Soil evaporation test system

Abstract

The invention relates to the technical field of soil evaporation test systems, in particular to a soil evaporation test system. Comprising the steps that a soil column simulation device contains a simulation soil column; the temperature control evaporation device regulates and controls the current evaporation temperature; the precipitation device controls the current precipitation amount; the salt concentration collecting device collects the current salt concentration; the gas permeability sensor collects current gas diffusion rates of the simulated soil column at different depths; the salt concentration supply device provides a salt solution; the hierarchical data acquisition device acquires corresponding temperatures, pH values, conductivity and nitrogen migration amounts at different depths; and the electronic equipment takes the current evaporation temperature, the current precipitation, the current salt concentration and each current gas diffusion rate as input, and takes the temperature, the pH value, the conductivity and the nitrogen migration amount corresponding to each depth as output, the initial soil column data prediction network is trained, and a target soil column data prediction model is obtained. The accurate simulation of the process of water-salt migration along with nitrogen migration of the saline-alkali soil is realized.

Description

Technical Field

[0001] This invention relates to the field of soil evaporation testing system technology, and specifically to a soil evaporation testing system. Background Technology

[0002] Saline-alkali soils, as an important global reserve of arable land, directly determine soil fertility levels and crop suitability through water and salt transport processes, making them a core research direction for saline-alkali land improvement and efficient utilization. Nitrogen, a crucial nutrient essential for crop growth, exhibits a close coupling relationship between its migration and transformation in saline-alkali soils and water and salt transport. Soil moisture evaporation drives salt accumulation towards the surface, while changes in salt concentration affect the solubility, adsorption-desorption balance, and microbial transformation efficiency of nitrogen (especially nitrates), thereby altering the spatial distribution and availability of nitrogen. However, current research on saline-alkali soils, both domestically and internationally, still has significant shortcomings. Specifically, the dynamic laws governing nitrogen migration driven by water and salt are insufficiently revealed, and a systematic theory of "water-salt-nitrogen" synergistic regulation has not yet been formed. This situation severely restricts the breakthrough and implementation of technologies for efficient nutrient utilization in saline-alkali soils.

[0003] The functional limitations of existing saline-alkali soil simulation experimental devices further exacerbate the aforementioned research bottlenecks. Traditional devices are designed for water and salt transport studies and do not include dedicated components for nitrogen sampling and analysis. They can only indirectly estimate nitrogen content through subsequent full-scale soil analysis, and cannot accurately capture the amount of nitrogen transported with water evaporation, let alone distinguish the dynamic changes in forms such as nitrates.

[0004] Therefore, how to accurately capture the amount of nitrogen transported with water evaporation, thereby improving the accuracy and continuity of nitrogen migration simulation experiments in saline-alkali soils, has become an urgent problem to be solved. Summary of the Invention

[0005] In view of this, the present invention provides a soil evaporation test system to solve the problem of accurately capturing the amount of nitrogen transported with water evaporation, thereby improving the accuracy and continuity of nitrogen migration simulation experiments in saline-alkali soil.

[0006] In a first aspect, the present invention provides a soil evaporation testing system, comprising: a soil column simulation device, a temperature-controlled evaporation device, a stratified data acquisition device, a precipitation device, a salt concentration acquisition device, a salt concentration supply device, an air permeability sensor, and electronic equipment; wherein, the temperature-controlled evaporation device is installed above the soil column simulation device, the precipitation device is installed above the soil column simulation device, the stratified data acquisition device and the air permeability sensor are both installed inside the simulated soil column within the soil column simulation device, the salt concentration acquisition device is installed below the soil column simulation device, and the salt concentration supply device is installed on one side of the soil column simulation device; the temperature-controlled evaporation device, the stratified data acquisition device, the precipitation device, and the salt concentration acquisition device are all communicatively connected to the electronic equipment; wherein: Soil column simulation device, used to hold simulated soil columns; A temperature-controlled evaporation device is used to regulate the current evaporation temperature of the simulated soil column surface under the control of electronic equipment; A precipitation device is used to control the current precipitation corresponding to a simulated soil column under the control of electronic equipment; A salt concentration acquisition device is used to acquire the current salt concentration below a simulated soil column; A permeability sensor is used to collect the current gas diffusion rate at different depths of a simulated soil column; A salt concentration supply device is used to provide salt solution to the simulated soil column; The layered data acquisition device is used to simulate a soil column under the conditions of current evaporation temperature, current precipitation, current salt concentration, and current gas diffusion rate at each depth. After a preset time, it collects the measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport at different depths. An electronic device is used to input the current evaporation temperature, current precipitation, current salt concentration, and current gas diffusion rate at each depth as input parameters, and the measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport at each depth as output parameters. The input parameters are fed into the initial soil column data prediction network to train the initial soil column data prediction network and obtain the target soil column data prediction model.

[0007] In one optional implementation, the electronic device is used to: input the current evaporation temperature, current precipitation, current salt concentration, and the current gas diffusion rate, measured soil column temperature, measured soil column pH value, measured soil column conductivity, and measured soil column nitrogen transport at each depth into the input layer of the initial soil column data prediction network; the embedding layer in the initial soil column data prediction network inputs the current evaporation temperature, current precipitation, current salt concentration, and the current gas diffusion rate, measured soil column temperature, measured soil column pH value, measured soil column conductivity, and measured soil column nitrogen transport at each depth. Feature extraction is performed on the transport volume to obtain target fusion features; the target fusion features are input into the output layer of the initial soil column data prediction network; the output layer outputs the virtual soil column temperature, virtual soil column pH value, and virtual soil column nitrogen transport volume corresponding to each depth; based on the first relationship error between the measured soil column temperature and the virtual soil column temperature, the second relationship error between the virtual soil column pH value and the measured soil column pH value, and the third relationship error between the virtual soil column nitrogen transport volume and the measured soil column nitrogen transport volume, the initial soil column data prediction network is trained to obtain the target soil column data prediction model.

[0008] In one optional implementation, the input layer includes a first sub-input layer and a second sub-input layer. The electronic device is used to: input the current evaporation temperature, current precipitation, and current salt concentration as shared features into the first sub-input layer; and input the current gas diffusion rate, measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport amount corresponding to each depth into the second sub-input layer according to a preset acquisition gradient.

[0009] In one optional implementation, the electronic device is configured to: Encode the current evaporation temperature, current precipitation, current salt concentration, and the current gas diffusion rate, measured soil column temperature, measured soil column pH value, measured soil column conductivity, and measured soil column nitrogen transport amount at each depth, according to the data type corresponding to each input data; Encode the current evaporation temperature, the measured soil column temperature at each depth, and the measured soil column nitrogen transport amount as features, generating evaporation-temperature-nitrogen coupled features; Encode the current salt concentration, the measured soil column conductivity at each depth, and the measured soil column nitrogen transport amount as features, generating evaporation-temperature-nitrogen coupled features; Encode the current salt concentration, the measured soil column conductivity at each depth, and the measured soil column nitrogen transport amount as features, generating evaporation-temperature-nitrogen coupled features. Nitrogen transport in the column is feature-extracted to generate a salt concentration-pH-nitrogen speciation coupling feature based on conductivity. The embedding layer extracts features from the current precipitation and the measured soil column conductivity at each depth to obtain precipitation-conductivity coupling features. Based on a preset duration, the evaporation-temperature-nitrogen coupling features, salt concentration-pH-nitrogen speciation coupling features, and precipitation-conductivity coupling features are determined, and their corresponding target weights are assigned. Based on the target weights, the evaporation-temperature-nitrogen coupling features, salt concentration-pH-nitrogen speciation coupling features, and precipitation-conductivity coupling features are fused to generate target fused features.

[0010] In one optional embodiment, the electronic device is configured to: acquire the measured electrical conductivity of the soil column corresponding to each depth; detect whether the measured electrical conductivity of the soil column is less than or equal to a preset electrical conductivity threshold; if the measured electrical conductivity of the soil column is less than or equal to the preset electrical conductivity threshold, calculate the initial salt concentration-pH-nitrogen speciation coupling characteristic based on a first relationship between the current salt concentration, the measured soil column pH value, and the measured nitrogen transport amount in the soil column; if the measured soil column electrical conductivity is greater than the preset electrical conductivity threshold, calculate the initial salt concentration-pH-nitrogen speciation coupling characteristic based on a second relationship between the current salt concentration, the measured soil column pH value, and the measured nitrogen transport amount in the soil column; and correct the initial salt concentration-pH-nitrogen speciation coupling characteristic according to the depth corresponding to the simulated soil column to obtain the salt concentration-pH-nitrogen speciation coupling characteristic.

[0011] In one optional implementation, the electronic device is configured to: determine initial weights corresponding to the evaporation-temperature-nitrogen coupling characteristics, salt concentration-pH-nitrogen speciation coupling characteristics, and precipitation-conductivity coupling characteristics, respectively, based on a preset duration; adjust each initial weight according to the depth of the simulated soil column to obtain standby weights corresponding to the evaporation-temperature-nitrogen coupling characteristics, salt concentration-pH-nitrogen speciation coupling characteristics, and precipitation-conductivity coupling characteristics at each depth; and adjust each standby weight according to the current evaporation temperature and the current precipitation to obtain each target weight.

[0012] In one alternative implementation, the first relational error is: ; Among them, T 实测，i Let T be the measured temperature of the soil column corresponding to the depth of the i-th layer. 虚拟，i K represents the virtual soil column temperature corresponding to depth i. t This is the time decay factor; The error in the second relation is: ; Among them, pH 实测，i Here is the measured pH value of the soil column corresponding to the depth of the i-th layer, pH 虚拟，i Let EG be the pH value of the virtual soil column corresponding to the depth of the i-th layer. i Let be the measured electrical conductivity of the soil column at the depth of the i-th layer; The error in the third relation is:

[0013] Where, N 实测，i N represents the measured nitrogen transport in the soil column corresponding to the depth of the i-th layer. 虚拟，i Let D be the nitrogen transport amount of the virtual soil column corresponding to the depth of the i-th layer. i Direction coefficient; Based on the first relational error, the second relational error, and the third relational error, the total loss error is generated. Based on the total loss error, the initial soil column data prediction network is trained to obtain the target soil column data prediction model.

[0014] In one optional embodiment, the soil column simulation device includes a soil column container and a base; the base is installed below the soil column container and has a water inlet, which can supply salt solution to the bottom of the simulated soil column by connecting a salt concentration supply device to simulate salt transport.

[0015] In one optional embodiment, the stratified data acquisition device includes a temperature sensor group, a conductivity sensor group, a pH sensor group, and a non-destructive sampling device, wherein: the temperature sensor group is used to monitor the measured temperature of the simulated soil column at different depths in real time after a preset duration; the temperature sensor group consists of multiple electronic thermometers, each electronic thermometer employing a high-precision sensor; each temperature probe is inserted into the simulated soil column at different depths through temperature probe mounting holes on the wall of the soil column container in the soil column simulation device, with at least one temperature probe at each depth; the conductivity sensor group is used to monitor the measured conductivity of the simulated soil column at different depths in real time after a preset duration; the conductivity sensor group consists of multiple conductivity sensors, each conductivity sensor being inserted into the simulated soil column at different depths through conductivity sensor mounting holes on the wall of the soil column container in the soil column simulation device, with at least one conductivity sensor at each depth; the pH sensor group is used to monitor the measured pH value of the simulated soil column at different depths in real time after a preset duration; the pH sensor group consists of multiple pH sensors. The system consists of pH sensors, each inserted into a simulated soil column at different depths through a conductivity sensor mounting hole on the wall of the soil column container within the soil column simulation device. At least one pH sensor is installed at each depth. A non-destructive sampling device is used to monitor the measured nitrogen transport in the simulated soil column at different depths after a preset time. The non-destructive sampling device includes a pore tube, a vacuum tube, and a nitrogen transport measurement device for the measured soil column. The pore tube is made of inert material, its outer diameter matches the sampling hole on the soil column container wall, and its length is determined based on the soil... The height of the column container is determined. One end of the pore tube is closed, and the other end is connected to a vacuum tube. The wall of the pore tube is distributed with micropores of the first size in diameter, allowing only pore water to pass through and preventing soil particles from entering. In use, the pore tube is inserted into different depths in the simulated soil column through the sampling hole. Pore water enters the pore tube under negative pressure and flows into the vacuum tube, achieving non-destructive sampling. The measured nitrogen transport in the soil column is used to detect the pore water at each depth of the simulated soil column, determine the amount of nitrogen transport in the pore water, and obtain the measured nitrogen transport in the soil column.

[0016] In one optional embodiment, the temperature-controlled evaporation device includes an infrared lamp, a lampshade, a temperature controller, and a temperature sensor; the infrared lamp is installed above the soil column simulation device, and the lampshade focuses infrared radiation so that the radiant heat is uniformly applied to the surface of the simulated soil column; the temperature sensor is installed on the surface of the simulated soil column for real-time monitoring of the surface temperature of the simulated soil column; the temperature controller is used to control the infrared lamp according to the temperature detected by the temperature sensor.

[0017] The soil evaporation test system provided in this application embodiment uses a soil column simulation device to hold a simulated soil column. Combined with a temperature-controlled evaporation device (regulating the current evaporation temperature), a precipitation device (controlling the current precipitation), and a salt concentration supply device (providing salt solution), it can accurately reproduce soil environments under different climatic conditions (such as evaporation temperatures of 25℃, 40℃, and 55℃) and degrees of salinization. This multi-parameter synergistic control method overcomes the limitations of traditional devices with single evaporation control and coarse simulation of saline-alkali environments, providing a more realistic experimental platform for studying the coupling relationship between water-salt transport and nitrogen migration. After a preset time, the stratified data acquisition device simultaneously collects measured soil column temperature, pH value, electrical conductivity, and nitrogen transport at different depths. Combined with the gas diffusion rate collected by the permeability sensor, a complete dataset of "environmental factors - soil physicochemical properties - nitrogen migration" is formed. The electronic device takes the current evaporation temperature, current precipitation, current salt concentration, and current gas diffusion rate at each depth as input, and the measured soil column temperature, pH value, electrical conductivity, and nitrogen transport at each depth as output parameters to train an initial soil column data prediction network to obtain the target model. This model can quantify the correlation between environmental factors and soil response (such as the driving effect of increased evaporation temperature on nitrogen surface accumulation), providing a quantitative tool for revealing the nitrogen migration patterns in saline-alkali soils and assisting in practical applications such as optimizing water-saving irrigation technologies and evaluating the effectiveness of soil amendment materials. The soil evaporation test system, through a closed-loop process of "simulation-acquisition-modeling," achieves accurate simulation and parameter determination of the process of water and salt transport accompanied by nitrogen migration in saline-alkali soils. Furthermore, data modeling enhances the analytical capability for complex processes, providing reliable experimental support for research on the nitrogen cycle mechanism in saline-alkali lands, thereby improving the accuracy and continuity of nitrogen migration simulation experiments in saline-alkali soils. Attached Figure Description

[0018] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the structure of a soil evaporation test system according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the process of "obtaining a target soil column data prediction model" according to an embodiment of the present invention. Figure 3This is a schematic diagram of "determining the initial weights corresponding to the evaporation-temperature-nitrogen coupling characteristics, salt concentration-pH-nitrogen speciation coupling characteristics, and precipitation-conductivity coupling characteristics according to a preset duration" according to an embodiment of the present invention. Figure 4 This is a schematic diagram of another soil evaporation test system according to an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0021] According to embodiments of the present invention, a soil evaporation testing system is provided, such as... Figure 1 As shown, the soil evaporation test system includes: a soil column simulation device 1, a temperature-controlled evaporation device 2, a stratified data acquisition device 3, a precipitation device 4, a salt concentration acquisition device 5, a salt concentration supply device 6, an air permeability sensor 7, and electronic equipment 8; wherein, the temperature-controlled evaporation device 2 is installed above the soil column simulation device 1, the precipitation device 4 is installed above the soil column simulation device 1, the stratified data acquisition device 3 and the air permeability sensor 7 are both installed inside the simulated soil column in the soil column simulation device 1, the salt concentration acquisition device 5 is installed below the soil column simulation device 1, and the salt concentration supply device 6 is installed on one side of the soil column simulation device 1; the temperature-controlled evaporation device 2, the stratified data acquisition device 3, the precipitation device 4, and the salt concentration acquisition device 5 are all communicatively connected to the electronic equipment 8; wherein: Soil column simulation device 1, used to hold simulated soil columns; Temperature-controlled evaporation device 2 is used to regulate the current evaporation temperature of the simulated soil column surface under the control of electronic device 8; The precipitation device 4 is used to control the current precipitation corresponding to the simulated soil column under the control of the electronic device 8; Salt concentration acquisition device 5 is used to acquire the current salt concentration below the simulated soil column; The air permeability sensor 7 is used to collect the current gas diffusion rate at different depths of the simulated soil column; Salt concentration supply device 6 is used to provide salt solution to the simulated soil column; The layered data acquisition device 3 is used to simulate the soil column under the conditions of current evaporation temperature, current precipitation, current salt concentration and current gas diffusion rate at each depth. After a preset time, it collects the measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity and measured soil column nitrogen transport at different depths. Electronic device 8 is used to input the current evaporation temperature, current precipitation, current salt concentration, and current gas diffusion rate corresponding to each depth as input parameters, and the measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport amount corresponding to each depth as output parameters to the initial soil column data prediction network, train the initial soil column data prediction network, and obtain the target soil column data prediction model.

[0022] Specifically, the soil column simulation device 1 serves as the foundation of the soil evaporation test system, and it uses a transparent acrylic container (20cm inner diameter and 50cm height) to fill a simulated soil column.

[0023] The temperature-controlled evaporation device 2 is controlled by electronic equipment 8 to control the temperature (25℃, 40℃, 55℃), precisely regulate the evaporation intensity on the surface of the soil column, and match the radiative heat-driven evaporation process under different climatic conditions.

[0024] The precipitation device 4 is installed above the soil column and simulates different amounts of precipitation under the control of electronic equipment 8. The amount and frequency of spraying are controlled by the program (such as simulating light rain, moderate rain, and heavy rain) to study the effect of precipitation on nitrogen leaching and re-evaporation in the soil surface. The spraying affects the distribution of water and salt leaching from the soil surface to the deep layers, and interacts with the evaporation process.

[0025] Salt concentration supply device 6 supplies salt solution (wherein the salt solution is nitrite solution) to the bottom of the soil column. By controlling the concentration and total amount, different degrees of salinization (such as electrical conductivity 0.5%-2.0%) are simulated, providing a variable basis for studying the effect of salt on nitrogen migration.

[0026] The layered data acquisition device 3, by inserting sensors at different depths, collects the measured soil column temperature (accuracy ±0.1℃), measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport amount after a preset time.

[0027] Permeability sensors 7 are embedded at different depths in the soil column to monitor changes in soil pore structure in real time by monitoring gas diffusion rates. For example, when soil cracks due to evaporation, permeability increases significantly, and the system can automatically correlate this with nitrogen transport rate to analyze the impact of structural damage on nutrient migration.

[0028] Salt concentration acquisition device 5 monitors the base salt concentration below the simulated soil column, providing data support for analyzing the impact of environmental factors on nitrogen transformation.

[0029] After the experiment was started, the soil evaporation test system recorded the current evaporation temperature, current precipitation, current salt concentration, and current gas diffusion rate at each depth corresponding to the simulated soil column. Then, after a preset time period, the system collected the measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport at different depths of the simulated soil column through the stratified data acquisition device 3 at regular intervals (e.g., several hours to several days).

[0030] Electronic device 8 takes environmental driving parameters (current evaporation temperature, precipitation, salt concentration, and gas diffusion rate) as input and measured parameters from stratified monitoring (measured soil column temperature, measured soil column pH, measured soil column conductivity, and measured nitrogen transport in the soil column) as output, and inputs them into the initial soil column data prediction network. The network learns the correlation between environmental parameters and soil response (such as how an increase in evaporation temperature drives nitrogen migration to the surface) and obtains the target model through iterative optimization.

[0031] Ultimately, the trained model can predict soil conditions at different depths based on input environmental parameters, providing a reliable tool for revealing the patterns of water and salt transport accompanied by nitrogen migration in saline-alkali soils, optimizing water-saving irrigation technologies, and evaluating the effectiveness of soil amendment materials.

[0032] The soil evaporation test system provided in this application embodiment uses a soil column simulation device 1 to hold a simulated soil column, combined with a temperature-controlled evaporation device 2 (to regulate the current evaporation temperature), a precipitation device 4 (to control the current precipitation), and a salt concentration supply device 6 (to provide salt solution). This system can accurately reproduce soil environments under different climatic conditions (such as evaporation temperatures of 25℃, 40℃, and 55℃) and degrees of salinization. This multi-parameter synergistic control method overcomes the limitations of traditional devices with single evaporation control and coarse simulation of saline-alkali environments, providing a more realistic experimental platform for studying the coupling relationship between water-salt transport and nitrogen migration. After a preset time, the stratified data acquisition device 3 simultaneously collects measured soil column temperatures, pH values, electrical conductivity, and nitrogen transport amounts at different depths. Combined with the gas diffusion rate collected by the permeability sensor 7, this forms a complete dataset of "environmental factors - soil physicochemical properties - nitrogen migration." Electronic device 8 takes the current evaporation temperature, current precipitation, current salt concentration, and current gas diffusion rate at each depth as input, and the measured soil column temperature, pH value, electrical conductivity, and nitrogen transport at each depth as output parameters to train an initial soil column data prediction network to obtain the target model. This model can quantify the correlation between environmental factors and soil response (such as the driving effect of increased evaporation temperature on nitrogen surface accumulation), providing a quantitative tool for revealing the nitrogen migration law in saline-alkali soils, and assisting in practical applications such as optimizing water-saving irrigation technology and evaluating the effectiveness of soil amendment materials. The soil evaporation test system, through a closed-loop process of "simulation-acquisition-modeling," achieves accurate simulation and parameter determination of the process of water and salt transport accompanied by nitrogen migration in saline-alkali soils. Furthermore, data modeling enhances the analytical capability for complex processes, providing reliable experimental support for the study of nitrogen cycling mechanisms in saline-alkali lands, thereby improving the accuracy and continuity of nitrogen migration simulation experiments in saline-alkali soils.

[0033] In one optional embodiment of this application, "the current evaporation temperature, current precipitation, current salt concentration, and current gas diffusion rate corresponding to each depth are used as input parameters, and the measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport amount corresponding to each depth are used as output parameters. These parameters are input into an initial soil column data prediction network, and the initial soil column data prediction network is trained to obtain a target soil column data prediction model." Figure 2 As shown, it may include the following steps: Step S101: Input the current evaporation temperature, current precipitation, current salt concentration, current gas diffusion rate at each depth, measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport amount into the input layer of the initial soil column data prediction network.

[0034] Specifically, step S101 above may include the following steps: Step S1011: Input the current evaporation temperature, current precipitation, and current salt concentration as shared features into the first sub-input layer.

[0035] Specifically, the electronic device 8 generates shared features based on the current evaporation temperature, current precipitation, and current salt concentration, and then inputs the shared features into the first sub-input layer.

[0036] For example, shared features can be as follows: .

[0037] Step S1012: Input the current gas diffusion rate, measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport amount corresponding to each depth into the second sub-input layer according to the preset acquisition gradient.

[0038] Specifically, the electronic device 8 can input the current gas diffusion rate, measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport amount corresponding to each depth to the second sub-input layer according to the preset acquisition gradient.

[0039] In step S102, the embedding layer in the initial soil column data prediction network extracts features from the current evaporation temperature, current precipitation, current salt concentration, current gas diffusion rate at each depth, measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport, thereby obtaining the target fusion features.

[0040] Specifically, step S102 above may include the following steps: Step S1021: The embedding layer encodes the current evaporation temperature, current precipitation, current salt concentration, current gas diffusion rate, measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport amount according to the data type corresponding to each input data.

[0041] Specifically, the embedding layer encodes the current evaporation temperature, current precipitation, and current salt concentration according to the data type corresponding to each input data.

[0042] For example, for the current evaporation temperature, assuming the radiation characteristics of the 250W infrared lamp 21 of the "infrared temperature control evaporation system", the "temperature-radiation intensity mapping code" is used: the target temperature (25℃, 40℃, 55℃) is converted into a normalized value of 0.5, 0.8, 1.0 according to the power ratio of the infrared lamp 21 (e.g., 55℃ corresponds to 100% power), and at the same time, a "radiation influence radius label" (5cm, 15cm, 25cm) is attached, and the label value decreases with depth (1.0→0.7→0.4).

[0043] For the current precipitation, if the current precipitation is <10mm (not reaching effective leaching): the code is 0.3, with an additional "surface wet label" (affecting only 5cm depth); if the current precipitation is 10mm≤current precipitation≤30mm (effective leaching): the code is 0.7, with an additional "middle layer infiltration label" (affecting 5-25cm depth); if the current precipitation is >30mm (excessive leaching): the code is 1.0, with an additional "deep migration label" (affecting the entire depth).

[0044] Based on the current salt concentration, the current salt concentration (0.5%-2.0%) is divided into three levels: "low", "medium", and "high" (1, 2, 3). The typical conductivity range under this level is simultaneously associated (e.g., high salt concentration → conductivity 1.5%-2.0%), and a "diffusion rate coefficient" is added (the coefficient is low at high salt concentrations because soil pores are easily blocked), which strengthens the association with the characteristic of inhibiting gas exchange in saline-alkali environments.

[0045] To address the current gas diffusion rate, a "permeability-depth matrix coding" approach is adopted: the gas diffusion rates at each depth are arranged in a matrix in the order of 5cm→45cm. The surface layer (5cm, 15cm) is assigned an "evaporation influence weight" (the weight is increased by 20% when the diffusion rate is high, because evaporation requires oxygen), and the deep layer (35cm, 45cm) is assigned a "salt and alkali inhibition weight" (the weight is increased by 20% when the diffusion rate is low, because high salt content clogs the pores).

[0046] For the measured soil column temperature, a "temperature gradient coding" method is adopted: not only is the temperature value at each depth recorded, but the temperature difference between adjacent depths (such as the temperature difference between 5cm and 15cm) is also calculated. If the temperature difference is greater than 5℃ (the upper limit of reasonable gradient), an "abnormal gradient label" (with a value of -0.1) is added to the temperature value at that depth to constrain the physical rationality of the temperature distribution.

[0047] For the measured soil column pH value and measured soil column electrical conductivity, a "co-coding" method is adopted: the measured soil column pH value and measured soil column electrical conductivity at each depth are multiplied (pH×EC). If the result is >15 (high salinity and alkalinity area), the coding value is amplified by 1.2 times and an "nitrification inhibition tag" (value 1, representing inhibition of nitrifying bacteria activity) is added.

[0048] For the measured nitrogen transport in the soil column, a "migration direction coding" method is adopted: while recording the nitrogen value at each depth, the "surface / deep ratio" (5cm / 45cm) is calculated. If the ratio is >2 (surface accumulation), an "evaporation-dominated label" is added; if the ratio is <0.5 (deep migration), a "leaching-dominated label" is added, which conforms to the rule that the direction of nitrogen transport is dominated by environmental conditions.

[0049] In step S1022, the embedded layer extracts features from the current evaporation temperature, the measured soil column temperature at each depth, and the measured nitrogen transport in the soil column to generate evaporation-temperature-nitrogen coupled features.

[0050] Specifically, electronic device 8 can extract the current evaporation temperature, the measured soil column temperature corresponding to each depth, and the measured nitrogen transport in the soil column. Then, based on the extracted current evaporation temperature, the measured soil column temperature corresponding to each depth, and the measured nitrogen transport in the soil column, electronic device 8 can calculate the evaporation-temperature-nitrogen coupling characteristics.

[0051] For example, electronic device 8 can calculate the evaporation-temperature-nitrogen coupling characteristics based on the following formula.

[0052] Evaporation-temperature-nitrogen coupling characteristic = T 温度 ×T 实测,i ×N 实测,i ×K i ×(1+0.01×T d ).

[0053] Among them, T 温度 T is the current evaporation temperature. 实测,i N represents the measured temperature of the soil column corresponding to the depth of the i-th layer. 实测,i K represents the measured nitrogen transport volume of the soil column corresponding to the depth of the i-th layer. i This refers to the radiation attenuation coefficient. For example, the radiation attenuation coefficient at a depth of 5cm is 1.0 (directly exposed to radiation); at a depth of 15cm, it is 0.7 (radiation is weakened after conduction through the surface); and at a depth of 45cm, it is 0.3 (the radiation effect is negligible). T d The thermal conduction time lag coefficient increases with depth (5cm=10min, 15cm=30min, 45cm=90min), reflecting the characteristic that the temperature response of deep soil lags behind that of the surface layer.

[0054] Step S1023: The embedded layer extracts features from the current salt concentration, the measured electrical conductivity of the soil column at each depth, the measured pH value of the soil column, and the measured nitrogen transport amount of the soil column, and generates a salt concentration-pH-nitrogen speciation coupling feature based on electrical conductivity.

[0055] Specifically, step S1023 above may include the following steps: Step a1: Obtain the measured electrical conductivity of the soil column at each depth, and check whether the measured electrical conductivity of the soil column is less than or equal to the preset electrical conductivity threshold.

[0056] Specifically, the electronic device 8 can calculate the preset conductivity threshold based on the following formula.

[0057] ; Wherein, the base value "2" is the benchmark threshold (unit: mS / cm) for a low-salt environment (0.5% salt concentration); the coefficient "3" reflects the weight of the influence of salt concentration on conductivity; when the current salt concentration reaches 2.0%, the threshold linearly increases to 5 mS / cm; C 盐 δ(d) represents the current salt concentration; δ(d) is the depth correction term (+0.5 for depths of 35cm and 45cm, and -0.3 for surface depths of 5cm and 15cm), simulating the pattern that "deep layers are continuously supplied with salt solution from base 12, resulting in more significant salt accumulation".

[0058] Then, the electronic device 8 compares the measured electrical conductivity of the soil column at each depth with the preset electrical conductivity threshold to detect whether the measured electrical conductivity of the soil column is less than or equal to the preset electrical conductivity threshold.

[0059] Step a2: If the measured electrical conductivity of the soil column is less than or equal to the preset electrical conductivity threshold, the initial salt concentration-pH-nitrogen speciation coupling characteristics are calculated based on the first relationship between the current salt concentration, the measured pH value of the soil column, and the measured nitrogen transport amount of the soil column.

[0060] Step a3: If the measured electrical conductivity of the soil column is greater than the preset electrical conductivity threshold, the initial salt concentration-pH-nitrogen speciation coupling characteristics are calculated based on the second relationship between the current salt concentration, the measured pH value of the soil column, and the measured nitrogen transport amount of the soil column.

[0061] Specifically, the electronic device 8 can determine the relationship between the measured electrical conductivity of the soil column at each depth and the preset electrical conductivity threshold.

[0062] If the measured electrical conductivity of the soil column is less than or equal to the preset electrical conductivity threshold, the initial salt concentration-pH-nitrogen speciation coupling characteristics are calculated based on the first relationship between the current salt concentration, the measured pH value of the soil column, and the measured nitrogen transport amount of the soil column.

[0063] For example, electronic device 8 can calculate the initial salt concentration-pH-nitrogen speciation coupling characteristics based on the following formula: ; Among them, S i For the initial salt concentration-pH-nitrogen speciation coupling characteristics, C 盐 For the current salt concentration, EC 实测,i Let pH be the measured electrical conductivity of the soil column corresponding to the depth of the i-th layer. 实测,i S represents the measured pH value of the soil column corresponding to the depth of the i-th layer. t As a correction factor, for example, soda salting S t =1.3, Chloride salting S t =1.0.

[0064] Step a4: Based on the depth corresponding to the simulated soil column, the initial salt concentration-pH-nitrogen speciation coupling characteristics are corrected to obtain the salt concentration-pH-nitrogen speciation coupling characteristics.

[0065] Specifically, the electronic device 8 can determine the depth stability coefficient K based on the depth corresponding to the simulated soil column. d For example, deep layers (35cm, 45cm): K d =1.2, because the deep saline-alkali environment is stable and has a more lasting inhibitory effect on nitrogen, so the characteristic value needs to be amplified; Middle layer (25cm): K d =1.0, because the medium properties are medium, no additional correction is needed; Surface (5cm, 15cm): K d =0.8, because the surface saline-alkali environment fluctuates greatly, and the instantaneous value has a weak long-term effect on nitrogen, so the characteristic value needs to be reduced.

[0066] Then, the electronic device 8 multiplies the initial salt concentration-pH-nitrogen speciation coupling features corresponding to each depth by the corresponding depth stability coefficient to obtain the salt concentration-pH-nitrogen speciation coupling features.

[0067] In step S1024, the embedded layer extracts features from the current precipitation and the measured electrical conductivity of the soil column at each depth to obtain precipitation-electrical conductivity coupling features.

[0068] Specifically, the embedded layer can obtain the current precipitation, as well as the measured electrical conductivity of the soil column at each depth and the current gas diffusion rate.

[0069] Electronic device 8 can calculate the conductivity difference between adjacent depths in the i-th group. If the conductivity difference between adjacent depths in the i-th group is greater than 0, and the current precipitation is greater than a preset precipitation threshold, then the shower direction is determined to be positive, and the label is "positive leaching" (surface salt migrates to deeper layers), with a positive feature value. If the conductivity difference between adjacent depths in the i-th group is greater than 0, and the current precipitation is less than a preset precipitation threshold, then the label is "false gradient" (only surface is wet, no leaching), and the feature value is multiplied by 0.5 (to reduce weight). If the conductivity difference between adjacent depths in the i-th group is less than 0, then the label is "reverse accumulation" (evaporation causes surface salt to be higher than deeper layers), and the feature value is negative.

[0070] Then, the embedding layer can calculate the precipitation-conductivity coupling characteristics based on the following formula.

[0071]

[0072] Among them, P i For precipitation-conductivity coupling characteristics, P 当前 This represents the current precipitation (EC). 上层,i -EC 下层,iD represents the conductivity difference between adjacent depths in the i-th group (e.g., 5cm and 15cm). 扩散,i Let L be the current gas diffusion rate corresponding to the depth of the i-th layer. i P is the leaching direction label. c Pc is a pore connectivity index, inversely derived from the gas diffusion rate (higher diffusion rate results in higher Pc). It is used to reinforce the characteristic that "soil pore structure is a channel for precipitation infiltration and salt migration." For example, higher diffusion rate → higher Pc. c =1.2, low diffusion rate → P c =0.8.

[0073] Step S1025: Based on the preset duration, determine the target weights corresponding to the evaporation-temperature-nitrogen coupling characteristics, salt concentration-pH-nitrogen speciation coupling characteristics, and precipitation-conductivity coupling characteristics.

[0074] Specifically, step S1025 above may include the following steps: Step b1: Based on the preset duration, determine the initial weights corresponding to the evaporation-temperature-nitrogen coupling characteristics, salt concentration-pH-nitrogen speciation coupling characteristics, and precipitation-conductivity coupling characteristics.

[0075] Specifically, the electronic device 8 can determine the initial weights corresponding to the evaporation-temperature-nitrogen coupling feature, the salt concentration-pH-nitrogen speciation coupling feature, and the precipitation-conductivity coupling feature based on the preset duration and the corresponding initial weights.

[0076] For example, in the short term (e.g., within 6 hours): precipitation (if any) has a significant leaching effect on surface salts, and evaporation has not yet formed a stable temperature gradient; in the medium term (e.g., 24-48 hours): the cumulative effect of infrared radiation (evaporation temperature) becomes prominent, and the upward migration of salts and nitrogen with water evaporation intensifies; in the long term (e.g., more than 72 hours): soil moisture tends to be in equilibrium, and the interaction between salt concentration and pH value becomes the dominant factor affecting nitrogen migration.

[0077] like Figure 3The diagram illustrates the initial weights of the evaporation-temperature-nitrogen coupling characteristics, salt concentration-pH-nitrogen speciation coupling characteristics, and precipitation-conductivity coupling characteristics, determined according to preset durations. Specifically, the short-term weighting is based on the fact that precipitation from the "sprinkler system" primarily affects the top 5-15cm layer within 12 hours, at which point the "precipitation-conductivity coupling" has the strongest effect on nitrogen leaching (weight 70%), while the cumulative effect of infrared evaporation has not yet manifested (weight 20%). The medium-term weighting is based on the fact that after 12 hours of operation of the "infrared temperature-controlled evaporation system," the surface temperature gradient of the soil column stabilizes (the temperature difference between 5cm and 45cm can reach over 10℃), and evaporation-driven nitrogen surface accumulation becomes the core process ("evaporation-temperature-nitrogen coupling" weight 50%). The long-term weighting is based on the fact that after 72 hours, the soil moisture evaporation rate slows down, the interaction between salt concentration and pH value tends to stabilize (e.g., pH value remains consistently high in high-salt environments), and the inhibitory effect on nitrogen migration becomes dominant (weight 60%).

[0078] Step b2: Adjust the initial weights according to the depth of the simulated soil column to obtain the backup weights for the evaporation-temperature-nitrogen coupling characteristics, salt concentration-pH-nitrogen speciation coupling characteristics, and precipitation-conductivity coupling characteristics at each depth.

[0079] Specifically, the electronic device 8 adjusts the initial weights according to the depth of the simulated soil column to obtain the backup weights corresponding to the evaporation-temperature-nitrogen coupling characteristics, salt concentration-pH-nitrogen speciation coupling characteristics, and precipitation-conductivity coupling characteristics at each depth.

[0080] For example, for the surface layer (5cm, 15cm): the weight of the evaporation-temperature-nitrogen coupling characteristic is increased by 20% because it is directly affected by the infrared temperature-controlled evaporation system, resulting in the strongest evaporation-driven effect; the weight of the precipitation-conductivity coupling characteristic is increased by 10% because it is close to the "spray device," where precipitation leaching has a significant impact; the weight of the salt concentration-pH-nitrogen form coupling characteristic is decreased by 30%—due to its distance from the base 12 salt solution supply, the saline-alkali environment has a weaker impact. For the middle layer (25cm), the weight remains at the initial value because this depth is affected by the combined effects of evaporation, precipitation, and salinity, with no obvious dominant factor. For the deep layer (35cm, 45cm): the weight of the salt concentration-pH-nitrogen form coupling characteristic is increased by 30% because the "base 12 water inlet" is most affected by the salt solution system supply; the weight of the evaporation-temperature-nitrogen coupling characteristic is decreased by 20% because infrared radiation attenuates after being conducted through multiple layers, reducing its impact; the weight of the precipitation-conductivity coupling characteristic is decreased by 10% because the amount of precipitation penetrating into the deep layer is small, resulting in weak leaching.

[0081] Step b3: Adjust the standby weights according to the current evaporation temperature and current precipitation to obtain the target weights.

[0082] Specifically, the electronic device 8 can adjust the backup weights according to the current evaporation temperature and the current precipitation to obtain the target weights.

[0083] For example, the current evaporation temperature adjustment rule is based on the fact that "the infrared lamp 21 can be set to three temperature levels: 25℃, 40℃, and 55℃", and is adjusted according to the temperature range: when the current evaporation temperature is ≥55℃ (strong radiation): the weight of the evaporation-temperature-nitrogen coupling characteristic of the surface layer (5cm, 15cm) is increased by 10% (e.g., 5cm from 50% to 55%), to strengthen the effect of "high temperature accelerating evaporation"; when the current evaporation temperature is ≤25℃ (weak radiation): the weight of the evaporation-temperature-nitrogen coupling characteristic of the surface layer is decreased by 10%, and the weight of the salt concentration-pH-nitrogen form coupling characteristic is increased by 5%, to adapt to the rule of "low temperature and saline-alkali environment dominating".

[0084] The current precipitation adjustment rules are based on the "threshold effect of precipitation on leaching" (<10mm is invalid precipitation, ≥10mm is valid precipitation): When the current precipitation is ≥10mm (valid precipitation), the weight of the precipitation-conductivity coupling feature at the surface (5cm, 15cm) is increased by 10% (e.g., 5cm from 35% to 40%), highlighting the effect of "precipitation significantly changing salinity distribution"; when the current precipitation is 0 (no precipitation), the weight of the precipitation-conductivity coupling feature at all depths is decreased by 50% (e.g., 5cm from 35% to 17.5%), weakening the interference when there is no precipitation and focusing on the effects of evaporation and salinity.

[0085] Step S1026: Based on the target weight, feature fusion is performed on the evaporation-temperature-nitrogen coupling feature, salt concentration-pH-nitrogen speciation coupling feature, and precipitation-conductivity coupling feature to generate the target fused feature.

[0086] Specifically, the electronic device 8 can perform feature fusion on the evaporation-temperature-nitrogen coupling features, salt concentration-pH-nitrogen speciation coupling features, and precipitation-conductivity coupling features based on the target weights to generate target fused features.

[0087] Step S103: Input the target fusion features into the output layer of the initial soil column data prediction network.

[0088] Specifically, the electronic device 8 can input the target fusion features into the output layer of the initial soil column data prediction network.

[0089] Step S104: The output layer outputs the virtual soil column temperature, virtual soil column pH value, and virtual soil column nitrogen transport amount corresponding to each depth.

[0090] Specifically, based on the simulated soil column at various depths (5cm, 15cm, 25cm, 35cm, 45cm), the virtual parameters of the output layer are strictly divided according to the depth dimension, forming a matrix output of "depth dimension × parameter dimension". For example, with five depths: 5cm, 15cm, 25cm, 35cm, and 45cm, the output is a matrix of five depths × three parameters. The parameters for each depth correspond independently, ensuring that the virtual results are consistent with the physical zoning of the stratified monitoring. For instance, the virtual nitrogen transport at a depth of 5cm specifically reflects the surface nitrogen accumulation characteristics dominated by surface evaporation, while the depth of 45cm focuses on reflecting the deep nitrogen migration patterns under the influence of the base 12 salt solution.

[0091] Electronic device 8 can constrain the virtual soil column temperature, virtual soil column pH value, and virtual soil column nitrogen transport amount based on the constraints corresponding to the virtual soil column temperature, virtual soil column pH value, and virtual soil column nitrogen transport amount, and then output the virtual soil column temperature, virtual soil column pH value, and virtual soil column nitrogen transport amount that meet the corresponding constraints in a "depth-parameter" manner.

[0092] For example, the constraints corresponding to the temperature of the virtual soil column include: vertical gradient constraint formula: Where d is the depth (5cm, 15cm, ..., 45cm), T v (d) represents the temperature of the virtual soil column at depth d. This formula ensures that the temperature difference between adjacent depths is ≤5℃ and the temperature decreases with depth, which conforms to the measured gradient law of the "layered temperature monitoring system".

[0093] The constraint formula relating to evaporation temperature: Among them, T 蒸发 The current evaporation temperature is (25℃, 40℃, 55℃), and Tv(5) is the virtual temperature of the 5cm surface layer. The formula limits the error between the surface temperature and the target evaporation temperature to within ±2℃, which is consistent with the physical characteristic that "the infrared lamp 21 acts directly on the surface layer".

[0094] The constraints on the pH value of the virtual soil column include: salt concentration coupled constraint formula: ; ;wherein: pH pred(d) represents the virtual soil column pH value corresponding to depth d (5cm, 15cm, 25cm, 35cm, 45cm), and pHv(d) represents the baseline soil column pH value corresponding to depth d (5cm, 15cm, 25cm, 35cm, 45cm). Csalt represents the current salt concentration (0.5%-2.0%). ΔpH(d) is a depth correction term (+0.3 for deep layers 35cm and 45cm, and -0.2 for surface layers 5cm and 15cm), simulating the characteristic that "deeper layers are more affected by salt solution supply." If the absolute deviation between the virtual soil column pH value and the corresponding baseline soil column pH value exceeds the range, the baseline soil column pH value pHv(d) is used as the benchmark for correction to ensure that the virtual soil column pH value is consistent with the rule that "the higher the salt concentration and the greater the influence of salt solution supply on deeper layers, the higher the pH value," reflecting the fundamental driving role of salt concentration on saline-alkali environments.

[0095] Precipitation impact constraint formula: ; ; in: The pH value of the soil column at depth d after precipitation correction is given. final (d) represents the final soil column pH value at depth d after correction based on the salt concentration coupling constraint formula, where P is the current precipitation; ω(d) is the depth weight (1.0 for the surface layer and 0.2 for the deep layer), reflecting that "precipitation has a stronger leaching effect on surface salts and alkalis".

[0096] The constraints on nitrogen transport in the virtual soil column include: the formula for the migration direction constraint: ; Among them, for evaporation-dominant (T) 蒸发 ≥40℃ and P<10mm).

[0097] For leaching-dominant (P≥30mm): Among them, N pred (d) represents the nitrogen transport volume of the virtual soil column at depth d, N. v (d) represents the nitrogen transport volume of the standard soil column corresponding to depth d, and N0 represents the nitrogen transport volume of the standard soil column corresponding to depth 0, which is the nitrogen transport volume of the standard soil column corresponding to the surface of the simulated soil column.

[0098] Step S105: Based on the first relationship error between the measured soil column temperature and the virtual soil column temperature, the second relationship error between the virtual soil column pH value and the measured soil column pH value, and the third relationship error between the virtual soil column nitrogen transport amount and the measured soil column nitrogen transport amount, the initial soil column data prediction network is trained to obtain the target soil column data prediction model.

[0099] Specifically, the error of the first relation is: ; Among them, T 实测，i Let T be the measured temperature of the soil column corresponding to the depth of the i-th layer. 虚拟，i K represents the virtual soil column temperature corresponding to depth i. t This is the time decay factor.

[0100] The error in the second relation is: ; Among them, pH 实测，i Here is the measured pH value of the soil column corresponding to the depth of the i-th layer, pH 虚拟，i Let EG be the pH value of the virtual soil column corresponding to the depth of the i-th layer. i Let be the measured electrical conductivity of the soil column at the depth of the i-th layer.

[0101] The error in the third relation is: ; Where, N 实测，i N represents the measured nitrogen transport in the soil column corresponding to the depth of the i-th layer. 虚拟，i Let D be the nitrogen transport amount of the virtual soil column corresponding to the depth of the i-th layer. i This is the direction coefficient.

[0102] The total loss error is generated based on the first relation error, the second relation error, and the third relation error.

[0103] Specifically, the total loss error = a1 first relation error + a2 second relation error + a3 third relation error.

[0104] Where a1, a2, and a3 are the weighting coefficients corresponding to the first relation error, the second relation error, and the third relation error, respectively.

[0105] In one optional embodiment of this application, when the temperature difference between virtual soil columns at adjacent depths exceeds a preset temperature difference threshold, the first relationship error value is increased to a first preset multiple. This first preset multiple can be 1.3, 1.5, or 2. For example, the temperature difference between adjacent depths (such as 5cm and 15cm, 15cm and 25cm) is typically ≤5℃ (gentle gradient). If the model predicts a temperature difference between virtual soil columns at adjacent depths >5℃ (e.g., 5cm temperature 25℃ and 15cm temperature 10℃, a temperature difference of 15℃), it is determined to be an "anti-physical gradient," and the value of LT (temperature gradient error) is doubled.

[0106] In one optional embodiment of this application, when the measured conductivity corresponding to the depth of the i-th layer is greater than a preset conductivity threshold, the second relationship error is amplified to a second preset multiple, wherein the second preset multiple can be 1.2, 1.3, or 1.5. Since the pH value increases exponentially with increasing salinity when the conductivity of saline-alkali soil is >1.5% (high salinity area) (e.g., pH 8.5 at 1.5% conductivity, pH 9.8 at 2.0%), and the pH value in this range has a strong inhibitory effect on the activity of nitrifying bacteria, directly affecting nitrogen migration, when the measured conductivity corresponding to the depth of the i-th layer is >1.5%, the pH error weight predicted by the model is automatically increased by 50% (e.g., the original error of 1.0 is amplified to 1.5).

[0107] Based on the total loss error, the initial soil column data prediction network is trained to obtain the target soil column data prediction model.

[0108] Specifically, in the initial stage (loss > 0.3): a high learning rate (0.01) is used to quickly converge the basic parameters; in the intermediate stage (0.1 < loss ≤ 0.3): the learning rate is reduced to 0.005 to focus on error correction in complex scenarios such as high salinity and strong evaporation; in the convergence stage (loss ≤ 0.1): the learning rate is dynamically reduced by "loss × 0.1" to avoid overfitting. After each iteration, specific verification is performed to eliminate parameter updates that violate the rules. Among these, infrared radiation response verification ensures that the sensitivity of nitrogen transport at a depth of 5cm to evaporation temperature is more than three times that at 45cm (radiation attenuation characteristics); salinity suppression verification: when conductivity increases from 1.0% to 2.0%, the nitrogen transport predicted by the model should decrease by ≥40% (high salinity suppression effect); leaching effect verification: when precipitation is ≥20mm, nitrogen transport at a depth of 15cm should be higher than that at 5cm (precipitation leaching trend). If the verification fails, the corresponding layer parameters are fine-tuned in reverse.

[0109] After a preset number of iterations, or after the total loss error tends to stabilize, the target soil column data prediction model is obtained.

[0110] The soil evaporation testing system provided in this application uses the current evaporation temperature, current precipitation, and current salt concentration as shared features input to the first sub-input layer. These three types of parameters are global environmental factors affecting the overall soil condition (such as evaporation rate and salt distribution). Separate input allows the model to prioritize capturing their macroscopic effects on the soil, avoiding interference from local parameters and strengthening the learning of global patterns. The current gas diffusion rate, measured soil column temperature, measured soil column pH, measured soil column conductivity, and measured soil column nitrogen transport at each depth are input to the second sub-input layer according to a preset acquisition gradient. This fully preserves the parameter differences at different soil depths (such as the difference between surface and deep temperatures and permeability), allowing the model to clearly learn the spatial distribution patterns of soil conditions and avoiding prediction bias caused by confusion of layered information. The input is split into "global shared features + layered local features," enabling subsequent network layers to design differentiated processing strategies for the two types of features (e.g., shared features focus on overall correlation analysis, while layered features focus on spatial gradient modeling), reducing cross-interference between features and improving the model's processing efficiency and accuracy for information of different dimensions.

[0111] The embedding layer encodes the current evaporation temperature, current precipitation, current salt concentration, and the current gas diffusion rate, measured soil column temperature, measured soil column pH, measured soil column electrical conductivity, and measured soil column nitrogen transport at each depth, based on the data type of each input data. This ensures that similar data (such as temperature and salt classifications) have consistent representation in the feature space, providing a unified basis for subsequent coupling analysis and reducing interference from data heterogeneity. The embedding layer extracts features from the current evaporation temperature, the measured soil column temperature at each depth, and the measured soil column nitrogen transport, generating evaporation-temperature-nitrogen coupling features to capture the linkage effect of evaporation on soil temperature changes and nitrogen migration (e.g., high-temperature evaporation accelerates nitrogen volatilization). The embedding layer obtains the measured soil column electrical conductivity at each depth and checks whether the measured soil column electrical conductivity is less than or equal to a preset conductivity threshold. If the measured soil column conductivity is less than or equal to a preset conductivity threshold, the initial salt concentration-pH-nitrogen speciation coupling characteristics are calculated based on the first relationship between the current salt concentration, the measured soil column pH value, and the measured nitrogen transport in the soil column. If the measured soil column conductivity is greater than the preset conductivity threshold, the initial salt concentration-pH-nitrogen speciation coupling characteristics are calculated based on the second relationship between the current salt concentration, the measured soil column pH value, and the measured nitrogen transport in the soil column. First, the relationship between the measured soil column conductivity and the preset conductivity threshold is detected. The first relationship is used for low-salt environments (≤ preset conductivity threshold), and the second relationship is used for high-salt environments (> preset conductivity threshold). This allows for accurate matching of the "salt concentration-pH-nitrogen" interaction pattern under different salt levels (e.g., high salt concentration enhances the effect of pH on nitrogen speciation), avoiding misjudgments caused by a single calculation method. Based on the depth of the simulated soil column, the initial salt concentration-pH-nitrogen speciation coupling characteristics are corrected to obtain the final salt concentration-pH-nitrogen speciation coupling characteristics. Correcting the initial salt concentration-pH-nitrogen speciation coupling features based on the simulated soil column depth (e.g., adjusting salt concentration weights due to higher salt accumulation in deeper soil layers) reflects the spatial differences in the "salt-pH-nitrogen" relationship at different soil depths. This avoids directly applying surface patterns to deeper layers and improves the adaptability of the coupling features to changes in the vertical dimensions of the soil. Through dual optimization of "conductivity level classification + depth correction," the salt concentration-pH-nitrogen speciation coupling features can reflect both the influence of salt levels and depth differences, more realistically depicting the complex relationships among these three elements in the soil system and providing more reliable feature support for subsequent model predictions.

[0112] The embedded layer extracts features from the measured soil column electrical conductivity at various depths based on the current precipitation, yielding precipitation-conductivity coupling features that reflect the dynamic relationship between precipitation leaching and soil salinity migration. These three types of coupling features accurately cover the core mechanisms of the soil system, avoiding feature isolation. Initial weights are assigned to the evaporation-temperature-nitrogen coupling features, salt concentration-pH-nitrogen speciation coupling features, and precipitation-conductivity coupling features based on preset durations. (For example, the evaporation-temperature-nitrogen coupling effect is more significant in short-term experiments, so it is assigned a higher initial weight; the salt concentration-pH-nitrogen speciation coupling effect is more persistent in long-term experiments, so its weight is increased). This allows the initial weights to align with the core driving factors of soil changes at different time scales, establishing a reasonable weighting foundation. Based on the depth of the simulated soil column, the initial weights are adjusted to obtain the backup weights for the evaporation-temperature-nitrogen coupling features, salt concentration-pH-nitrogen speciation coupling features, and precipitation-conductivity coupling features at each depth. (For example, surface soil is greatly affected by evaporation and precipitation, so the reserve weights for evaporation-temperature-nitrogen and precipitation-conductivity coupling characteristics are increased; deep soil is prone to salt accumulation, so the reserve weights for salt concentration-pH-nitrogen speciation coupling characteristics are increased.) This ensures that the weights reflect the differences in effects across the vertical dimensions of the soil, avoiding the one-sidedness of a "uniform weight across the entire depth." Based on the current evaporation temperature and precipitation, the reserve weights are adjusted to obtain the target weights. The reserve weights are further adjusted based on the current evaporation temperature and precipitation (e.g., strengthening the weight of evaporation-temperature-nitrogen coupling characteristics during high temperatures and increasing the weight of precipitation-conductivity coupling characteristics during heavy rainfall), allowing the target weights to match environmental changes in real time. This ensures that the influence of the three coupling characteristics can dynamically adjust with the actual environment during fusion, resulting in target features that better reflect the actual effects of the soil. Based on the target weights, the evaporation-temperature-nitrogen coupling characteristics, salt concentration-pH-nitrogen speciation coupling characteristics, and precipitation-conductivity coupling characteristics are fused to generate the target fused features. By integrating three types of coupled features with target weights, target features that combine multi-factor correlation and time adaptability are formed. This not only preserves the core laws of each dimension but also achieves information complementarity, providing more comprehensive and accurate feature support for subsequent predictions.

[0113] The output layer outputs virtual soil column temperature, pH value, and nitrogen transport at each depth. These indicators directly reflect soil thermal conditions, acid-base balance, and nutrient loss, aligning with the research focus on soil evaporation and salt / nutrient transport, resulting in highly practical output values. Based on the first relationship error between measured and virtual soil column temperatures, the second relationship error between virtual and measured soil column pH values, and the third relationship error between virtual and measured soil column nitrogen transport, the initial soil column data prediction network is trained to obtain the target soil column data prediction model. The model's bias in predicting different indicators is simultaneously corrected to avoid the "over-compensation" effect caused by optimizing a single error, ensuring high accuracy in multi-dimensional indicator predictions and meeting the prediction needs of actual experiments. The first relational error calculation incorporates a time decay factor to reflect the decay pattern of soil temperature over experimental time (e.g., significant temperature changes in the short term, Kt approaches 1; after a long period, the temperature tends to stabilize, and Kt decreases), avoiding the neglect of the time dimension's influence on temperature prediction and improving the accuracy of temperature error calculation. The second relational error combines pH error with the measured conductivity EPi at the corresponding depth. Since soil pH is closely related to salinity (conductivity reflects salinity levels) (e.g., high salinity environments easily affect pH stability), this design allows pH error calculation to adapt to different salinity conditions, reducing the one-sidedness of comparing a single pH value. The third relational error incorporates a direction coefficient. Nitrogen transport exhibits directional differences (e.g., upward volatilization or downward leaching), and the direction coefficient Di can correct errors under different transport directions (e.g., Di takes a positive value for leaching and a negative value for volatilization), making the nitrogen transport error calculation more closely match actual migration patterns and avoiding error distortion caused by directional confusion. Based on the total loss error generated by three types of targeted errors, the biases in the prediction of temperature, pH value and nitrogen transport can be corrected simultaneously when training the network in reverse. This avoids the "loss of one thing for another" caused by optimizing a single error, and enables the target model to have high accuracy in predicting multi-dimensional soil indicators, meeting the prediction needs of complex characteristics of soil systems.

[0114] In one alternative implementation, such as Figure 4 As shown, the soil column simulation device 1 includes a soil column container 11 and a base 12; the base 12 is installed below the soil column container 11, and the base 12 is provided with a water inlet, which can supply salt solution to the bottom of the simulated soil column by connecting to the salt concentration supply device 6 to simulate salt transport.

[0115] Specifically, the soil column simulation device 1, as the core load-bearing structure of the entire soil evaporation test system, achieves accurate simulation of the saline-alkali soil environment through the coordinated design of the soil column container 11 and the base 12. Its working principle is as follows: The soil column container 11 is made of transparent acrylic material (20cm inner diameter, 50cm height), providing a closed space for filling the simulated soil column. Its transparency facilitates direct observation of the soil profile's moisture, salt distribution, and nitrogen migration. Sampling holes and temperature probe mounting holes (corresponding to depths of 5cm, 15cm, etc.) are set every 10cm along the vertical direction of the container wall, providing installation interfaces for non-destructive sampling systems and stratified temperature monitoring systems, enabling precise collection of soil parameters at different depths.

[0116] The base 12 is installed below the soil column container 11. Its water inlet is connected to the salt concentration supply device 6, which can precisely supply salt solutions of different concentrations to the bottom of the simulated soil column. This design simulates the process of "groundwater replenishing salt" in natural saline-alkali soil. By controlling the total amount and concentration of salt supplied, it can simulate soil environments with different degrees of salinization, providing a variable basis for studying the influence of salt concentration on nitrogen transport and evaporation processes.

[0117] The soil column container 11 and the base 12 together constitute a controllable simulation system: the salt solution supplied by the base 12 moves upward through the soil pores, simulating the distribution of salt in the soil. Combined with the temperature-controlled evaporation device 2 and the precipitation device 4 above, it can simulate evaporation and precipitation processes under different climatic conditions; the transparent container facilitates monitoring the water and salt transport and nitrogen migration status of the soil throughout the process, providing a stable environmental carrier for subsequent experimental data collection and analysis.

[0118] The soil evaporation test system provided in this application embodiment has a base 12 with a water inlet connected to a salt concentration supply device 6, which can supply salt solution to the bottom of the simulated soil column, reproducing the process of salt migration from deep soil to the surface in the natural environment. This provides a realistic test scenario for studying the effects of salt on soil temperature, pH value, and other indicators, avoiding test deviations caused by unrealistic salt supply methods. The soil column container 11 is made of transparent material, allowing for direct observation of the layered state of the simulated soil column (such as salt accumulation layer and water infiltration layer) and physical changes during the test (such as soil cracking and water infiltration). It also facilitates the installation and positioning of the layered data acquisition device 3 and the air permeability sensor 7, reducing monitoring inconvenience caused by container obstruction.

[0119] In one alternative implementation, such as Figure 4 As shown, the stratified data acquisition device 3 includes a temperature sensor group 31, a conductivity sensor group 32, a pH sensor group 33, and a non-destructive sampling device 34, wherein: Temperature sensor group 31 is used to monitor the measured soil column temperature at different depths of the simulated soil column in real time after a preset time. Temperature sensor group 31 consists of multiple electronic thermometers, and the temperature probe of each electronic thermometer adopts a high-precision sensor. Each temperature probe is inserted into the simulated soil column at different depths through the temperature probe mounting hole on the wall of the soil column container 11 in the soil column simulation device 1. At least one temperature probe is set at each depth. The conductivity sensor group 32 is used to monitor the measured conductivity of the simulated soil column at different depths in real time after a preset time. The conductivity sensor group 32 consists of multiple conductivity sensors. Each conductivity sensor 32 is inserted into the simulated soil column at different depths through the conductivity sensor 32 mounting hole on the wall of the soil column container 11 in the soil column simulation device 1. At least one conductivity sensor 32 is set at each depth. pH sensor group 33 is used to monitor the measured pH value of the simulated soil column at different depths in real time after a preset time. The pH sensor group 33 consists of multiple pH sensors. Each pH sensor is inserted into the simulated soil column at different depths through the installation hole of the conductivity sensor 32 on the wall of the soil column container 11 in the soil column simulation device 1. At least one pH sensor is set at each depth. The non-destructive sampling device 34 is used to monitor the measured nitrogen transport in the simulated soil column at different depths in real time after a preset time. The non-destructive sampling device 34 includes a pore tube, a vacuum tube, and a measuring device for the measured nitrogen transport in the soil column. The pore tube is made of inert material, and its outer diameter matches the sampling hole on the wall of the soil column container 11. Its length is determined according to the height of the soil column container 11. One end of the pore tube is closed, and the other end is connected to the vacuum tube. The wall of the pore tube is distributed with micropores of the first size in diameter, which only allow pore water to pass through and prevent soil particles from entering. In use, the pore tube is inserted into the simulated soil column at different depths through the sampling hole. Pore water enters the pore tube under negative pressure and flows into the vacuum tube, thus achieving non-destructive sampling. The measured nitrogen transport in soil columns is used to simulate the pore water at various depths of a soil column, detect the amount of nitrogen transported in the pore water, and obtain the measured nitrogen transport in the soil column.

[0120] Specifically, the temperature sensor group 31 consists of multiple electronic thermometers equipped with high-precision sensors. Its temperature probes measure a range of -50℃ to 110℃ with an accuracy of ±0.1℃. During operation, each temperature probe is inserted into a simulated soil column at different depths (e.g., 5cm, 15cm) through dedicated mounting holes on the wall of the soil column container 11. At least one probe is installed at each depth to ensure complete coverage of the soil profile. After a preset time, the probes transmit real-time temperature data to a digital display screen via sensor lines, accurately reflecting the soil temperature differences at different depths. This provides fundamental data for analyzing the impact of temperature on nitrogen transformation (such as nitrification) and evaporation rates.

[0121] The conductivity sensor group 32 consists of multiple conductivity sensors 32. These sensors are inserted into simulated soil columns at different depths through dedicated mounting holes on the wall of the soil column container 11, with at least one sensor at each depth. Its core function is to monitor the conductivity of the soil at each depth in real time. Since conductivity is directly related to soil salinity, the conductivity data can indirectly reflect the vertical distribution and transport patterns of salt. This monitoring result corresponds to the salt concentration supplied by the salt solution system, providing key parameters for studying the impact of salt concentration on nitrogen transport.

[0122] The pH sensor group 33 consists of multiple pH sensor groups 33. The sensors are inserted into simulated soil columns at different depths through mounting holes (which can be shared with or independently set up with conductivity sensors 32) on the wall of the soil column container 11, with at least one sensor at each depth. Its function is to monitor the pH value of the soil at each depth in real time, quantifying the differences in acidity and alkalinity in saline-alkali soils. Since pH value is an important factor affecting nitrogen form transformation (such as nitrate stability), this data, combined with conductivity and nitrogen transport, can reveal the constraint mechanism of nitrogen migration on saline-alkali environments.

[0123] The non-destructive sampling device 34 consists of a pore tube, a vacuum tube, and a nitrogen transport measurement device, and is the core component for achieving "non-destructive sampling".

[0124] Sampling process: The pore tube is made of inert material, with micropores of 0.15 μm in diameter distributed on the tube wall (allowing only pore water to pass through while preventing soil particles from entering). One end is closed, and the other end is connected to a vacuum tube. In use, the pore tube is inserted into simulated soil columns of different depths through the sampling holes on the wall of the soil column container 11. Pore water enters the pore tube under negative pressure and flows into the vacuum tube, achieving sampling without damaging the soil structure.

[0125] Nitrogen determination: The nitrogen transport measurement equipment of the soil column (such as the ultraviolet-visible spectrophotometer in the laboratory) is used to test the collected pore water, focusing on the nitrate content, and then determining the nitrogen transport at each depth, filling the gap that traditional devices cannot accurately capture the dynamic changes of nitrogen.

[0126] The four sets of devices work together to synchronously collect data on temperature, electrical conductivity, pH value, and nitrogen transport after a preset time period, forming a complete dataset of "environmental factors - soil physicochemical properties - nitrogen migration". This data not only reflects the vertical distribution of a single parameter but also reveals the coupling relationship between "water and salt transport accompanied by nitrogen migration" in saline-alkali soils through multi-parameter correlation analysis, providing reliable experimental evidence for research on nitrogen cycle mechanisms in saline-alkali lands and optimization of water-saving irrigation technologies.

[0127] The soil evaporation test system provided in this application embodiment employs a high-precision electronic thermometer as its temperature sensor group 31. At least one probe is inserted into simulated soil columns at different depths, accurately capturing subtle differences in temperature at each depth (e.g., the temperature difference between the surface and deeper layers) while avoiding the randomness of single-point monitoring. This provides a reliable temperature data foundation for subsequent temperature error calculation and model training. Conductivity sensors 32 are inserted into the soil columns at different depths, with at least one at each depth, to monitor the conductivity (reflecting salinity levels) of the soil at different depths in real time. This clearly presents the vertical distribution pattern of salt in the soil (e.g., whether salt accumulates at a certain depth), providing accurate salinity data support for the extraction of salt concentration-pH-nitrogen speciation coupling features. pH sensors are inserted into the soil columns at different depths to monitor soil pH values ​​in real time, accurately reflecting the acid-base differences at different depths (e.g., the difference between the pH value of the surface layer leached by precipitation and that of the deeper layers). This avoids pH averaging errors caused by mixed sampling, providing accurate data for subsequent pH error calculation and soil acid-base balance research. The pore tubes, made of inert material and allowing only pore water to pass through, can collect pore water at different depths without damaging the simulated soil column structure (avoiding soil particle disturbance). Combined with vacuum tube negative pressure sampling, the operation is convenient and can accurately acquire pore water at various depths, providing contamination-free and structurally unaffected samples for nitrogen transport measurement. Simultaneously, it ensures the soil column can be reused for subsequent experiments, improving experimental efficiency. For nitrogen transport detection at various pore water depths, it can directly and accurately obtain nitrogen migration data from soils at different depths (such as whether nitrogen is leached to deeper layers with water), avoiding errors from traditional sampling methods. This provides a reliable basis for nitrogen transport-related feature extraction and error calculation, assisting the model in learning nitrogen migration patterns.

[0128] In one alternative implementation, such as Figure 4 As shown, the temperature-controlled evaporation device 2 includes an infrared lamp 21, a lamp cover 22, a temperature controller 23, and a temperature sensor 24. The infrared lamp 21 is installed above the soil column simulation device 1, and the infrared radiation is focused by the lamp cover 22 so that the radiant heat is evenly applied to the surface of the simulated soil column. The temperature sensor 24 is installed on the surface of the simulated soil column and is used to monitor the temperature of the simulated soil column surface in real time. The temperature controller 23 is used to control the infrared lamp 21 according to the temperature detected by the temperature sensor 24.

[0129] Specifically, infrared lamp 21 is installed above the soil column simulation device 1 as a radiant heat source to simulate the heat generated by solar radiation. Lamp cover 22 can focus infrared radiation, so that the radiant heat is evenly applied to the surface of the simulated soil column, ensuring that the surface of the soil column is heated evenly, which is closer to the effect of solar radiation in the natural environment.

[0130] Temperature sensor 24 is installed on the surface of the simulated soil column, which can monitor the temperature of the soil column surface in real time and transmit the monitored temperature data to temperature controller 23.

[0131] The temperature controller 23 compares the real-time temperature detected by the temperature sensor 24 with the preset target evaporation temperature (e.g., 25℃, 40℃, 55℃) and automatically adjusts the power of the infrared lamp 21. When the real-time temperature is lower than the target temperature, the temperature controller 23 increases the power of the infrared lamp 21 to enhance the radiant heat output; when the real-time temperature is higher than the target temperature, it decreases the power of the infrared lamp 21 to reduce the radiant heat output, thereby achieving precise control of the evaporation temperature on the surface of the simulated soil column and providing a stable temperature environment for exploring the nitrogen migration process under different evaporation intensities.

[0132] This temperature-controlled evaporation device 2 overcomes the limitations of traditional devices with their single evaporation regulation through this closed-loop control method, and can more realistically simulate the evaporation process driven by radiative heat under natural conditions, thereby improving the reliability of experimental results.

[0133] The soil evaporation test system provided in this application embodiment uses an infrared lamp 21 to provide a radiant heat source. The lamp cover 22 focuses the radiant heat, ensuring that the heat is evenly distributed on the surface of the simulated soil column. This avoids uneven evaporation rates caused by localized overheating, ensuring a consistent temperature distribution on the soil column surface. This provides stable heat source conditions for simulating natural evaporation environments and reduces interference from uneven heating on test data. A temperature sensor 24 is installed on the surface of the simulated soil column, providing real-time feedback on the current surface temperature. This allows the electronic equipment 8 to accurately obtain the heating effect, providing real-time data for subsequent temperature control and avoiding the inaccuracy of relying on the power of the infrared lamp 21 to indirectly determine the temperature. The temperature controller 23 adjusts the infrared lamp 21 in real-time based on the sensor monitoring data (increasing power when the temperature is below the target value and decreasing power when it is above), forming a closed-loop control. This ensures that the surface temperature of the simulated soil column remains stable at a preset value, accurately simulating evaporation temperature conditions under different environments and improving the controllability and repeatability of the test.

[0134] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A soil evaporation testing system, characterized in that, The soil evaporation test system includes: a soil column simulation device, a temperature-controlled evaporation device, a stratified data acquisition device, a precipitation device, a salt concentration acquisition device, a salt concentration supply device, an air permeability sensor, and electronic equipment; wherein, the temperature-controlled evaporation device is installed above the soil column simulation device, the precipitation device is installed above the soil column simulation device, the stratified data acquisition device and the air permeability sensor are both installed inside the simulated soil column in the soil column simulation device, the salt concentration acquisition device is installed below the soil column simulation device, and the salt concentration supply device is installed on one side of the soil column simulation device; the temperature-controlled evaporation device, the stratified data acquisition device, the precipitation device, and the salt concentration acquisition device are all communicatively connected to the electronic equipment; wherein: The soil column simulation device is used to hold the simulated soil column; The temperature-controlled evaporation device is used to regulate the current evaporation temperature of the simulated soil column surface under the control of the electronic device; The precipitation device is used to control the current precipitation corresponding to the simulated soil column under the control of the electronic device; The salt concentration acquisition device is used to acquire the current salt concentration below the simulated soil column; The air permeability sensor is used to collect the current gas diffusion rate of the simulated soil column at different depths; The salt concentration supply device is used to provide a salt solution to the simulated soil column; The layered data acquisition device is used to collect the measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity and measured soil column nitrogen transport at different depths after a preset time, under the conditions of the current evaporation temperature, the current precipitation, the current salt concentration and the current gas diffusion rate corresponding to each depth. The electronic device is used to input the current evaporation temperature, the current precipitation, the current salt concentration, and the current gas diffusion rate corresponding to each depth as input parameters, and the measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport amount corresponding to each depth as output parameters, to an initial soil column data prediction network, and to train the initial soil column data prediction network to obtain a target soil column data prediction model.

2. The soil evaporation test system according to claim 1, characterized in that, The electronic device is used for: The current evaporation temperature, the current precipitation, the current salt concentration, the current gas diffusion rate corresponding to each depth, the measured soil column temperature, the measured soil column pH value, the measured soil column electrical conductivity, and the measured soil column nitrogen transport amount are input into the input layer of the initial soil column data prediction network. The embedding layer in the initial soil column data prediction network extracts features from the current evaporation temperature, current precipitation, current salt concentration, current gas diffusion rate at each depth, measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport to obtain target fusion features. The target fusion features are input into the output layer of the initial soil column data prediction network; The output layer outputs the virtual soil column temperature, virtual soil column pH value, and virtual soil column nitrogen transport amount corresponding to each depth. Based on the first relationship error between the measured soil column temperature and the virtual soil column temperature, the second relationship error between the virtual soil column pH value and the measured soil column pH value, and the third relationship error between the virtual soil column nitrogen transport amount and the measured soil column nitrogen transport amount, the initial soil column data prediction network is trained to obtain the target soil column data prediction model.

3. The soil evaporation test system according to claim 2, characterized in that, The input layer includes a first sub-input layer and a second sub-input layer, and the electronic device is used for: The current evaporation temperature, the current precipitation, and the current salt concentration are input as shared features into the first sub-input layer; The current gas diffusion rate, measured soil column temperature, measured soil column pH value, measured soil column electrical conductivity, and measured soil column nitrogen transport amount corresponding to each depth are input to the second sub-input layer according to a preset acquisition gradient.

4. The soil evaporation test system according to claim 2, characterized in that, The electronic device is used for: The embedded layer encodes the current evaporation temperature, the current precipitation, the current salt concentration, and the current gas diffusion rate, the measured soil column temperature, the measured soil column pH value, the measured soil column electrical conductivity, and the measured soil column nitrogen transport amount corresponding to each depth, based on the data type corresponding to each input data. The embedded layer extracts features from the current evaporation temperature, the measured soil column temperature at each depth, and the measured nitrogen transport in the soil column to generate evaporation-temperature-nitrogen coupled features. The embedded layer extracts features from the current salt concentration, the measured electrical conductivity of the soil column at each depth, the measured pH value of the soil column, and the measured nitrogen transport amount of the soil column, generating a salt concentration-pH-nitrogen speciation coupling feature based on electrical conductivity. The embedded layer extracts features from the current precipitation and the measured electrical conductivity of the soil column corresponding to each depth to obtain precipitation-electrical conductivity coupling features. Based on the preset duration, the target weights corresponding to the evaporation-temperature-nitrogen coupling feature, the salt concentration-pH-nitrogen speciation coupling feature, and the precipitation-conductivity coupling feature are determined respectively. Based on the target weights, the evaporation-temperature-nitrogen coupling features, the salt concentration-pH-nitrogen speciation coupling features, and the precipitation-conductivity coupling features are fused to generate the target fused features.

5. The soil evaporation test system according to claim 4, characterized in that, The electronic device is used for: Obtain the measured electrical conductivity of the soil column at each depth, and detect whether the measured electrical conductivity of the soil column is less than or equal to a preset electrical conductivity threshold. If the measured electrical conductivity of the soil column is less than or equal to the preset electrical conductivity threshold, then the initial salt concentration-pH-nitrogen speciation coupling characteristics are calculated based on the first relationship between the current salt concentration, the measured pH value of the soil column, and the measured nitrogen transport amount of the soil column. If the measured electrical conductivity of the soil column is greater than the preset electrical conductivity threshold, the initial salt concentration-pH-nitrogen speciation coupling characteristics are calculated based on the second relationship between the current salt concentration, the measured pH value of the soil column, and the measured nitrogen transport amount of the soil column. Based on the depth corresponding to the simulated soil column, the initial salt concentration-pH-nitrogen speciation coupling feature is corrected to obtain the salt concentration-pH-nitrogen speciation coupling feature.

6. The soil evaporation test system according to claim 4, characterized in that, The electronic device is used for: Based on the preset duration, the initial weights corresponding to the evaporation-temperature-nitrogen coupling characteristics, the salt concentration-pH-nitrogen speciation coupling characteristics, and the precipitation-conductivity coupling characteristics are determined respectively. Based on the depth corresponding to the simulated soil column, the initial weights are adjusted to obtain the backup weights corresponding to the evaporation-temperature-nitrogen coupling characteristics, the salt concentration-pH-nitrogen speciation coupling characteristics, and the precipitation-conductivity coupling characteristics at each depth. Based on the current evaporation temperature and the current precipitation, the standby weights are adjusted to obtain the target weights.

7. The soil evaporation test system according to claim 4, characterized in that, The error of the first relationship is: ； Among them, T 实测，i T represents the measured soil column temperature corresponding to the depth of the i-th layer. 虚拟，i K represents the temperature of the virtual soil column corresponding to depth i. t This is the time decay factor; The error of the second relationship is: ； Among them, pH 实测，i The measured pH value of the soil column corresponding to the depth of the i-th layer is given by pH. 虚拟，i EG represents the pH value of the virtual soil column corresponding to the depth of the i-th layer. i Let be the measured electrical conductivity of the soil column at the depth of the i-th layer; The error of the third relationship is: ; Where, N 实测，i N represents the measured nitrogen transport amount in the soil column corresponding to the depth of the i-th layer. 虚拟，i D represents the nitrogen transport amount of the virtual soil column corresponding to the depth of the i-th layer. i Direction coefficient; Based on the first relational error, the second relational error, and the third relational error, a total loss error is generated; Based on the total loss error, the initial soil column data prediction network is trained to obtain the target soil column data prediction model.

8. The soil evaporation test system according to claim 1, characterized in that, The soil column simulation device includes a soil column container and a base; the base is installed below the soil column container and has a water inlet, which can supply salt solution to the bottom of the simulated soil column by connecting a salt concentration supply device to simulate salt transport.

9. The soil evaporation test system according to claim 1, characterized in that, The stratified data acquisition device includes a temperature sensor group, a conductivity sensor group, a pH sensor group, and a non-destructive sampling device, wherein: The temperature sensor group is used to monitor the measured soil column temperature at different depths of the simulated soil column in real time after the preset time. The temperature sensor group consists of multiple electronic thermometers, and the temperature probe of each electronic thermometer adopts a high-precision sensor. Each temperature probe is inserted into the simulated soil column at different depths through the temperature probe mounting hole on the wall of the soil column container in the soil column simulation device, and at least one temperature probe is set at each depth. The conductivity sensor group is used to monitor the measured conductivity of the simulated soil column at different depths in real time after the preset time. The conductivity sensor group consists of multiple conductivity sensors, and each conductivity sensor is inserted into the simulated soil column at different depths through the conductivity sensor mounting hole on the wall of the soil column container in the soil column simulation device. At least one conductivity sensor is set at each depth. The pH sensor group is used to monitor the measured pH value of the simulated soil column at different depths in real time after the preset time. The pH sensor group consists of multiple pH sensors, and each pH sensor is inserted into the simulated soil column at different depths through the conductivity sensor mounting hole on the wall of the soil column container in the soil column simulation device. At least one pH sensor is set at each depth. A non-destructive sampling device is used to monitor the measured nitrogen transport in the simulated soil column at different depths in real time after a preset time period. The non-destructive sampling device includes a pore tube, a vacuum tube, and a nitrogen transport measurement device for the measured soil column. The pore tube is made of inert material, with an outer diameter matching the sampling hole on the wall of the soil column container. Its length is determined according to the height of the soil column container. One end of the pore tube is closed, and the other end is connected to the vacuum tube. The wall of the pore tube has micropores with a diameter of a first size distributed on it, allowing only pore water to pass through and preventing soil particles from entering. In use, the pore tube is inserted into the simulated soil column at different depths through the sampling hole. Pore water enters the pore tube under negative pressure and flows into the vacuum tube, thus achieving non-destructive sampling. The measured nitrogen transport amount in the soil column is used to detect the pore water at each depth of the simulated soil column, determine the amount of nitrogen transport in the pore water, and obtain the measured nitrogen transport amount in the soil column.

10. The soil evaporation test system according to claim 1, characterized in that, The temperature-controlled evaporation device includes an infrared lamp, a lampshade, a temperature controller, and a temperature sensor. The infrared lamp is installed above the soil column simulation device, and the lampshade focuses infrared radiation so that the radiant heat is evenly applied to the surface of the simulated soil column. The temperature sensor is installed on the surface of the simulated soil column to monitor the temperature of the simulated soil column surface in real time. The temperature controller is used to control the infrared lamp based on the temperature detected by the temperature sensor.

Images