration of the drip irrigation pulse. Within the pulse duration, the sampling frequency was set to ensure that the number of samples taken within a single pulse duration was no less than 5, achieving continuous capture of the wetting process. Three-dimensional spatial interpolation was performed on the soil volumetric moisture content of each grid node to generate a continuously distributed soil moisture spatial field. The three-dimensional spatial interpolation calculation adopted the Kriging interpolation method. During the interpolation process, the spatial heterogeneity of soil texture was considered, and corresponding variograms were set for soil regions with different textures to ensure that the interpolated soil moisture spatial field accurately reflects the spatial distribution of soil water.

[0095] The set of spatial points in the soil moisture content spatial field whose soil volumetric moisture content equals the sum of the initial moisture content before irrigation and the set moisture content increment is defined as the moist front. The set moisture content increment is predetermined based on soil texture and the water requirements of citrus, ensuring that the moist front can accurately define the boundary between moist body and non-moist soil. The spatial area enclosed by the moist front constitutes the three-dimensional volume data of the moist body. The three-dimensional volume data of the moist body includes the spatial volume of the moist body, the coordinate set of the moist front, and the soil volumetric moisture content information of each spatial point within the moist body.

[0096] Table 3. Parameters for the 3D Grid Layout of Soil Moisture Dielectric Sensor

[0097]

[0098] This table is used to clarify the three-dimensional spatial layout of the soil moisture dielectric sensor and the corresponding sampling control parameters, providing a node coordinate reference for the acquisition of soil volumetric water content and three-dimensional spatial interpolation, and providing parameter basis for switching the sampling frequency between the pulse period and the intermittent period, so as to ensure that the acquisition of three-dimensional volumetric data of the wet body can match the different stages of the irrigation process.

[0099] In the process of constructing three-dimensional volumetric data of the wetted body, during the interval between two adjacent pulse durations, the sampling frequency of the soil moisture dielectric sensor is switched to the interval sampling frequency. The interval sampling frequency is lower than the sampling frequency during the pulse duration. This reduces the power consumption and data redundancy of data acquisition while ensuring that the soil moisture redistribution process can be captured. Based on the soil volumetric water content of each grid node collected at the interval sampling frequency, three-dimensional spatial interpolation calculation is re-performed to generate the interval soil moisture spatial field. The interval soil moisture spatial field can reflect the redistribution process of soil water under the action of gravity and capillary force after the pulse irrigation stops.

[0100] Spatial points where the rate of decrease in soil volumetric moisture content equals a preset critical value are extracted from the spatial field of soil moisture content during the intermittent period. The preset critical value is determined in advance based on the soil hydraulic properties and corresponds to the boundary features between the wet body and the non-wet area during the soil water redistribution process. The extracted spatial points are connected to form the water redistribution boundary. The water redistribution boundary replaces the original wet front and updates the three-dimensional volume data of the wet body. The updated three-dimensional volume data of the wet body can accurately reflect the morphological changes of the wet body caused by the redistribution of soil water during the intermittent period, avoiding the calculation error of the wet body volume caused by the original wet front.

[0101] The rate of decrease in soil volumetric water content is calculated as shown in formula (8):

[0102] (8)

[0103] in, The time t coordinate The rate of decrease in soil volumetric moisture content at that location Let be the soil volumetric water content at time t. for Soil volumetric moisture content over time The sampling time interval during the interval period.

[0104] After confirming the steps of the infiltration model, during the pause phase of the pulse duration, the actual soil volumetric water content decrease curves of multiple spatial nodes at the moist front are extracted. The actual soil volumetric water content decrease curves are continuous curves showing the change of soil volumetric water content of the corresponding spatial nodes over time during the pause phase. The number of extracted spatial nodes is no less than 3 and they are evenly distributed along the moist front to ensure that the soil hydrodynamic changes at the moist front can be fully reflected.

[0105] The actual soil volumetric moisture content decrease curve is input into the soil hydraulic diffusivity inversion algorithm to calculate the actual soil hydraulic diffusivity corresponding to the resting phase. The algorithm is based on the one-dimensional Richards equation for soil water movement and calculates the corresponding soil hydraulic diffusivity value by inverting the measured moisture content change curve. The actual soil hydraulic diffusivity is then used to replace the soil hydraulic diffusivity in the correction equation, and the parameters in the infiltration model are locally updated and calibrated. The scope of this local update and calibration is limited to the control area of ​​the corresponding drip irrigation head to avoid parameter calibration errors caused by differences in soil properties in different areas. In subsequent irrigation cycles, the locally updated and calibrated infiltration model is used for inverse solving to ensure that the parameters of the infiltration model match the dynamic changes in soil hydraulic properties, thereby improving the accuracy of the infiltration model solution.

[0106] The soil hydraulic diffusivity inversion control equation based on the Richards equation is shown in Equation (9):

[0107] (9)

[0108] in, Soil volumetric water content, For time, The coordinates are in the direction of soil depth. The soil hydraulic diffusivity is determined by actual measurement. Follow and The change data is obtained through inversion solution. The actual value.

[0109] The calibration relationship for updating the infiltration model parameters is shown in formula (10):

[0110] (10)

[0111] in, To update the calibrated soil hydraulic diffusivity, To correct the initial soil hydraulic diffusivity function in the equation, The actual soil hydraulic diffusivity obtained from the inversion calculation, This is a reference value for the initial soil hydraulic diffusivity function at the corresponding moisture content.

[0112] In this embodiment, soil volumetric water content is collected by a soil moisture dielectric sensor deployed through a three-dimensional grid. The soil moisture spatial field is generated by three-dimensional spatial interpolation, and the three-dimensional volume data of the wet body is defined. During the pulse interval, the water content decrease rate is used to determine the water redistribution boundary and update the wet body data. At the same time, the soil hydraulic diffusivity is inverted by the measured water content decrease curve during the rest phase, and the infiltration model parameters are locally updated and calibrated. This enables dynamic tracking of changes in soil hydrodynamic parameters and improves the solution accuracy of the infiltration model and the accuracy of the wet body data.

[0113] In an alternative embodiment, refer to Figure 5 and Figure 6 In establishing the infiltration model, the initial moisture content, saturated moisture content, and soil hydraulic diffusivity of the citrus root zone soil were obtained. The initial moisture content was the measured volumetric moisture content of the root zone soil before irrigation, and the saturated moisture content was the volumetric moisture content of the root zone soil under saturated conditions, obtained through in-situ field ring cutter testing. The soil hydraulic diffusivity was a diffusivity function within the corresponding moisture content range of the root zone soil, obtained through indoor soil column tests. The drip irrigation working pressure was converted into the drip head outlet flow rate. The functional relationship between the drip head outlet flow rate and the drip irrigation working pressure was pre-calibrated through the hydraulic performance test of the drip head. The calibration process covered the normal working pressure range of the drip irrigation pipe to ensure the accuracy of flow conversion across the entire pressure range. The outlet flow rate and pulse duration were substituted as input variables into the modified equation based on the Green-Ampt model. The modified equation was optimized for the three-dimensional infiltration characteristics of point source drip irrigation, considering the coupling effect of radial diffusion and vertical infiltration of the drip head outlet, and correcting the simulation bias of the one-dimensional Green-Ampt model for the three-dimensional infiltration process. Using initial moisture content, saturated moisture content, and soil hydraulic diffusivity as parameter variables in the modified equation, the modified equation is solved to obtain the functional relationship between the advancing distance of the wetting front and the cumulative infiltration. This functional relationship is confirmed as the infiltration model. The infiltration model can quantitatively describe the dynamic changes of soil water infiltration and the migration trajectory of the wetting front under different combinations of drip irrigation working pressure and pulse duration.

[0114] During the spatial overlap calculation, the three-dimensional space containing the dielectric constant matrix of the root spatial distribution and the three-dimensional space containing the three-dimensional volume data of the wetting body are divided into three-dimensional grids of the same scale according to a preset voxel size. The preset voxel size is determined comprehensively based on the arrangement spacing of the sensor array and the spatial resolution of soil water infiltration, ensuring that the divided three-dimensional grid voxels can simultaneously match the spatial resolution of the root distribution and the spatial resolution of the wetting body. For each divided three-dimensional grid voxel, the root density feature represented by the dielectric constant and the wetting body feature represented by the volume water content are extracted. In the process of extracting the root density feature, a quantitative conversion relationship between the measured value of the dielectric constant and the root volume density is established. The root distribution state within the voxel is determined by the converted root volume density value. When the root volume density within the voxel is greater than the preset density threshold, the voxel is determined to have root density features. In the process of extracting the wetting body feature, when the voxel is located within the spatial region enclosed by the wetting front, the voxel is determined to have wetting body features. The proportion of grid voxels that simultaneously possess root density and wet body characteristics to the total number of grid voxels possessing root density characteristics is statistically analyzed. This proportion is then identified as the overlap rate between root density and wet body moisture at a specific soil depth. The division of specific soil depths corresponds to the layered layout depth of drip irrigation pipes, ensuring that each layered drip irrigation pipe corresponds to a unique specific soil depth, thereby enabling the calculation of layered overlap rates and adjustment of irrigation parameters.

[0115] The conversion relationship between dielectric constant and root volume density is shown in equation (11):

[0116] (11)

[0117] in, Root volume density This is the measured value of the dielectric constant. and The conversion factor was determined in advance through calibration experiments using a mixture of citrus root and soil samples.

[0118] During the adjustment of the working pressure and pulse width of the drip irrigation pipe at the corresponding depth, the difference between the root density and the overlap rate of the wetting body at a specific soil depth at the current moment and a preset threshold is extracted. The preset threshold is determined in advance based on the water requirement characteristics and root distribution characteristics of citrus at different phenological stages. The preset threshold for different soil depths can be set differently according to the root distribution density. This difference is input into the infiltration model as a target approximation variable. The target outflow rate and target pulse duration corresponding to the target wetting front advance distance are obtained through the inverse solution of the infiltration model. The inverse solution process adopts a nonlinear least squares optimization algorithm to ensure that the obtained target outflow rate and target pulse duration can make the overlap rate reach above the preset threshold.

[0119] The adjustment increment of the working pressure of the drip irrigation pipe at the corresponding depth is calculated based on the target outflow rate, and the adjustment increment of the pulse width is calculated based on the target pulse duration. The adjustment increments of the working pressure and the pulse width are output to the pressure regulating valve and the solenoid valve. The pressure regulating valve adjusts the working pressure of the drip irrigation pipe in real time according to the adjustment increment, and the solenoid valve adjusts the pulse width in real time according to the adjustment increment, thereby realizing closed-loop control of the drip irrigation process.

[0120] The calculation of the pressure adjustment increment for drip irrigation is shown in formula (12):

[0121] (12)

[0122] in, To adjust the working pressure increment of drip irrigation, To obtain the target effluent flow rate through reverse engineering, The flow rate coefficient of the drip irrigation head. This reflects the current pressure on drip irrigation operations.

[0123] The pulse width adjustment increment is calculated as shown in formula (13):

[0124] (13)

[0125] in, This is the adjustment increment for the pulse width. The target pulse duration is obtained by reverse engineering. This represents the pulse duration at the current moment.

[0126] In the step of outputting the adjustment increments of working pressure and pulse width to the pressure regulating valve and solenoid valve, the shallow root distribution area and the deep root distribution area are divided. The division depth of the shallow root distribution area and the deep root distribution area is predetermined based on the vertical distribution characteristics of citrus roots, with a soil depth of 0.4m as the boundary. Above 0.4m is the shallow root distribution area, and below 0.4m is the deep root distribution area. The first overlap rate corresponding to the shallow root distribution area and the second overlap rate corresponding to the deep root distribution area are obtained respectively. The first overlap rate is the weighted average of the overlap rates of all soil layers in the shallow root distribution area, and the second overlap rate is the weighted average of the overlap rates of all soil layers in the deep root distribution area. The weighting weight is the sum of the root density in the corresponding soil layer.

[0127] When the first overlap rate meets a preset threshold and the second overlap rate is lower than a preset threshold, the adjustment increment of the working pressure of the shallow drip irrigation pipe is reduced and the adjustment increment of the pulse width is shortened to reduce the irrigation water volume in the shallow area, avoiding water redundancy in the shallow area and leakage in the deep area. At the same time, the adjustment increment of the working pressure of the deep drip irrigation pipe is increased and the adjustment increment of the pulse width is lengthened to increase the irrigation water volume in the deep area, promoting the movement of the wetting front towards the deep root distribution area and improving the overlap rate in the deep area. The pulse duration of the shallow and deep drip irrigation pipes is staggered in time to avoid interference between soil water infiltration caused by simultaneous water output from the shallow and deep drip irrigation systems, ensuring that the formation processes of the shallow and deep wetting bodies are independent of each other, and improving the control accuracy of the wetting front movement trajectory.

[0128] Table 4. Timing Control and Parameter Adjustment of Shallow and Deep Drip Irrigation Pipes

[0129]

[0130] This table is used to clarify the current working parameters, target adjustment parameters, and timing control parameters of shallow and deep drip irrigation pipes, providing a quantitative basis for adjusting drip irrigation working pressure and pulse width. At the same time, it provides an offset benchmark for the pulse timing staggered control of shallow and deep drip irrigation pipes, avoiding mutual interference between the infiltration processes of shallow and deep irrigation.

[0131] The weighted average overlap rate of the shallow and deep root distribution areas is calculated as shown in formulas (14) and (15):

[0132] (14)

[0133] (15)

[0134] in, This represents the first overlap rate corresponding to the shallow root distribution area. This represents the second overlap rate corresponding to the deep root distribution area. This is the depth that marks the boundary between the shallow and deep root distribution areas. This represents the maximum depth at which the root system is distributed. For depth Overlap rate of soil layers This represents the total root density of the soil layer at depth z.

[0135] In this embodiment, an infiltration model is established by optimizing the three-dimensional infiltration characteristics of point source drip irrigation using a modified equation. The root density characteristics are accurately extracted by quantitatively converting the dielectric constant and root volume density, and the spatial overlap rate of the layers is calculated. The adjustment increment of irrigation parameters is obtained by inversely solving the infiltration model. At the same time, shallow and deep root distribution areas are divided. Based on the difference in the corresponding overlap rate, the differentiated parameter adjustment and time-sequence staggered control of the deep and shallow drip irrigation pipes are realized, further improving the matching degree between the spatial distribution of irrigation water and the root absorption space.

Claims

1. A water-saving citrus cultivation method based on green production, characterized in that, include: During the citrus planting period, the spatial distribution dielectric constant matrix of the root system was obtained through a multi-channel time-domain reflectometry sensor array. During the irrigation period, three-dimensional volume data of the wetted body around the drip irrigation head are obtained through a soil moisture dielectric sensor. An infiltration model was established with drip irrigation working pressure and pulse duration as independent variables and soil hydrodynamic parameters as dependent variables. The spatial overlap rate of the dielectric constant matrix of the root system spatial distribution and the three-dimensional volume data of the wetted body is calculated. When the overlap rate between the root density and the moisture content of the soil at a specific depth is lower than a preset threshold, the working pressure and pulse width of the drip irrigation pipe at the corresponding depth are calculated and adjusted according to the infiltration model, so that the movement trajectory of the wetting front coincides with the spatial distribution of high-density roots.

2. The water-saving citrus cultivation method based on green production according to claim 1, characterized in that, The steps for obtaining the root spatial distribution dielectric constant matrix include: establishing a three-dimensional rectangular coordinate system with the center of the base of the citrus tree trunk as the origin, and arranging the multi-channel time-domain reflectometry sensor array in the three-dimensional rectangular coordinate system according to the rule of increasing radial distance and increasing depth layer by layer; Within a set acquisition period, the dielectric constant measurement value of each sensor probe in the multi-channel time-domain reflectometry sensor array is recorded; The dielectric constant measurement value of each sensor probe is fused and mapped with the spatial coordinates in the three-dimensional rectangular coordinate system to generate the root spatial distribution dielectric constant matrix containing spatial coordinate information and dielectric constant information.

3. The water-saving citrus cultivation method based on green production according to claim 1, characterized in that, The steps for obtaining the three-dimensional volume data of the wetted body include: arranging the soil moisture dielectric sensor around the drip irrigation head according to the three-dimensional grid nodes, and collecting the soil volume moisture content of each grid node according to the set sampling frequency during irrigation; The soil volumetric water content of each grid node is calculated by three-dimensional spatial interpolation to generate a continuously distributed soil water content spatial field. The set of spatial points in the soil moisture content spatial field whose soil volumetric moisture content is equal to the sum of the initial moisture content before irrigation and the set moisture content increment is defined as the moist front, and the spatial region enclosed by the moist front constitutes the three-dimensional volume data of the moist body.

4. The water-saving citrus cultivation method based on green production according to claim 3, characterized in that, The steps for establishing the infiltration model include: obtaining the initial moisture content, saturated moisture content, and soil hydraulic diffusivity of the citrus root zone soil; The drip irrigation working pressure is converted into the drip irrigation head water flow rate, and the water flow rate and the pulse duration are substituted as input variables into the correction equation based on the Green-Ampt model; Using the initial moisture content, the saturated moisture content, and the soil hydraulic diffusivity as parameter variables in the modified equation, the modified equation is solved to obtain the functional relationship between the advancing distance of the wetting front and the cumulative infiltration amount, and the functional relationship is confirmed as the infiltration model.

5. The water-saving citrus cultivation method based on green production according to claim 1, characterized in that, The steps for calculating the spatial overlap rate include: dividing the three-dimensional space where the root spatial distribution dielectric constant matrix is ​​located and the three-dimensional space where the three-dimensional volume data of the wetted body is located into three-dimensional meshes of the same scale according to a preset voxel size; For each divided 3D mesh voxel, the root density feature represented by the dielectric constant and the wetted body feature represented by the volume water content are extracted. The proportion of grid voxels that simultaneously possess the root density characteristic and the moist body characteristic is determined to be the root density and moist body water overlap rate of the soil layer at the specific depth.

6. The water-saving citrus cultivation method based on green production according to claim 1, characterized in that, The steps for adjusting the working pressure and pulse width of the drip irrigation pipe at the corresponding depth include: extracting the difference between the root density and the overlap rate of the wetted body of the soil layer at the specific depth at the current moment and the preset threshold; The difference is input into the infiltration model as a target approximation variable, and the target outflow rate and target pulse duration corresponding to the target wetting front advance distance are obtained by inverse solution through the infiltration model; The adjustment increment of the working pressure of the drip irrigation pipe at the corresponding depth is calculated based on the target outflow rate, and the adjustment increment of the pulse width is calculated based on the target pulse duration. The adjustment increment of the working pressure and the adjustment increment of the pulse width are then output to the pressure regulating valve and the solenoid valve.

7. A water-saving citrus cultivation method based on green production according to claim 2, characterized in that, After generating the root spatial distribution dielectric constant matrix, the method further includes: obtaining the current phenological stage of the citrus and retrieving the root growth expansion prediction model corresponding to the phenological stage; The grid coordinates in the dielectric constant matrix of the root system spatial distribution are input into the root system growth and expansion prediction model, and the root extension direction vector and extension length prediction value corresponding to each grid coordinate in the phenological stage are output. The root spatial distribution dielectric constant matrix is ​​coordinately translated based on the root extension direction vector and the predicted extension length, and the translated dielectric constant measurement value is weighted and attenuated based on the radial distance from the origin, thereby generating the dynamically corrected root spatial distribution dielectric constant matrix.

8. A water-saving citrus cultivation method based on green production according to claim 3, characterized in that, The steps for constructing the three-dimensional volume data of the wet body further include: during the interval between two adjacent pulse durations, switching the sampling frequency of the soil moisture dielectric sensor to the interval sampling frequency; Based on the soil volumetric water content of each grid node collected at the intermittent sampling frequency, the three-dimensional spatial interpolation calculation is re-executed to generate the intermittent soil water content spatial field; In the intermittent period, spatial points are extracted from the soil moisture content spatial field where the rate of decrease in soil volume moisture content is equal to a preset critical value. These spatial points are connected to form a water redistribution boundary. The water redistribution boundary is then used to replace the wetting front to update the three-dimensional volume data of the wetting body.

9. A water-saving citrus cultivation method based on green production according to claim 4, characterized in that, After confirming the infiltration model, the method further includes: during the pause phase of the pulse duration, extracting the actual soil volumetric water content decrease curves of multiple spatial nodes at the wetting front; The actual soil volumetric moisture content decrease curve is input into the soil hydraulic diffusivity inversion algorithm to calculate the actual soil hydraulic diffusivity corresponding to the resting stage. The actual soil hydraulic diffusivity is used to replace the soil hydraulic diffusivity in the modified equation, and the parameter variables in the infiltration model are locally updated and calibrated. In subsequent irrigation cycles, the infiltration model, after local updates and calibration, is used for reverse solving.

10. A water-saving citrus cultivation method based on green production according to claim 6, characterized in that, The step of outputting the adjustment increment of the working pressure and the adjustment increment of the pulse width to the pressure regulating valve and the solenoid valve further includes: dividing the shallow root distribution area and the deep root distribution area, and obtaining the first overlap rate corresponding to the shallow root distribution area and the second overlap rate corresponding to the deep root distribution area respectively. When the first overlap rate meets the preset threshold and the second overlap rate is lower than the preset threshold, the adjustment increment of the working pressure of the shallow drip irrigation pipe is reduced and the adjustment increment of the pulse width is shortened, while the adjustment increment of the working pressure of the deep drip irrigation pipe is increased and the adjustment increment of the pulse width is lengthened. The shallow drip irrigation tube and the deep drip irrigation tube are controlled to be staggered in timing from the pulse duration.

Images