Smart model based plant root zone water retaining agent targeted application method and system

Abstract

The application belongs to the technical field of plant planting, and discloses a plant root zone water-retaining agent targeted application method and system based on an intelligent model, which comprises the following steps: acquiring plant parameter data, planting pit structure data and soil texture data; determining a root zone radius and a root zone effective depth, calculating a root zone cross-sectional area and a root zone volume, and obtaining a layer number and a volume of each layer; acquiring a water supply event record and a water content time sequence, obtaining a root zone equivalent infiltration volume, and acquiring a root zone effective water retention volume; combining the root zone effective water retention volume to calculate a gap volume of the root zone effective water supply, and giving a gap level mark; pre-building a water-retaining agent application prediction model, and outputting a water-retaining agent target amount and a hierarchical delivery proportion set; correcting the output result of the water-retaining agent application prediction model; and the application models and analyzes the root zone water retention and the gap through the model, outputs the water-retaining agent amount and the hierarchical proportion, and improves the pertinence and stability of the water-retaining agent application.

Description

Technical Field

[0001] This invention relates to the field of plant cultivation technology, and more specifically, to a method and system for targeted application of plant root zone water-retaining agents based on intelligent models. Background Technology

[0002] In vegetation restoration and afforestation projects in desert and arid regions, target plants generally face insufficient effective water supply in the early stages of planting. On the one hand, surface evaporation is strong, diurnal temperature range is large, and shallow water retention time is short; on the other hand, sandy soil has large pores and a high permeability coefficient, allowing precipitation or irrigation water to quickly infiltrate and bypass the soil layer available to the roots, making it difficult for rhizosphere water to form a stable buffer, resulting in low germination rate, slow seedling establishment, and large fluctuations in survival rate. Against this backdrop, water-retaining agents, as materials that improve soil water retention and slow-release capacity, are widely used in planting pits, near drip irrigation tapes, and in locally improved soil layers to enhance available water in the root zone and drought resistance stability.

[0003] Current methods for applying water-retaining agents rely heavily on experience, using fixed amounts per plant or hole, which makes it difficult to consider the differences in root width, soil thickness, and soil texture among different plants. Inappropriate application can lead to the water-retaining agent absorbing water and swelling, which may alter the soil pore structure, reduce aeration, or cause localized water retention, thereby leading to problems such as root hypoxia and inhibited root growth.

[0004] In view of this, the present invention proposes a method and system for targeted application of plant root zone water-retaining agents based on intelligent models to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a method for targeted application of plant root zone water-retaining agents based on an intelligent model, comprising:

[0006] Acquire plant parameter data, planting pit structure data, and soil texture data;

[0007] Based on plant parameter data and planting pit structure data, the root zone radius and effective root zone depth are determined, and the root zone cross-sectional area and root zone volume are calculated. A layered grid index is generated to obtain the number of layers and the volume of each layer in the root zone.

[0008] Obtain water supply event records and water content time series. Based on the water supply event records, obtain the equivalent infiltration volume of the root zone. Based on the water content time series, determine the infiltration rate characterization parameters and effective water holding characterization parameters. Calculate the evaporation loss and infiltration loss to obtain the effective water retention volume of the root zone.

[0009] The target retention volume is preset, and the gap volume of effective water supply in the root zone is calculated by combining the effective water retention volume in the root zone, and the gap level is assigned.

[0010] The model is constructed by inputting feature vectors and a pre-constructed water-retaining agent application prediction model, and outputting a set of target dosage and stratified application ratios for water-retaining agents.

[0011] The water absorption ratio, apparent density, and porosity of each layer of the water-retaining agent were obtained, and the upper limit of the absorbable water volume was used as the equivalent infiltration volume of the root zone to correct the output of the water-retaining agent application prediction model.

[0012] Furthermore, the method for obtaining the root region radius is as follows:

[0013] By selecting a batch of samples of the same seedling species and age, the maximum radial distance that can be reached in the root zone of each sample was measured by the root probe method.

[0014] For each sample, the ratio of the maximum radial reach of the root zone to the crown diameter is calculated, and the average of all ratios is used as the root zone radius mapping coefficient for that seedling at that seedling age.

[0015] Based on the crown diameter in the plant parameter data, the initial value of the root zone radius is calculated by combining the root zone radius mapping coefficient. The initial value of the root zone radius is then compared with the planting pit radius in the planting pit structure data. If the initial value of the root zone radius is greater than the planting pit radius, the planting pit radius is used to replace the initial value of the root zone radius. If the initial value of the root zone radius is not greater than the planting pit radius, the initial value of the root zone radius is used as the root zone radius.

[0016] Furthermore, the method for obtaining the number of layers and the volume of each layer is as follows:

[0017] The effective root zone depth of plant groups of the same species and seedling age is obtained as the total layer depth, and the minimum layer thickness constraint that is feasible for construction is obtained.

[0018] The total depth of each layer is divided by the minimum layer thickness constraint to obtain the number of layers. The calculated value is then rounded down to obtain the number of basic layers, and the number of basic layers is set as the number of layers S. When S is less than 2, S is limited to 2.

[0019] The starting and ending depths of the first S layers are generated sequentially from the surface downwards according to the minimum layer thickness constraint, and the layer thickness of the first S minus 1 layers is taken from the minimum layer thickness constraint.

[0020] When the total depth of the layers cannot be divided by the minimum layer thickness constraint, the remaining depth is incorporated into the last layer, so that the termination depth of the last layer is equal to the total depth of the layers. When S is 2 and the total depth of the layers is less than twice the minimum layer thickness constraint, the thickness of the two layers is set to 1 / 2 of the total depth of the layers.

[0021] Each layer outputs the starting depth, ending depth, layer center depth, and layer volume, where the layer center depth is the average of the starting depth and ending depth, and the layer volume of the Kth layer is the product of the root region cross-sectional area and the thickness of the Kth layer.

[0022] Furthermore, the method for obtaining the infiltration rate characterization parameters is as follows:

[0023] Within the preset observation window, the moisture content at each sampling time is traversed, the maximum value is taken as the peak moisture content, and the sampling time at which the first maximum value is reached is taken as the peak time.

[0024] After obtaining the peak times at two depths, the upper and lower layers of the root zone, the time difference between the peak arrival times is calculated. The average propulsion speed of the water content peak per unit time is obtained by dividing the vertical distance between the two depths by the time difference between the peak arrival times. The average propulsion speed is used as a parameter to characterize the infiltration rate.

[0025] The upper layer of the root zone refers to the first layer vertically downward from the surface, while the lower layer of the root zone refers to the last layer.

[0026] Furthermore, the method for obtaining the effective water-holding characterization parameters is as follows:

[0027] Taking the time series of water content in the upper layer of the root zone as the object, the sampling time is traversed within the observation window to determine whether the water content in the upper layer of the root zone is higher than the lower limit of available water content of the plant, and the time periods that are continuously higher than the lower limit of available water content of the plant are accumulated to obtain the effective water holding characterization parameters.

[0028] The wilting coefficient of the plant can be obtained by using the lower limit of the water content of the planting plot.

[0029] Furthermore, the method for obtaining the amount of infiltration loss is as follows:

[0030] The root crossing time is obtained by dividing the infiltration rate by the effective depth of the root zone;

[0031] A preset accounting time window is set, and the root crossing time is compared with the accounting time window length. If the root crossing time is less than the accounting time window length, the root crossing risk is marked as high; if the root crossing time is not less than the accounting time window length, the root crossing risk is marked as low.

[0032] When the risk of crossing the root zone is marked as high, the water content corresponding to the peak water content of the lower layer in the root zone is used as the starting water content. Within the calculation time window, the reduction of the water content of the lower layer relative to the starting water content is accumulated at each sampling time starting from the peak time of the lower layer. The accumulated reduction is converted into the equivalent infiltration loss volume of the root zone according to the volume of the corresponding lower layer in the root zone, and is used as the infiltration loss. The accumulation is only carried out for the continuous decline section, and the accumulation stops when the water content of the lower layer rises again after the decline.

[0033] When the risk of infiltration is marked as low, the infiltration loss is set to 0.

[0034] Furthermore, the method for pre-setting a target retention volume, calculating the gap volume of effective water supply in the root zone in conjunction with the effective water retention volume in the root zone, and assigning a gap level label includes:

[0035] The total loss volume is obtained by adding the evaporation loss to the equivalent infiltration loss volume in the root zone, and the effective water retention volume in the root zone is obtained by subtracting the total loss volume from the equivalent infiltration volume in the root zone.

[0036] The effective water retention volume in the root zone is subtracted from the preset target retention volume. When the difference is less than or equal to 0, the gap is taken as 0. When the difference is greater than 0, the gap is taken as the difference.

[0037] The gap volumes of the most recent N water supply events are statistically analyzed. For each water supply event, the gap volumes are sorted from smallest to largest. The P1 quantile statistical value is taken as the first volume threshold for low gaps and medium gaps, and the P2 quantile statistical value is taken as the second volume threshold for medium gaps and high gaps. The first volume threshold is less than the second volume threshold.

[0038] The gap volume of the water supply incident is compared with the first volume threshold and the second volume threshold. When the gap volume is less than the first volume threshold, the gap level is marked as 1; when the gap volume is not less than the first volume threshold and less than the second volume threshold, the gap level is marked as 2; when the gap volume is not less than the second volume threshold, the gap level is marked as 3.

[0039] Furthermore, methods for correcting the output of the water-retaining agent application prediction model include:

[0040] Based on the target dosage of water-retaining agent and the set of stratified application ratios output by the water-retaining agent application prediction model, the target dosage of water-retaining agent for each layer is calculated, and the volume of each layer is read.

[0041] The pore volume of each layer is calculated based on the volume of each layer and the porosity of each layer, and the allowable pore volume of each layer is determined by combining the preset upper limit threshold of the pore occupancy of each layer.

[0042] Calculate the equivalent expansion volume of the water-retaining agent in the Kth layer and compare it with the allowable pore volume of the Kth layer. When the equivalent expansion volume of the Kth layer exceeds the allowable pore volume of the Kth layer, perform a correction on the target amount of the water-retaining agent in the Kth layer so that the corrected equivalent expansion volume of the Kth layer does not exceed the allowable pore volume of the Kth layer; when it does not exceed the allowable pore volume, keep the target amount of the water-retaining agent in the Kth layer unchanged.

[0043] The corrected target usage for each layer is summed to obtain the corrected total usage, and the layer-by-layer delivery ratio is updated accordingly.

[0044] Furthermore, the method for calculating the equivalent expansion volume of the water-retaining agent in each layer is as follows:

[0045] The equivalent infiltration volume of the root zone is taken as the upper limit of the absorbable water volume, and the upper limit of the absorbable water volume of each layer is obtained by distributing it according to the proportion of pore volume of each layer.

[0046] The target dosage of the water-retaining agent in the Kth layer is taken as the dry mass, and the apparent density of the water-retaining agent is converted into the equivalent dry volume.

[0047] The estimated water absorption of the dry material is obtained by multiplying the dry material mass by the water absorption ratio. The smaller value between the estimated water absorption of the dry material converted into volume and the upper limit of the water absorption capacity of the Kth layer is taken as the water absorption of the dry material. The water absorption of the dry material is added to the equivalent volume of the dry material to obtain the equivalent expansion volume after water absorption.

[0048] The intelligent model-based targeted application system for plant root zone water-retaining agents includes the following steps:

[0049] Data acquisition module: Acquires plant parameter data, planting pit structure data, and soil texture data;

[0050] Root zone stratification module: Based on plant parameter data and planting pit structure data, determine the root zone radius and effective root zone depth, calculate the root zone cross-sectional area and root zone volume, and generate a stratified grid index to obtain the number of layers and the volume of each layer in the root zone;

[0051] Water supply analysis module: acquires water supply event records and water content time series, obtains the equivalent infiltration volume of the root zone based on the water supply event records, determines the infiltration rate characterization parameters and effective water holding characterization parameters based on the water content time series, and calculates the evaporation loss and infiltration loss to obtain the effective water retention volume of the root zone.

[0052] Gap marking module: presets the target retention volume, calculates the gap volume of effective water supply in the root zone in combination with the effective water retention volume in the root zone, and assigns gap level markings;

[0053] Model building module: Constructs the model input feature vector, inputs the pre-built water-retaining agent application prediction model, and outputs a set of target dosage and stratified application ratio of water-retaining agent;

[0054] Results Correction Module: Obtains the water absorption ratio, apparent density, and porosity of each layer of the water-retaining agent, and uses the equivalent infiltration volume of the root zone as the upper limit of the water absorption capacity to correct the output results of the water-retaining agent application prediction model.

[0055] The technical effects and advantages of the intelligent model-based targeted application method and system for plant root zone water-retaining agents proposed in this invention are as follows:

[0056] This invention integrates plant parameters, soil structure, and soil texture into a structured framework. First, it determines the root zone radius and effective root zone depth, generating a layered grid index and the volume of each layer. Then, it combines water supply event records and moisture content time series to extract infiltration rate and effective water holding capacity parameters. Finally, it calculates infiltration, evaporation, and infiltration losses, obtaining the effective water retention volume and deficit volume in the root zone and assigning deficit level labels. This achieves a quantitative closed-loop system centered on available water in the root zone. This eliminates reliance on empirical estimations for water-retaining agent application, providing a consistent and actionable application guideline across different seedling ages, soil layers, and water supply conditions. This reduces the risk of under-application or over-application due to root zone scale misjudgments, soil layer differences, and water supply fluctuations, improving the replicability and cross-site portability of engineering deployments.

[0057] This invention introduces a pre-constructed prediction model for water-retaining agent application, enabling application decisions to adapt to multi-factor coupled scenarios. Addressing engineering risks such as excessive pore occupancy, limited gas-phase channels, and upper-layer water retention caused by the water-retaining agent's water absorption and expansion, this invention corrects and constrains the model output based on upper limits for water absorption ratio, apparent density, porosity, and pore occupancy before application. Furthermore, it controls the water absorption volume with an upper limit on absorbable water capacity, ensuring that the stratified application scheme can still be implemented under conditions of material expansion and soil pore capacity. This improves water retention efficiency while also ensuring root zone aeration safety and application stability. Attached Figure Description

[0058] Figure 1 This is a flowchart of the targeted application method of plant root zone water-retaining agent based on intelligent model in Embodiment 1 of the present invention;

[0059] Figure 2 This is a flowchart of the process for obtaining infiltration loss in Embodiment 1 of the present invention;

[0060] Figure 3 This is a block diagram of the plant root zone water-retaining agent targeted application system based on an intelligent model in Embodiment 2 of the present invention. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] Example 1

[0063] See Figure 1 As shown, this embodiment provides a method for targeted application of plant root zone water-retaining agents based on an intelligent model, including:

[0064] Obtain plant parameter data, planting pit structure data, and soil texture data.

[0065] The plant parameter data includes species, seedling age, and crown diameter. The crown diameter is measured by averaging the crown projection lengths in two mutually perpendicular directions. The crown diameter of a batch of seedlings of the same species and age is measured by sampling and the average value is taken as the crown diameter of that batch.

[0066] Obtain planting pit structure data and soil texture data.

[0067] The planting pit structure data includes the planting pit diameter and the effective soil layer thickness. Preferably, the construction design parameters are first read as target values, and the maximum and minimum inner diameters of the pit opening are checked and measured after the planting pit is excavated and before backfilling, and the average value is taken as the planting pit diameter. The effective soil layer thickness of the planting pit is preferably measured by checking the small profile after excavation to the interface of obvious texture or dense layer. The interface criterion is a sudden increase in penetration resistance or a sudden change in soil color / structure.

[0068] The soil texture data includes the mass fraction of sand particles and the mass fraction of clay particles, preferably obtained based on a third-party soil testing report; when a testing report is unavailable, samples can be taken from the upper and middle layers of the root zone, and the mass fraction of each particle size can be calculated by sieving and sedimentation.

[0069] Obtain the root zone radius of the plant.

[0070] The method for obtaining the root region radius is as follows:

[0071] Obtain the root zone radius mapping coefficient, which is a dimensionless parameter used to characterize the mapping relationship between the horizontal outward range of the root system and the crown size in the early stage of planting.

[0072] The initial value of the root zone radius is obtained by multiplying the crown diameter by the root zone radius mapping coefficient. The initial value of the root zone radius is then compared with the planting pit radius. If the initial value of the root zone radius is greater than the planting pit radius, the planting pit radius is used instead of the initial value of the root zone radius. If the initial value of the root zone radius is not greater than the planting pit radius, the initial value of the root zone radius is used as the root zone radius.

[0073] The root zone radius mapping coefficient is obtained from sample statistics. Specifically, a batch of samples of the same seedling species and age is selected, and the maximum radial reach of the root zone is measured for each sample using a root probe method. The ratio of the maximum radial reach of the root zone to the crown diameter is calculated for each sample, and the average of all ratios is taken as the root zone radius mapping coefficient for that seedling species and age. The corresponding root zone radius mapping coefficient is obtained by changing the seedling species and age.

[0074] Obtain the effective depth of the root region.

[0075] The effective depth of the root zone is constrained by both the effective soil layer thickness of the planting pit and the current depth of the seedling root system, taking the smaller of the two. The current depth of the seedling root system is used to characterize the vertical reach of the seedling root system before planting.

[0076] The optimal root depth for seedlings is obtained through sampling inspection: representative seedlings of the same species and age are selected from each batch, and the root ball height or taproot length is measured. The median value is taken as the current root depth of the batch. When seedlings are supplied in container or soil ball form, the root ball height is preferably taken as the measurement benchmark.

[0077] Calculate the cross-sectional area and volume of the root region.

[0078] The root region cross-section is approximated as a circle, and the radius of the root region cross-section is the root region radius. The root region volume is calculated by approximating it as a cylinder, with the cylinder radius being the root region radius and the height being the effective depth of the root region. The cross-sectional area and volume of the root region are calculated based on the area formula of a circle and the volume formula of a cylinder.

[0079] The root region is divided into layers to generate a hierarchical grid index.

[0080] For plant groups of the same species and at the same seedling age, all plants in the plant group use the same number of layers and layer depth, instead of setting individual layers for each plant.

[0081] Specifically, the effective root zone depth of the plant group is first obtained as the total layer depth, and the minimum layer thickness constraint for construction is obtained. The total layer depth is divided by the minimum layer thickness constraint to obtain the calculated value of the number of layers. The calculated value is rounded down to obtain the number of basic layers, and the number of basic layers is set as the layer number S. The layer number S is limited to not less than 2. When the calculated S is less than 2, S is set to 2.

[0082] The starting and ending depths of the first S layers are generated sequentially from the surface downwards according to the minimum layer thickness constraint. The layer thickness of the first S minus 1 layers is taken from the minimum layer thickness constraint. When the total layer depth cannot be divided by the minimum layer thickness constraint, the remaining depth is incorporated into the last layer, so that the ending depth of the last layer is equal to the total layer depth. To avoid the layer thickness being too thin when the total layer depth is small, when S=2 and the total layer depth is less than twice the minimum layer thickness constraint, the layer thickness of the two layers is set to 1 / 2 of the total layer depth.

[0083] Each layer outputs a layer number, starting depth, ending depth, layer center depth, and layer volume, where the layer center depth is the average of the starting depth and ending depth, and the layer volume of the Kth layer is the product of the root region cross-sectional area and the thickness of the Kth layer.

[0084] The minimum layer thickness constraint is set based on the minimum layer thickness that can be stably achieved in the on-site construction mixing and backfilling process, and its repeatability is verified through multiple backfilling tests before curing.

[0085] Obtain water supply event records and water content time series.

[0086] The water supply event records are obtained from rain gauges and irrigation meters, including the start time of the event, the end time of the event, and the water supply volume.

[0087] The water supply volume can be recorded in terms of water depth or volume, and will be uniformly converted into the equivalent infiltration volume of the root zone during the gap calculation stage.

[0088] When the water supply event is a precipitation event, the water supply is usually given by the rain gauge in terms of water depth. The equivalent infiltration volume of the root zone is obtained by multiplying the water depth by the cross-sectional area of ​​the root zone.

[0089] When the water supply event is an irrigation event, if the water supply volume is given in volume, the water supply volume is divided by the number of plants to obtain the equivalent infiltration volume of the root zone; if the water supply volume is given in water depth, the equivalent infiltration volume of the root zone is obtained by multiplying the water depth by the cross-sectional area of ​​the root zone.

[0090] The moisture content time series is obtained by moisture content sensors buried in the upper and lower layers of the root zone according to a preset sampling period, including the sampling time and the corresponding moisture content, and the sampling times of sensors at different depths are aligned to the same time index.

[0091] The upper layer of the root region is defined as the depth range corresponding to the first layer in the hierarchical grid index, and the lower layer of the root region is defined as the depth range corresponding to the deepest layer in the hierarchical grid index; the burial depth of the moisture content sensor is taken as the center depth of the corresponding layer.

[0092] Using the start time of the water supply event as the time reference, a baseline value and observation window are determined: a preset observation window is established, with the water content before the start time of the water supply event used as the baseline value, and the observation window after the start time of the water supply event is extracted for peak identification. Based on time alignment, the peak time is identified and infiltration rate characterization parameters are calculated.

[0093] Peak identification is performed on the upper and lower depths of the root region respectively. The peak identification rule is: traverse the water content at each sampling time within the observation window, take the maximum value as the water content peak, and take the sampling time that first reaches the maximum value as the peak time.

[0094] After obtaining the peak times at two depths, the time difference between the peak arrival times is calculated, and the average propulsion velocity of the water content peak per unit time is obtained by dividing the vertical distance between the two depths by the time difference between the peak arrival times. The average propulsion velocity is used as a parameter characterizing the infiltration rate.

[0095] The observation window is a continuous time interval starting from the start time of a water supply event, used to identify the peak water content and its arrival time, and is obtained by observing the most recent water supply events. For example, if the peak arrival time of the most recent 30 water supply events is statistically distributed between 12 minutes and 68 minutes, the maximum value of 68 minutes is taken, adjusted upwards by 10%, and then rounded up as the upper limit of the observation window length.

[0096] See Figure 2 As shown, a root crossing risk marker is generated.

[0097] The root crossing time is obtained by dividing the effective depth of the root zone by the infiltration rate, a characterization parameter.

[0098] A preset accounting time window is set, and the root crossing time is compared with the accounting time window length. If the root crossing time is less than the accounting time window length, the root crossing risk is marked as high; if the root crossing time is not less than the accounting time window length, the root crossing risk is marked as low.

[0099] Subsequently, effective water-holding characterization parameters were obtained: taking the time series of water content in the upper layer of the root zone as the object, the sampling time was traversed within the observation window to determine whether the water content was higher than the lower limit of the plant's available water content, and the time periods that were continuously higher than the lower limit of the plant's available water content were accumulated to obtain the effective water-holding characterization parameters.

[0100] The lower limit of available water content for plants is determined using the wilting coefficient: the wilting coefficient given in the soil texture report or agronomic parameter table of the planting plot is read, and the wilting coefficient is preferably the value of the parameter table that matches the seedling or crop type; when the wilting coefficient is given in terms of volumetric water content, the wilting coefficient is directly used as the lower limit of available water content for plants; when the wilting coefficient is given in terms of mass water content, the dry bulk density of the same soil layer is further read, and the mass water content is converted into volumetric water content according to the dry bulk density and then used as the lower limit of available water content for plants.

[0101] Calculate evaporation loss and perform time window conversion, and output all data in a unified volumetric format. Evaporation loss is the converted value of water loss caused by surface evaporation within the calculation time window.

[0102] The method for obtaining evaporation loss is to set up an evaporation measurement device at the engineering site and output the cumulative evaporation within the calculation time window. The cumulative evaporation is generally given in terms of water depth and diameter, and the evaporation loss is calculated by multiplying the water depth by the cross-sectional area of ​​the root zone.

[0103] Calculate the infiltration loss and output the effective water supply gap in the root zone, and output all data in volumetric format.

[0104] Infiltration loss is used to characterize the amount of water that leaks across the lower boundary of the effective depth of the root zone after a water supply event, i.e., the amount of deep water lost that the target plant cannot utilize within the accounting time window.

[0105] When the risk of crossing the root zone is marked as high, the water content corresponding to the peak water content of the lower layer in the root zone is used as the starting water content. Within the calculation time window, the reduction of the water content of the lower layer relative to the starting water content is calculated cumulatively at each sampling time starting from the peak water content of the lower layer. The cumulative reduction is converted into the equivalent infiltration loss volume of the root zone according to the volume of the corresponding lower layer in the root zone, and is used as the infiltration loss. The accumulation is only performed on the continuous falling section, and the accumulation stops when the water content of the lower layer rises again after falling.

[0106] When the risk of leakage beyond the boundary is marked as low, it is determined that the contribution of leakage beyond the boundary to the accounting time window is limited, and the amount of leakage loss is taken as 0.

[0107] The effective water retention volume in the root zone is calculated based on evaporation loss and infiltration loss. The calculation rule is as follows: add the evaporation loss to the equivalent infiltration loss volume in the root zone to obtain the total loss volume; subtract the total loss volume from the equivalent infiltration volume in the root zone to obtain the effective water retention volume in the root zone. Furthermore, a target retention volume is introduced, which characterizes the minimum effective water storage requirement in the root zone to meet the water holding target threshold. The calculation rule for the effective water supply gap volume in the root zone is as follows: subtract the effective water retention volume in the root zone from the target retention volume. When the difference is less than or equal to 0, the gap is set to 0; when the difference is greater than 0, the difference is taken as the gap volume, and a gap level label is further output.

[0108] The method for setting the target retention volume includes: selecting no less than N water supply events within a plant group of the same seedling species and age; calculating the effective water holding characteristic parameters and the effective water retention volume in the root zone for each event; screening out events whose effective water holding characteristic parameters are not less than the target water holding threshold; sorting the retention volumes of the events according to size; and taking the median value as the target retention volume.

[0109] Methods for obtaining gap level markers include:

[0110] The gaps in the most recent N water supply events are statistically analyzed. For each water supply event, the gap volume is sorted from smallest to largest. The P1 quantile statistical value is taken as the first volume threshold for low gaps and medium gaps, and the P2 quantile statistical value is taken as the second volume threshold for medium gaps and high gaps. The first volume threshold is less than the second volume threshold.

[0111] The gap volume of the water supply incident is compared with the first volume threshold and the second volume threshold. When the gap volume is less than the first volume threshold, the gap level is marked as 1; when the gap volume is not less than the first volume threshold and less than the second volume threshold, the gap level is marked as 2; when the gap volume is not less than the second volume threshold, the gap level is marked as 3.

[0112] For example, N is set to 30. The gap water depths of the most recent 30 water supply events are statistically analyzed. For each water supply event, the gap water depth is calculated and sorted from smallest to largest. The 20th percentile statistical value is 5mm, and the 80th percentile statistical value is 15mm. When the cross-sectional area of ​​a root zone is 0.20 square meters, 5mm is multiplied by 0.20 square meters to obtain the first volume threshold of 0.0010 cubic meters, and 15mm is multiplied by 0.20 square meters to obtain the second volume threshold of 0.0030 cubic meters. If the calculated gap volume for this event is 0.0008 cubic meters, the gap level is marked as 1; if the gap volume is 0.0020 cubic meters, the gap level is marked as 2; and if the gap volume is 0.0040 cubic meters, the gap level is marked as 3.

[0113] The calculation time window is a continuous time interval starting from the start time of a water supply event. It is used to summarize the equivalent infiltration volume, evaporation loss, and equivalent downward infiltration loss volume of the root zone during the calculation period, and to calculate the gap volume. This time window is obtained through observations of the most recent water supply events. Specifically, the effective water holding time at the end of the most recent water supply events is statistically analyzed. The effective water holding time at the end of the effective water holding time is the time difference between the first drop in the upper layer water content of the root zone from the start of the water supply event to the lower limit of the plant's available water content. The larger quantile statistical value of the effective water holding time at the end of the effective water holding time is taken and superimposed with the residual time to obtain the upper limit of the calculation time window length. This upper limit of the calculation time window length is not less than the upper limit of the observation window length. For example, if the effective water holding time at the end of the most recent 30 water supply events is distributed between 4.2 hours and 15.6 hours, the maximum value of 15.6 hours is taken, adjusted upwards by 10%, and then rounded up to obtain the upper limit of the calculation time window length.

[0114] A pre-constructed prediction model for water-retaining agent application was used to obtain the target dosage and stratified application ratio of the water-retaining agent.

[0115] The training method for the water-retaining agent application prediction model includes:

[0116] Under actual planting conditions, L sets of training data are pre-collected, where L is a positive integer greater than 1. These L sets of training data form event-level records indexed by water supply events. Each event-level record corresponding to a water supply event includes a model input feature vector. This vector characterizes the soil texture data, root zone cross-sectional area, effective root zone depth, number of layers S, volume of each layer, prior evaporation feature field, and equivalent infiltration volume of the root zone for that water supply event, and is concatenated in a preset order. During training, the prior evaporation feature field can be the cumulative evaporation value within the calculation time window or the measured output of the evaporation measurement device; during deployment and inference, it is replaced by historical statistical values ​​from the same period. In addition to the model input feature vector, the event-level record also includes measured result fields for generating supervision labels, including effective water holding capacity parameters, gap volume, and gap level markers. These measured result fields are not used as model input features in training. Under actual planting conditions, multiple application schemes were collected for each water supply event to form supervised tagged data. These application schemes were obtained through a gradient application experiment: for plant groups of the same seedling species and age, Q representative plants were selected, where Q is a positive integer greater than 1; under the same water supply event conditions, M different application quality levels were set for the Q representative plants, and corresponding stratified application ratio templates were set, where M is a positive integer greater than 1. For each application scheme, after execution, the effective water-holding characteristic parameters of the scheme were recorded within the calculation time window, and the gap volume corresponding to the water supply event was calculated. The gap volumes were then categorized to obtain gap level labels. The criteria for compliance were that the gap level label was not higher than the target gap level and the effective water-holding characteristic parameters were not less than the target water-holding threshold. The scheme with the lowest application quality was selected from the compliant schemes, and its application quality was used as the optimal application quality label for the water-retaining agent in that water supply event; the proportion of the actual application quality of each layer in this scheme to the total application quality was used as the optimal stratified ratio label. The optimal application quality label was a discrete optimal solution obtained based on discrete application quality levels, which met the compliance constraints and had engineering feasibility. When no compliant solution exists under the same water supply event conditions, the solution with the smallest gap level label is selected as a fallback candidate. When gap level labels are tied, the solution with the largest effective water holding capacity parameter is selected. If they are still tied, the solution with the smallest delivery quality is selected, and this event is marked as a non-compliant candidate event for separate statistics in subsequent training and evaluation. Furthermore, the above gradient delivery experiment is repeated for plant groups under different water supply events, so that the training data simultaneously covers individual differences among different plants and differences in different water supply conditions, and that each optimal label can be traced back to the corresponding event-level experimental record containing the model input fields.

[0117] For example, M is set to 4, and the feeding quality levels for a single plant are set to 40g, 60g, 80g, and 100g, with stratification ratio templates of [0.50, 0.30, 0.20], [0.40, 0.35, 0.25], [0.35, 0.35, 0.30], and [0.30, 0.35, 0.35], respectively. When the target water holding threshold is 6.0h and the target gap level is 1, if 60g corresponds to an effective water holding of 5.5h, the water holding criterion is not met; if 80g corresponds to an effective water holding of 6.4h and the gap level is marked as 1, the compliance criterion is met; if 100g corresponds to an effective water holding of 7.2h and the gap level is marked as 1, the compliance criterion is also met. Then, according to the principle of "minimum feeding quality in the compliance plan", the optimal feeding quality label is 80g, and the optimal stratification ratio label is [0.35, 0.35, 0.30].

[0118] The method for setting the water-holding target threshold includes: selecting no less than 10 typical water supply events within a plant group of the same seedling species and age, calculating the effective water-holding characterization parameters of each event according to the accounting time window and taking the average value, and then adjusting the average value by 20% as the water-holding target threshold and fixing it.

[0119] The water-retaining agent application prediction model employs a gradient boosting regression tree model and a two-stage modeling approach to establish the hierarchical connection between dosage prediction and proportion prediction: The first stage is a dosage prediction sub-model, with input consisting of event-level input records composed of model input fields, and output being the target dosage of water-retaining agent; the second stage is a proportion prediction sub-model, with output being a set of stratified application proportions consistent with the number of strata, S. The input of the second stage consists of event-level input records and the dosage field. During training the proportion prediction sub-model, the dosage field is preferentially replaced with the optimal application quality label to stabilize training; during inference deployment, the target dosage of water-retaining agent predicted in the first stage is used as the dosage field input. The output of the proportion prediction sub-model is the initial value of the proportion weights for each layer. These initial weights are truncated due to non-negativity and then normalized to obtain a set of stratified application proportions, ensuring that all elements are greater than or equal to 0 and the sum of all elements is 1.

[0120] Set the initial hyperparameters. The initial hyperparameters for the usage prediction sub-model include: 200 trees, a learning rate of 0.05, a maximum depth of 6, a minimum number of samples per leaf node of 20, a feature sampling ratio of 0.80, and a regularization coefficient of 0.10. The initial hyperparameters for the ratio prediction sub-model include: 300 trees, a learning rate of 0.05, a maximum depth of 6, a minimum number of samples per leaf node of 20, a feature sampling ratio of 0.80, and a regularization coefficient of 0.10. The collected training data is divided into training, validation, and test sets according to a preset ratio of 6:3:1, and further grouped by land parcel to avoid data from the same parcel being included in both the training and test sets simultaneously.

[0121] The model was trained using the training set, and hyperparameters were tuned. For the dosage prediction sub-model, mean squared error was used as the loss function during training; for the proportion prediction sub-model, the mean squared error between the hierarchical proportion label and the predicted proportion was used as the loss function. Hyperparameter tuning employed Bayesian optimization, with optimization ranges including: 100 to 600 trees, learning rate of 0.01 to 0.10, maximum depth of 3 to 10, minimum number of samples per leaf node of 10 to 60, regularization coefficient of 0.05 to 0.30, and feature sampling ratio of 0.60 to 1.00. An early stopping mechanism was introduced, using the validation set loss as a monitoring metric during gradient boosting iterations. Training was stopped when the validation set loss decreased by less than 0.1% over 50 consecutive tree-building iterations, preserving the model parameters with the best validation set performance.

[0122] The trained model is evaluated using a test set to determine deployment conditions. A dosage error index is calculated for the dosage prediction sub-model, and a proportion error index is calculated for the proportion prediction sub-model. Simultaneously, event-level consistency checks are performed: for test set events, the deviation between the target dosage of water-retaining agent output by the model and the optimal application quality label is compared, and the deviation between the stratified application proportion output by the model and the optimal stratified proportion label is also compared. If both types of deviations do not exceed a preset allowable range, the event is considered consistent. Preferably, the allowable range for dosage deviation uses a relative deviation threshold, preferably 15%; the allowable range for stratified proportion deviation preferably uses a parallel threshold of the maximum deviation threshold for each layer and the overall L1 distance threshold, preferably 0.10 for the maximum deviation threshold for each layer and 0.20 for the overall L1 distance threshold.

[0123] Further, the under-deployment rate is calculated, which is the percentage of events where the model output usage is lower than the optimal delivery quality tag usage. A preset threshold for the under-deployment rate is used as a safety constraint indicator, preferably 10%. When the percentage of consistent events in the test set is not less than a preset consistency ratio threshold and the under-deployment rate meets the safety constraint, the model performance evaluation is deemed satisfactory, and deployment is initiated. The consistency ratio threshold is preferably 80%. For "non-compliant candidate events," their consistency and under-deployment rate are calculated separately to indicate that the delivery strategy under extreme conditions requires manual review or a more conservative configuration with a higher safety margin.

[0124] The results of the prediction model for water-retaining agent application are corrected to reduce the excessive occupation of pore space by the water-retaining agent's water absorption and expansion.

[0125] Based on the target dosage of water-retaining agent and the set of stratified application ratios output by the water-retaining agent application prediction model, the target dosage of water-retaining agent for each layer is calculated, and the volume of each layer is read.

[0126] The water absorption ratio and apparent density of the water-retaining agent were obtained, the soil porosity of each layer was obtained, and the equivalent infiltration volume of the root zone during this water supply event was read as the upper limit of the absorbable water.

[0127] Preferably, the water absorption ratio and apparent density of the water-retaining agent are obtained from the manufacturer's test report; the porosity of each soil layer is obtained based on a third-party soil test report.

[0128] The pore volume of each layer is calculated based on the volume of each layer and the porosity of each layer. The allowable pore volume of each layer is determined by combining the preset upper limit threshold of the pore occupancy rate of each layer. The upper limit threshold of the pore occupancy rate of each layer is preferably 0.25 to 0.35 for the upper layer and 0.35 to 0.45 for the middle and lower layers to ensure that sufficient gas phase channels are still retained after pore occupancy.

[0129] The upper limit of water absorption capacity is allocated according to the proportion of pore volume in each layer to obtain the upper limit of water absorption capacity for each layer, so as to avoid the upper limit of water absorption capacity being reused in each layer.

[0130] The equivalent expansion volume of the water-retaining agent in each layer is calculated. The specific calculation method is as follows: first, take the target dosage of the water-retaining agent in the Kth layer as the dry mass of the Kth layer, and then convert it into the equivalent dry volume based on the apparent density of the water-retaining agent.

[0131] The estimated water absorption of the dry material is obtained by multiplying the dry material mass by the water absorption ratio. The smaller value between the estimated water absorption of the dry material converted into volume and the upper limit of the water absorption capacity of the Kth layer is taken as the water absorption of the dry material. The water absorption of the dry material is added to the equivalent volume of the dry material to obtain the equivalent expansion volume after water absorption.

[0132] The equivalent expansion volume of layer K is compared with the allowable pore volume of layer K. When the equivalent expansion volume of layer K exceeds the allowable pore volume of layer K, the target dosage of water-retaining agent in layer K is corrected to ensure that the corrected equivalent expansion volume of layer K does not exceed the allowable pore volume of layer K; if it does not exceed the allowable pore volume, the target dosage of water-retaining agent in layer K remains unchanged. The corrected target dosages of each layer are summed to obtain the corrected total dosage, and the layer application ratio is updated accordingly to ensure that the expansion and pore-occupancy risk of each layer meets the constraints before implementation.

[0133] In this embodiment, the layered application scheme output by the model is pre-corrected based on the material's water absorption characteristics and soil pore capacity constraints. This ensures that the application quality of each layer meets the allowable pore volume constraint before implementation, thereby reducing the excessive occupation of pore space by the water-retaining agent's water absorption and expansion. This reduces risks such as limited gas phase channels, waterlogging in the upper layer, and root hypoxia, and improves the feasibility and safety redundancy of the layered application scheme. It should be noted that the gating correction is based on pre-estimated constraints of water absorption ratio, apparent density, and upper limit of absorbable water. Its purpose is to provide conservative risk assessment and cutoff rules to guide dosage adjustments before application, rather than accurately reproducing the actual water migration and moisture content evolution process after application.

[0134] Example 2

[0135] See Figure 3 As shown, this embodiment provides a plant root zone water-retaining agent targeted application system based on an intelligent model. The implementation of the plant root zone water-retaining agent targeted application method based on the intelligent model includes:

[0136] Data acquisition module: Acquires plant parameter data, planting pit structure data, and soil texture data.

[0137] Root zone stratification module: Based on plant parameter data and planting pit structure data, determine the root zone radius and effective root zone depth, calculate the root zone cross-sectional area and root zone volume, and generate a stratified grid index to obtain the number of layers and the volume of each layer in the root zone.

[0138] Water supply analysis module: acquires water supply event records and water content time series, obtains the equivalent infiltration volume of the root zone based on the water supply event records, determines the infiltration rate characterization parameters and effective water holding characterization parameters based on the water content time series, and calculates the evaporation loss and infiltration loss to obtain the effective water retention volume of the root zone.

[0139] Gap marking module: It presets the target retention volume and calculates the gap volume of effective water supply in the root zone in combination with the effective water retention volume in the root zone, and assigns gap level marking.

[0140] Model building module: Constructs the model input feature vector, inputs the pre-built water-retaining agent application prediction model, and outputs a set of target dosage and stratified application ratio of water-retaining agent.

[0141] Results Correction Module: Obtains the water absorption ratio, apparent density, and porosity of each layer of the water-retaining agent, and uses the equivalent infiltration volume of the root zone as the upper limit of the water absorption capacity to correct the output results of the water-retaining agent application prediction model.

[0142] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0143] In conclusion, the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for targeted application of plant root zone water-retaining agents based on intelligent models, characterized in that, include: Acquire plant parameter data, planting pit structure data, and soil texture data; Based on plant parameter data and planting pit structure data, the root zone radius and effective root zone depth are determined, and the root zone cross-sectional area and root zone volume are calculated. A layered grid index is generated to obtain the number of layers and the volume of each layer in the root zone. The method for obtaining the number of layers and the volume of each layer is as follows: The effective root zone depth of plant groups of the same species and seedling age is obtained as the total layer depth, and the minimum layer thickness constraint that is feasible for construction is obtained. The total depth of each layer is divided by the minimum layer thickness constraint to obtain the number of layers. The calculated value is then rounded down to obtain the number of basic layers, and the number of basic layers is set as the number of layers S. When S is less than 2, S is limited to 2. The starting and ending depths of the first S layers are generated sequentially from the surface downwards according to the minimum layer thickness constraint, and the layer thickness of the first S minus 1 layers is taken from the minimum layer thickness constraint. When the total depth of the layers cannot be divided by the minimum layer thickness constraint, the remaining depth is incorporated into the last layer, so that the termination depth of the last layer is equal to the total depth of the layers. When S is 2 and the total depth of the layers is less than twice the minimum layer thickness constraint, the thickness of the two layers is set to 1 / 2 of the total depth of the layers. Each layer outputs the starting depth, ending depth, layer center depth, and layer volume, where the layer center depth is the average of the starting depth and ending depth, and the layer volume of the Kth layer is the product of the root region cross-sectional area and the thickness of the Kth layer. Obtain water supply event records and water content time series. Based on the water supply event records, obtain the equivalent infiltration volume of the root zone. Based on the water content time series, determine the infiltration rate characterization parameters and effective water holding characterization parameters. Calculate the evaporation loss and infiltration loss to obtain the effective water retention volume of the root zone. The method for obtaining the infiltration rate characterization parameters is as follows: Within the preset observation window, the moisture content at each sampling time is traversed, the maximum value is taken as the peak moisture content, and the sampling time at which the first maximum value is reached is taken as the peak time. After obtaining the peak times at two depths, the upper and lower layers of the root zone, the time difference between the peak arrival times is calculated. The average propulsion speed of the water content peak per unit time is obtained by dividing the vertical distance between the two depths by the time difference between the peak arrival times. The average propulsion speed is used as a parameter to characterize the infiltration rate. The upper layer of the root zone refers to the first layer that descends vertically from the surface of the earth, while the lower layer of the root zone refers to the last layer. The method for obtaining the effective water-holding characterization parameters is as follows: Taking the time series of water content in the upper layer of the root zone as the object, the sampling time is traversed within the observation window to determine whether the water content in the upper layer of the root zone is higher than the lower limit of available water content of the plant, and the time periods that are continuously higher than the lower limit of available water content of the plant are accumulated to obtain the effective water holding characterization parameters. The wilting coefficient of the plant can be obtained by taking the lower limit of the water content of the planting plot; The target retention volume is preset, and the gap volume of effective water supply in the root zone is calculated by combining the effective water retention volume in the root zone, and the gap level is assigned. The model is constructed by inputting feature vectors and a pre-constructed water-retaining agent application prediction model, and outputting a set of target dosage and stratified application ratios for water-retaining agents. The water absorption ratio, apparent density, and porosity of each layer of the water-retaining agent were obtained, and the upper limit of the absorbable water volume was used as the equivalent infiltration volume of the root zone to correct the output of the water-retaining agent application prediction model.

2. The method for targeted application of plant root zone water-retaining agent based on intelligent model according to claim 1, characterized in that, The method for obtaining the root region radius is as follows: By selecting a batch of samples of the same seedling species and age, the maximum radial distance that can be reached in the root zone of each sample was measured by the root probe method. For each sample, the ratio of the maximum radial reach of the root zone to the crown diameter is calculated, and the average of all ratios is used as the root zone radius mapping coefficient for that seedling at that seedling age. Based on the crown diameter in the plant parameter data, the initial value of the root zone radius is calculated by combining the root zone radius mapping coefficient. The initial value of the root zone radius is then compared with the planting pit radius in the planting pit structure data. If the initial value of the root zone radius is greater than the planting pit radius, the planting pit radius is used to replace the initial value of the root zone radius. If the initial value of the root zone radius is not greater than the planting pit radius, the initial value of the root zone radius is used as the root zone radius.

3. The method for targeted application of plant root zone water-retaining agent based on intelligent model according to claim 1, characterized in that, The method for obtaining the infiltration loss is as follows: The root crossing time is obtained by dividing the infiltration rate by the effective depth of the root zone; A preset accounting time window is set, and the root crossing time is compared with the accounting time window length. If the root crossing time is less than the accounting time window length, the root crossing risk is marked as high; if the root crossing time is not less than the accounting time window length, the root crossing risk is marked as low. When the risk of crossing the root zone is marked as high, the water content corresponding to the peak water content of the lower layer in the root zone is used as the starting water content. Within the calculation time window, the reduction of the water content of the lower layer relative to the starting water content is accumulated at each sampling time starting from the peak time of the lower layer. The accumulated reduction is converted into the equivalent infiltration loss volume of the root zone according to the volume of the corresponding lower layer in the root zone, and is used as the infiltration loss. The accumulation is only carried out for the continuous decline section, and the accumulation stops when the water content of the lower layer rises again after the decline. When the risk of infiltration is marked as low, the infiltration loss is set to 0.

4. The method for targeted application of plant root zone water-retaining agent based on intelligent model according to claim 1, characterized in that, The method for calculating the gap volume of effective water supply in the root zone by combining the preset target retention volume with the effective water retention volume in the root zone, and assigning a gap level label includes: The total loss volume is obtained by adding the evaporation loss to the equivalent infiltration loss volume in the root zone, and the effective water retention volume in the root zone is obtained by subtracting the total loss volume from the equivalent infiltration volume in the root zone. The effective water retention volume in the root zone is subtracted from the preset target retention volume. When the difference is less than or equal to 0, the gap is taken as 0. When the difference is greater than 0, the gap is taken as the difference. The gap volumes of the most recent N water supply events are statistically analyzed. For each water supply event, the gap volumes are sorted from smallest to largest. The P1 quantile statistical value is taken as the first volume threshold for low gaps and medium gaps, and the P2 quantile statistical value is taken as the second volume threshold for medium gaps and high gaps. The first volume threshold is less than the second volume threshold. The gap volume of the water supply incident is compared with the first volume threshold and the second volume threshold. When the gap volume is less than the first volume threshold, the gap level is marked as 1; when the gap volume is not less than the first volume threshold and less than the second volume threshold, the gap level is marked as 2; when the gap volume is not less than the second volume threshold, the gap level is marked as 3.

5. The method for targeted application of plant root zone water-retaining agent based on intelligent model according to claim 1, characterized in that, The method for correcting the output of the water-retaining agent application prediction model includes: Based on the target dosage of water-retaining agent and the set of stratified application ratios output by the water-retaining agent application prediction model, the target dosage of water-retaining agent for each layer is calculated, and the volume of each layer is read. The pore volume of each layer is calculated based on the volume of each layer and the porosity of each layer, and the allowable pore volume of each layer is determined by combining the preset upper limit threshold of the pore occupancy of each layer. Calculate the equivalent expansion volume of the water-retaining agent in the Kth layer and compare it with the allowable pore volume of the Kth layer. When the equivalent expansion volume of the Kth layer exceeds the allowable pore volume of the Kth layer, perform a correction on the target amount of the water-retaining agent in the Kth layer so that the corrected equivalent expansion volume of the Kth layer does not exceed the allowable pore volume of the Kth layer; when it does not exceed the allowable pore volume, keep the target amount of the water-retaining agent in the Kth layer unchanged. The corrected target usage for each layer is summed to obtain the corrected total usage, and the layer-by-layer delivery ratio is updated accordingly.

6. The method for targeted application of plant root zone water-retaining agent based on intelligent model according to claim 5, characterized in that, The method for calculating the equivalent expansion volume of the water-retaining agent in each layer is as follows: The equivalent infiltration volume of the root zone is taken as the upper limit of the absorbable water volume, and the upper limit of the absorbable water volume of each layer is obtained by distributing it according to the proportion of pore volume of each layer. The target dosage of the water-retaining agent in the Kth layer is taken as the dry mass, and the apparent density of the water-retaining agent is converted into the equivalent dry volume. The estimated water absorption of the dry material is obtained by multiplying the dry material mass by the water absorption ratio. The smaller value between the estimated water absorption of the dry material converted into volume and the upper limit of the water absorption capacity of the Kth layer is taken as the water absorption of the dry material. The water absorption of the dry material is added to the equivalent volume of the dry material to obtain the equivalent expansion volume after water absorption.

7. A plant root zone water-retaining agent targeted application system based on an intelligent model, implementing the plant root zone water-retaining agent targeted application method based on an intelligent model as described in any one of claims 1-6, characterized in that, include: Data acquisition module: Acquires plant parameter data, planting pit structure data, and soil texture data; Root zone stratification module: Based on plant parameter data and planting pit structure data, determine the root zone radius and effective root zone depth, calculate the root zone cross-sectional area and root zone volume, and generate a stratified grid index to obtain the number of layers and the volume of each layer in the root zone; Water supply analysis module: acquires water supply event records and water content time series, obtains the equivalent infiltration volume of the root zone based on the water supply event records, determines the infiltration rate characterization parameters and effective water holding characterization parameters based on the water content time series, and calculates the evaporation loss and infiltration loss to obtain the effective water retention volume of the root zone. Gap marking module: presets the target retention volume, calculates the gap volume of effective water supply in the root zone in combination with the effective water retention volume in the root zone, and assigns gap level markings; Model building module: Constructs the model input feature vector, inputs the pre-built water-retaining agent application prediction model, and outputs a set of target dosage and stratified application ratio of water-retaining agent; Results Correction Module: Obtains the water absorption ratio, apparent density, and porosity of each layer of the water-retaining agent, and uses the equivalent infiltration volume of the root zone as the upper limit of the water absorption capacity to correct the output results of the water-retaining agent application prediction model.

Images