Soil splash erosion simulation analysis system

Abstract

The present application relates to the technical field of soil splash erosion, and particularly relates to a soil splash erosion simulation analysis system, which comprises: a rainfall simulator for simulating rainfall with different intensities and different pH values; a wind simulator for adjusting wind speed and wind direction; a raindrop characteristic measurement device for measuring raindrop data; a splash erosion simulation device for simulating the process of raindrop falling on target soil to cause splash erosion and generate splash erosion material; a data acquisition device for acquiring the initial pH value and initial conductivity of the target soil before rainfall, and acquiring the target pH value, target conductivity, target erosion depth of the target soil after rainfall, and the target splash erosion material average weight diameter and target total splash erosion amount of the splash erosion material; and an electronic device for determining and establishing a splash erosion prediction model based on the raindrop pH value, raindrop data, wind speed, wind direction and data acquired by the data acquisition device, and developing a soil splash erosion simulation analysis system capable of real-time and quantitative analysis of soil splash erosion.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of soil splash erosion, and particularly relates to a soil splash erosion simulation analysis system. BACKGROUND

[0002] Soil erosion, particularly splash erosion caused by raindrop impact, widely affects agricultural production and ecological environment. As raindrops hit the soil surface, soil particles are not only compacted or separated, but also transported by saltation or surface runoff, causing soil nutrient loss and soil structure deterioration. This process poses a potential threat to soil water retention, plant root growth, and farmland productivity.

[0003] Currently, research on soil splash erosion is mostly focused on its mechanism and influencing factors, however, in practical applications, especially for different soil types and different climate conditions, there is still a great gap in the dynamic monitoring and analysis of splash erosion. Traditional soil erosion evaluation methods often rely on laboratory tests or large-scale monitoring, lacking real-time and accurate quantitative analysis tools for soil splash erosion, and unable to accurately reflect the complexity and differences of raindrop impact under different environments.

[0004] Therefore, it is particularly necessary to develop a soil splash erosion simulation analysis system that can analyze soil splash erosion in real time and quantitatively. SUMMARY

[0005] In view of this, the present application provides a soil splash erosion simulation analysis system to develop a soil splash erosion simulation analysis system that can analyze soil splash erosion in real time and quantitatively.

[0006] In a first aspect, the present invention provides a soil splash erosion simulation and analysis system, comprising: a rainfall simulator, a splash erosion simulation device, a raindrop characteristic measurement device, a data acquisition device, a wind simulator, and an electronic device, wherein the rainfall simulator is mounted above the raindrop characteristic measurement device, the raindrop characteristic measurement device is mounted above the splash erosion simulation device, and the data acquisition device is mounted on the splash erosion simulation device; the wind simulator is mounted on one side of the raindrop characteristic measurement device; the raindrop characteristic measurement device, the wind simulator, and the data acquisition device are all communicatively connected to the electronic device, wherein: the rainfall simulator is used to simulate rainfall of different intensities and different pH values; the wind simulator is used to adjust wind speed and direction under the control of the electronic device; the raindrop characteristic measurement device is used to measure raindrop data, wherein the raindrop data includes raindrop diameter, raindrop terminal velocity, raindrop kinetic energy, and raindrop movement posture; Splash erosion simulation equipment is used to simulate the process of raindrops landing on target soil, causing splash erosion and generating splash erosion materials; data acquisition equipment is used to collect the initial pH value and initial conductivity corresponding to the target soil before rainfall; and collect the corresponding target pH value, target conductivity, target erosion depth of the target soil after rainfall, as well as the target average weight diameter of the splash erosion materials and the target total splash erosion amount; electronic equipment is used to determine and establish a splash erosion prediction model between raindrop data and soil splash erosion data based on raindrop pH value, raindrop data, wind speed, wind direction, initial pH value, initial conductivity, target pH value, target conductivity, target erosion depth, target average weight diameter of the splash erosion materials and the target total splash erosion amount; and to predict the predicted erosion depth, predicted average weight diameter of the splash erosion materials and predicted total splash erosion amount corresponding to other soils to be tested based on the splash erosion prediction model.

[0007] In an optional embodiment, the rainfall simulator includes: a rain barrel, a regulating pump, a rain nozzle, a first liftable support frame and a motor; wherein the regulating pump is connected to the rain barrel and the rain nozzle, the motor is installed on the rain nozzle, the first liftable support frame is installed below the rain nozzle, and the regulating pump, the rain nozzle, the first liftable support frame and the motor are all communicatively connected to the electronic device, wherein: the rain barrel is used to hold rainwater with different pH values; the regulating pump is used to control the flow rate of water drawn to the rain nozzle through the rotation speed under the control of the electronic device to simulate rainfall of different intensities; the motor is used to drive the rain nozzle to rotate under the control of the electronic device to simulate the effect of natural rainfall; the rain nozzle is used to adjust different calibers under the control of the electronic device to adjust the diameter of raindrops, thereby simulating raindrops of different sizes; the first liftable support frame is used to adjust the height of the rain nozzle under the control of the electronic device, thereby controlling the height of rainfall.

[0008] In an alternative embodiment, the wind simulator comprises a rotatable base, a second elevatable support frame, a mixed-flow fan group, a honeycomb-shaped flow guide device installed at the air outlet of the mixed-flow fan group, and the rotatable base, the second elevatable support frame, and the mixed-flow fan group are all in communication connection with the electronic device, wherein: the rotatable base is used to rotate in a direction to adjust the wind direction under the control of the electronic device; the second elevatable support frame is used to adjust the height to match the height of the rain nozzle in the rainfall simulator under the control of the electronic device; the mixed-flow fan group is used to adjust the rotating speed frequency to adjust the wind speed under the control of the electronic device; and the honeycomb-shaped flow guide device is used to effectively reduce the turbulence phenomenon generated when the fan blows air.

[0009] In an alternative embodiment, the sputtering simulation device comprises a sputtering disc, a collection drawer, and a waterproof cover, the sputtering disc is composed of a central source area and a collection area; the central source area is used to place a soil disc with a drainage hole at the bottom; the soil disc is used to hold the target soil; the collection area is composed of a plurality of concentric rings, the rings are separated by steel plates, and each ring area includes two symmetrically small holes for collecting sputtering materials, the positions of the small holes in the ring areas of adjacent rings are staggered, and the ring areas are kept sealed and waterproof; the collection drawer is installed below the sputtering disc and is used to place a sputtering material sample collection basin to improve the collection efficiency of the sputtering materials; and the waterproof cover is installed on the periphery of the sputtering disc to ensure that the raindrops only act on the soil disc.

[0010] In an alternative embodiment, the electronic device is used to calculate an electrical conductivity difference value based on an initial electrical conductivity and a target electrical conductivity, calculate a pH value difference value based on an initial pH value and a target pH value, input the raindrop pH value, the raindrop data, the wind speed, and the wind direction into a first branch of an initial sputtering prediction network to perform feature extraction on the raindrop pH value and the raindrop data to obtain external driving features, input the electrical conductivity difference value and the pH value difference value into a second branch of the initial sputtering prediction network to perform feature extraction on the electrical conductivity difference value and the pH value difference value to obtain soil response features, perform fusion processing on the external driving features and the soil response features to obtain target features, input the target features into an output layer to output a virtual erosion depth, a virtual sputtering material average weight diameter, and a virtual total sputtering amount, and train the initial sputtering prediction network based on the relationship between the virtual erosion depth, the virtual sputtering material average weight diameter, and the virtual total sputtering amount and a target erosion depth, a target sputtering material average weight diameter, and a target total sputtering amount to obtain a sputtering prediction model.

[0011] In an optional implementation, the electronic device is configured to: correct the raindrop diameter based on the wind speed to obtain a diameter correction coefficient; calculate a raindrop falling time based on a terminal velocity of the raindrop; correct the raindrop falling time based on a soil type corresponding to the target soil to obtain a time correction coefficient; correct the terminal velocity of the raindrop based on the wind speed to obtain a raindrop velocity correction coefficient; calculate a uniformity coefficient based on the wind speed and a raindrop posture; calculate a basic spatial range feature based on the diameter correction coefficient, the time correction coefficient, and the raindrop velocity correction coefficient; calculate an impact range feature based on the basic spatial range feature and the uniformity coefficient; obtain a raindrop kinetic energy density based on the raindrop kinetic energy divided by the diameter correction coefficient; correct the raindrop kinetic energy based on the wind direction, the wind speed, and the raindrop running posture to obtain a kinetic energy correction coefficient; calculate a synergy factor based on the soil type corresponding to the target soil and a pH value of the raindrop; calculate a direction intensity coefficient based on the raindrop running posture and an included angle between the wind direction and an impact direction; calculate an energy intensity based on the kinetic energy correction coefficient, the raindrop kinetic energy density, and the synergy factor; calculate an impact intensity feature based on the energy intensity and the direction intensity coefficient; and fuse the impact range feature and the impact intensity feature to generate the external driving feature.

[0012] In an optional implementation, the electronic device is configured to: correct the conductivity difference and the pH value difference based on a relationship between the conductivity difference and the pH value difference to obtain a corrected conductivity coefficient and a corrected pH value coefficient; calculate a soil buffer index based on a first relationship between the corrected conductivity coefficient and the corrected pH value coefficient; calculate a structure integrity feature based on a second relationship between the corrected conductivity coefficient and the corrected pH value coefficient; and fuse the soil buffer index and the structure integrity feature to generate the soil response feature.

[0013] In an optional implementation, the external driving feature includes an impact range feature and an impact intensity feature, and the soil response feature includes a soil buffer index and a structure integrity feature. The electronic device is configured to: assign initial weights to the external driving feature and the soil response feature, respectively; construct a correlation coefficient matrix of the external driving feature and the soil response feature; determine a target correlation feature with a correlation coefficient greater than a preset correlation threshold based on the correlation coefficient matrix; perform cross operation on the target correlation feature to generate an interaction feature; correct the initial weights based on the interaction feature to obtain target weights corresponding to the external driving feature and the soil response feature, respectively; and perform feature fusion on the external driving feature and the soil response feature based on the target weights to generate a target feature.

[0014] In an optional implementation, the electronic device is configured to: calculate an erosion relative error and an erosion absolute error based on the virtual erosion depth and the target erosion depth; determine first weight information corresponding to the erosion relative error and the erosion absolute error according to a soil type corresponding to the target soil; calculate an erosion error based on the first weight information, the erosion relative error, and the erosion absolute error; calculate a sputtering material average weight diameter square error based on the virtual sputtering material average weight diameter and the target sputtering material average weight diameter; calculate a first particle proportion in the virtual sputtering material average weight diameter that is greater than a preset diameter and a second particle proportion in the target sputtering material average weight diameter that is greater than the preset diameter; calculate a sputtering material average weight diameter error based on a deviation between the first particle proportion and the second particle proportion and the sputtering material average weight diameter square error; calculate a total sputtering amount error based on the virtual total sputtering amount and the target total sputtering amount; generate a target loss function based on the erosion error, the sputtering material average weight diameter square error, and the total sputtering amount error; and train the initial sputtering prediction network based on the target loss function to obtain the sputtering prediction model.

[0015] In an optional implementation, the electronic device is further configured to: calculate an erosion resistance capability index based on the target erosion depth; calculate a cluster breakage resistance capability index based on the target sputtering material average weight diameter; calculate an overall loss resistance capability index based on the target total sputtering amount; and obtain a target erosion resistance capability corresponding to the target soil based on the erosion resistance capability index, the cluster breakage resistance capability index, and the overall loss resistance capability index.

[0016] The soil splash erosion simulation analysis system provided by the embodiment of the application realizes the synchronous collection of the whole-chain parameters of "rainfall-wind-soil response" through the structured layout of each device. The rainfall simulator can accurately control the rainfall intensity and pH value, and combined with the wind speed and direction adjustment of the wind simulator, different climate conditions can be reproduced, solving the problem that the traditional single rainfall simulation cannot reflect the influence of wind. The raindrop characteristic measuring device captures core parameters such as raindrop diameter and kinetic energy in real time, and the splash erosion simulation device records the complete change of the soil from the initial state to the splash erosion state in cooperation with the data collection device, forming a closed-loop data chain of "raindrop parameters-soil response", and providing a comprehensive basis for subsequent modeling. The electronic device establishes a splash erosion prediction model based on the collected full-amount data, breaking through the limitations of traditional experience evaluation. The splash erosion prediction model input covers raindrop pH value, raindrop data, environmental factors and soil initial properties, and the output can be directly related to the soil splash erosion result, and can quantitatively predict the splash erosion risk in different scenarios. Compared with the traditional laboratory test, the splash erosion prediction model can quickly deduce the splash erosion result of the unknown scene through the historical data, reduce the repeated experiment cost, and provide a scientific prediction tool for the agricultural water and soil conservation scheme. If the average weight diameter of the target splash erosion material decreases and the total splash erosion amount is small under the same rainfall condition, it indicates that the soil has strong erosion resistance; if the initial pH buffer capacity is weak, the soil erosion is intensified, and soil improvement measures can be taken accordingly. The erosion resistance evaluation result is directly related to the priority of the prevention and control measures - the soil with weak erosion resistance can be given priority to implement water and soil conservation engineering, avoiding blind investment and improving the accuracy of soil protection.

[0017] In addition, the erosion resistance related capabilities are quantified from the three key links of erosion, aggregate breakage and loss, avoiding one-sided judgment of soil erosion resistance performance by a single index. The indexes of the three stages are integrated to completely cover the whole process of soil erosion resistance performance from being splashed and stripped to particle breakage and finally loss, reflecting the comprehensive erosion resistance level. The weak links of soil erosion resistance (such as weak anti-erosion and easy breakage of aggregates) can be directly located, providing clear basis for targeted improvement (such as increasing organic fertilizer to improve the stability of aggregates). BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the specific embodiments or the prior art, the following will briefly introduce the drawings needed to be used in the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0019] Figure 1 It is a structural schematic diagram of the soil splash erosion simulation analysis system according to the embodiment of the application;

[0020] Figure 2is a schematic structural diagram of a raindrop characteristic measuring device according to an embodiment of the present invention;

[0021] Figure 3 1 is a flow chart of another soil splash erosion simulation and analysis system method according to an embodiment of the present invention;

[0022] Figure 4 is a structural block diagram of a sputtering simulation device according to an embodiment of the present invention;

[0023] Figure 5 1 is a flow chart of establishing a splash erosion prediction model between raindrop data and soil splash erosion data according to an embodiment of the present invention;

[0024] Figure 6 1 is a schematic diagram of a process for obtaining a target corrosion resistance corresponding to a target soil according to an embodiment of the present invention;

[0025] Among them, there are a rainfall simulator 1; a splash erosion simulation device 2; a raindrop characteristic measurement device 3; a data acquisition device 4; a wind simulator 5; an electronic device 6; a rain barrel 11; a regulating pump 12; a rainfall nozzle 13; a first liftable support frame 14; a motor 15; a rotatable base 51; a second liftable support frame 52; a mixed flow fan unit 53; a honeycomb guide device 54; a splash erosion plate 21; a collection drawer 22; a waterproof cover 23; a central source area 211; a collection area 212; a soil plate 213; and a steel plate 214. DETAILED DESCRIPTION

[0026] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0027] The present application embodiment provides a soil splash erosion simulation analysis system, such as Figure 1 As shown, the system includes: a rainfall simulator 1, a splash erosion simulation device 2, a raindrop characteristic measurement device 3, a data acquisition device 4, a wind simulator 5 and an electronic device 6, wherein the rainfall simulator 1 is installed above the raindrop characteristic measurement device 3, the raindrop characteristic measurement device 3 is installed above the splash erosion simulation device 2, and the wind simulator 5 is installed on one side of the raindrop characteristic measurement device; the raindrop characteristic measurement device 3, the wind simulator 5 and the data acquisition device 4 are all communicatively connected to the electronic device 6, wherein:

[0028] Rainfall simulator 1, used to simulate rainfall of different intensities and pH values;

[0029] a wind simulator 5 for adjusting the wind speed and the wind direction under the control of the electronic device 6;

[0030] a raindrop characteristic measuring device 3 for measuring raindrop data, the raindrop data including a raindrop diameter, a raindrop terminal velocity, a raindrop kinetic energy and a raindrop running posture;

[0031] a splash erosion simulation device 2 for simulating a process in which raindrops falling on target soil cause splash erosion and generate splash erosion materials;

[0032] a data acquisition device 4 for acquiring an initial pH value and an initial conductivity of the target soil before rainfall, and acquiring a target pH value, a target conductivity, a target erosion depth of the target soil after rainfall, and a target average weight diameter of the splash erosion materials and a target total splash erosion amount corresponding to the splash erosion materials;

[0033] an electronic device 6 for establishing a splash erosion prediction model between the raindrop data and the soil splash erosion data based on the raindrop pH value, the raindrop data, the wind speed, the wind direction, the initial pH value, the initial conductivity, the target pH value, the target conductivity, the target erosion depth, and the target average weight diameter of the splash erosion materials and the target total splash erosion amount, and for predicting a predicted erosion depth, a predicted average weight diameter of the splash erosion materials and a predicted total splash erosion amount corresponding to other to-be-measured soil based on the splash erosion prediction model.

[0034] Specifically, the layout is as follows: Figure 1 The devices are deployed in linkage: the rainfall simulator 1 is installed directly above the raindrop characteristic measuring device 3 (the distance is 2-3 m to ensure the completeness of the raindrop falling path), the raindrop characteristic measuring device 3 is located directly above the splash erosion simulation device 2 (the distance is 1.5 m to accurately capture the characteristics of the raindrops before reaching the soil), and the wind simulator 5 is placed aside the raindrop characteristic measuring device 3 (the horizontal distance is 1.5 m to simulate the influence of natural wind on the raindrop trajectory). The electronic device 6 is connected to each device through a communication interface to complete parameter calibration (such as the corresponding relationship between the aperture of the rainfall nozzle 13 and the raindrop diameter, and the linear calibration between the fan speed and the wind speed).

[0035] The target soil is loaded into the soil disc 213 of the splash erosion simulation device 2 (the thickness is 5 cm, which matches the specification of the soil disc 213 of the splash erosion simulation device 2), and after being left for 24 hours, the initial state parameters are measured by the data acquisition device 4: the initial pH value (measured by a soil pH meter, 3 times for an average value) and the initial conductivity (measured by a portable conductivity meter, reflecting the basic state of soil salt content).

[0036] Electronic device 6 sends control instructions to rainfall simulator 1 and wind simulator 5. Rainfall simulator 1 adjusts the speed of pump 12 to set the rainfall intensity (e.g., 20 mm / h) and selects rainwater from rain barrel 11 with a corresponding pH value (e.g., pH 4.5 simulates acidic rain), simulating rainfall of varying intensities and pH values. Wind simulator 5 can adjust wind speed and direction.

[0037] As raindrops fall from the rainfall simulator 1, the raindrop characteristic measurement device 3 synchronously captures data: raindrop diameter (measured by a laser particle size analyzer with an accuracy of 0.1 mm), raindrop terminal velocity (the trajectory is captured by a high-speed camera and the instantaneous velocity is calculated), raindrop kinetic energy (calculated based on mass and velocity: kinetic energy = 0.5 × raindrop mass × velocity²), and raindrop movement posture (tilt angle α, reflecting the impact of wind on the raindrop trajectory), and transmits the raindrop data in real time to the electronic device 6.

[0038] Specifically, if Figure 2 As shown, the raindrop characteristic measuring device 3 is a remote sensing sensor comprising a transmitting end and a receiving end. At the transmitting end, a laser transmitter is used to generate a parallel laser beam; at the receiving end, the laser signal is received by a combination of a photodiode and a lens. By analyzing the attenuation and amplitude change of the laser beam when passing through precipitation particles, the parameters of raindrops produced by drippers of different diameters are measured, including the mean diameter of raindrops, the terminal velocity of raindrops, and the kinetic energy of raindrops. The raindrop characteristic measuring device records the number of raindrops passing through the measurement area per minute and measures the diameter and velocity of each raindrop. The raindrop cumulative volume percentage curve of the rainfall is used to obtain the corresponding raindrop diameter when the cumulative volume reaches 50%. This diameter is called the mean diameter of raindrops and is represented by D50. Assuming that the raindrop is a sphere, its volume is then calculated.

[0039] The kinetic energy of raindrops KE (Jm-2h-1) is calculated as follows:

[0040] (1)

[0041] Where m i represents the mass of raindrop i in kg; v i Indicates the speed of raindrop i in ms -2 .

[0042] The calculation formula for rainfall erosivity is expressed by the product of rainfall kinetic energy and rainfall intensity. This product is considered to be a key indicator for measuring rainfall erosivity.

[0043] (2)

[0044] Where R represents rainfall erosivity, KE is rainfall kinetic energy, and I 30 Indicates the maximum rainfall intensity within 30 minutes.

[0045] Under the action of wind, raindrops will be affected by wind during falling, resulting in the phenomenon of inclined falling. This makes the kinetic energy of raindrops increase, so the kinetic energy needs to be corrected according to the inclination angle of rainfall.

[0046] (3)

[0047] Wherein, KE f is the corrected kinetic energy of rainfall under the influence of wind; and a represents the included angle between the direction of raindrop movement and the vertical direction of gravity. The following is the calculation process of the included angle a.

[0048] According to the uniform motion of raindrops during falling, the gravity of raindrops is equal to the air resistance, and the equation is as follows:

[0049] (4)

[0050] After transformation, the equation can be rewritten as:

[0051] (5)

[0052] After transformation, the above equation can obtain the speed V s of raindrops in the vertical direction.

[0053] (6)

[0054] When raindrops move uniformly in the horizontal direction, the force can be represented by the Newton drag force formula as follows:

[0055] (7)

[0056] Through the above equation, the speed V w of raindrops in the horizontal direction can be obtained.

[0057] (8)

[0058] According to the calculation method of trigonometric function, the included angle a between the direction of raindrop movement and the vertical direction of gravity is the ratio of the speed V s of raindrops in the vertical direction to the speed V w of raindrops in the horizontal direction. After being brought into equation (7) and equation (8), the calculation result is as follows:

[0059] (9)

[0060] In the above equation, m is the mass of a single raindrop, g is the acceleration of gravity, F a is the air resistance of raindrops during falling along the gravity, r is the radius of raindrops, and p wis the water density, C is the air drag coefficient, is the air density, S r is the cross-sectional area of ​​the raindrop, and F is the horizontal force on the raindrop when it falls.

[0061] After raindrops and wind force act synergistically on the target soil, the data acquisition device 4 starts multi-dimensional recording: immediately after rainfall, the target pH value and target conductivity are measured (compared with the initial value to reflect the physical and chemical changes of the soil).

[0062] Data collection equipment 4 may include a soil pH meter, a soil conductivity meter, a laser rangefinder, and a concentric ring collector. The soil pH meter can measure the initial pH value of the target soil before rainfall and the target pH value of the target soil after rainfall. The soil conductivity meter can measure the initial conductivity of the target soil before rainfall and the target conductivity of the target soil after rainfall.

[0063] A laser rangefinder measures the initial height from the target soil before rainfall. Then, after rainfall, the target height from the target soil is measured. The target erosion depth is calculated by subtracting the initial height from the target height. Simultaneously, data acquisition equipment 4 uses concentric ring collectors to weigh the total mass of the eroded material in each ring area to obtain the target total erosion volume. Furthermore, the average weight diameter of the eroded material is determined by sieving. All data are linked to timestamps, forming a complete data chain from "initial state - simulation parameters - erosion results."

[0064] The electronic device 6 data is based on the raindrop pH value, raindrop data, wind speed, wind direction, initial pH value, initial conductivity, target pH value, target conductivity, target erosion depth, target average weight diameter of splashing material and target total splashing amount, to establish a splashing prediction model between the raindrop data and the soil splashing data; based on the splashing prediction model, the predicted erosion depth, predicted average weight diameter of splashing material and predicted total splashing amount corresponding to other soils to be tested are predicted.

[0065] The soil splash erosion simulation analysis system provided by the embodiments of the present application realizes the synchronous collection of the "rainfall-wind-soil response" full-chain parameters through the structured layout of each device (vertical linkage of the rainfall simulator 1, the raindrop characteristic measuring device 3 and the splash erosion simulation device 2, lateral cooperation of the wind simulator 5). The rainfall simulator 1 can accurately control the rainfall intensity and pH value, and can reproduce different climate conditions (such as acidic rainfall + crosswind, neutral rainfall + no wind) in combination with the wind speed and direction adjustment of the wind simulator 5, thereby solving the problem that the traditional single rainfall simulation cannot reflect the influence of wind. The raindrop characteristic measuring device 3 can capture core parameters such as raindrop diameter and kinetic energy in real time, the splash erosion simulation device 2 cooperates with the data acquisition device 4 to record the complete change (target erosion depth, average weight diameter of splash erosion material, etc.) of the soil from the initial state (initial pH, conductivity) to after the splash erosion, and forms a closed-loop data chain of "raindrop parameters-soil response", thereby providing a comprehensive basis for subsequent modeling. The electronic device 6 establishes a splash erosion prediction model based on the collected full-amount data, thereby breaking through the limitations of traditional experience evaluation. The splash erosion prediction model input covers raindrop pH value, raindrop data, environmental factors (wind speed, wind direction) and soil initial properties, and the output can be directly associated with the soil splash erosion result (total splash erosion amount, erosion depth, etc.), and can quantitatively predict the splash erosion risk in different scenarios (such as the splash erosion amount of sandy soil under the combination of "large-diameter raindrop + high wind speed"). Compared with the traditional laboratory test, the splash erosion prediction model can quickly deduce the splash erosion result of unknown scenarios through historical data, thereby reducing the repeated experiment cost and providing a scientific prediction tool for agricultural water and soil conservation schemes (such as the selection of farmland covering methods). If the average weight diameter of the target splash erosion material decreases by a small amount and the total splash erosion amount is small under the same rainfall condition, it indicates that the soil has strong erosion resistance (such as clay soil is usually better than sandy soil); if the initial pH buffer capacity is weak (the target pH change is large), which leads to the intensification of splash erosion, soil improvement measures (such as applying organic fertilizer to improve the stability of aggregates) can be taken accordingly. The erosion resistance evaluation result is directly related to the priority of the prevention and control measures - the soil with weak erosion resistance can be given priority to implement water and soil conservation engineering (such as terrace construction), thereby avoiding blind investment and improving the accuracy of soil protection.

[0066] In an optional embodiment of the present application, as shown in Figure 3 The rainfall simulator 1 comprises a rain barrel 11, an adjusting pump 12, a rainfall spray head 13, a first liftable support frame 14 and a motor 15. The adjusting pump 12 is connected with the rain barrel 11 and the rainfall spray head 13. The motor 15 is installed above the rainfall spray head 13. The first liftable support frame 14 is installed below the rainfall spray head 13. The adjusting pump 12, the rainfall spray head 13, the first liftable support frame 14 and the motor 15 are in communication connection with the electronic device 6.

[0067] The rain barrel 11 is used for containing rainwater with different pH values.

[0068] The regulating pump 12 is used to control the water flow extracted to the rain shower head 13 by rotating speed under the control of the electronic device 6, so as to simulate different intensities of rainfall.

[0069] The motor 15 is used to drive the rain shower head 13 to rotate under the control of the electronic device 6, so as to simulate the effect of natural rainfall.

[0070] The rain shower head 13 is used to adjust different diameters under the control of the electronic device 6, so as to adjust the diameter of raindrops and simulate raindrops of different sizes.

[0071] The first liftable support frame 14 is used to adjust the height of the rain shower head 13 under the control of the electronic device 6, so as to control the height of the rainfall.

[0072] Specifically, different pH value rainwater (such as acid rainwater with pH=5.0 and neutral rainwater with pH=7.0) is pre-configured in the rain barrel 11, and a pH value label is marked. The rain barrel 11 is connected with the inlet of the regulating pump 12 through a pipeline, so as to ensure that the water flow can be pumped into the rain shower head 13.

[0073] In the initial state, the first liftable support frame 14 adjusts the rain shower head 13 to a reference height (such as 2.0 m, as the starting point of subsequent height adjustment); the rain shower head 13 is reset to a default diameter (such as 1.5 mm, corresponding to medium-sized raindrops); the motor 15 stops rotating, and the rain shower head 13 is in a static state;

[0074] Then, the electronic device 6 specifies the rainwater with the target pH value (such as selecting the rain barrel 11 with pH=5.0) through pipeline valve control (which is implied in the connection logic of the rain barrel 11 and the regulating pump 12), so as to ensure that the water flow entering the shower head meets the pH requirement.

[0075] The electronic device 6 sends a rotating speed instruction to the regulating pump 12 according to the "rainfall intensity-rotating speed corresponding relationship" (such as 20 mm / h rainfall intensity corresponding to regulating pump 12 3000 r / min). The regulating pump 12 controls the water extraction amount per unit time by changing the rotating speed: high rotating speed (such as 4000 r / min)→ large flow→ simulate heavy rainfall (such as 30 mm / h); low rotating speed (such as 1000 r / min)→ small flow→ simulate weak rainfall (such as 10 mm / h).

[0076] The electronic device 6 sends a diameter adjusting instruction to the rain shower head 13, and changes the water outlet diameter through the internal valve or aperture switching structure of the shower head: large diameter (such as 2.5 mm)→ large raindrops (diameter 2-3 mm); small diameter (such as 1.0 mm)→ small raindrops (diameter 0.5-1 mm). The corresponding relationship between the diameter and the raindrop diameter needs to be pre-calibrated (such as experimentally measured "2.0 mm diameter corresponding to raindrop diameter 2.2 mm±0.2 mm").

[0077] The electronic device 6 controls the first liftable support frame 14 to extend or retract according to the "raindrop falling height-end speed relationship" (the higher the height, the greater the end speed of the raindrop, and the stronger the impact kinetic energy). For example, if high-altitude rainfall (high kinetic energy) is simulated, the support frame is instructed to be raised to 2.5 m; if near-ground rainfall (low kinetic energy) is simulated, the support frame is instructed to be lowered to 1.5 m. The height adjustment accuracy is usually ±0.1 m, which ensures that the falling kinetic energy of the raindrop is controllable.

[0078] To simulate the randomness of raindrop distribution in natural rainfall, the electronic device 6 sends a rotation instruction to the motor 15: the motor 15 drives the rainfall nozzle 13 to rotate at a set speed (such as 30 r / min) to make the rainwater evenly cover the sputtering simulation device 2 below when it is sprayed from the nozzle (to avoid local rainfall concentration caused by fixed direction spraying); if "local heavy rainfall" is simulated, the motor 15 can be instructed to rotate intermittently (such as rotating for 10 seconds and stopping for 5 seconds).

[0079] Next, the electronic device 6 receives the running state feedback of each component (such as the actual speed of the adjusting pump 12 and the real-time height of the nozzle), and if a deviation (such as a speed fluctuation of ±5%) is found, a correction instruction is immediately sent: if the rainfall intensity is too low (the actual flow is less than the set value), the speed of the adjusting pump 12 is increased by 5%; if the height of the nozzle is deviated due to vibration, the first liftable support frame 14 is instructed to be fine-tuned to the set height.

[0080] After the experiment is completed, the electronic device 6 sends a stop instruction: the speed of the adjusting pump 12 gradually decreases to 0 (to avoid water pressure fluctuation in the pipeline caused by sudden pump stop), and the water supply to the nozzle is stopped. The motor 15 stops rotating, and the rainfall nozzle 13 returns to the initial position; the aperture of the rainfall nozzle 13 is reset to the default value (1.5 mm); the first liftable support frame 14 lowers the nozzle to the reference height (2.0 m);

[0081] The electronic device 6 records the complete parameters of this rainfall (pH value, intensity, raindrop diameter, etc.) for subsequent correlation analysis with sputtering data.

[0082] The soil splash erosion simulation analysis system provided by the embodiments of the present application can directly contain rainwater with different pH values, thereby providing a basis for simulating various rainfall such as acidic, neutral and alkaline rainfall. The regulating pump 12 can accurately simulate rainfall with different intensities by controlling the water flow through the rotating speed, and the regulating process is controlled by the electronic device 6, thereby ensuring the stability and repeatability of the rainfall intensity parameters. The rainfall nozzle 13 can be adjusted to different diameters, and the rainfall droplet diameter can be accurately adjusted by combining the control of the electronic device 6, thereby meeting the demand of simulating different sizes of rainfall droplets. These designs enable the simulator to comprehensively cover the core parameters of rainfall, thereby providing accurate “rainfall input” for subsequent soil splash erosion simulation. The motor 15 drives the rainfall nozzle 13 to rotate, so that the distribution of the rainfall droplets is more uniform and closer to the effect of the random distribution of the rainfall droplets in natural rainfall. The first liftable support frame 14 can adjust the height of the rainfall nozzle 13, thereby controlling the rainfall height and simulating the impact of the falling rainfall droplets on the soil at different heights. Through these designs, the simulator can reduce the deviation between the artificial simulated rainfall and the natural rainfall, so that the simulated scene is closer to the actual natural environment. The components are in communication connection with the electronic device 6, thereby realizing the automation and intelligentization of the parameter adjustment. The researchers can conveniently adjust the parameters such as the water flow, the rotating speed of the nozzle, the nozzle diameter and the support frame height through the electronic device 6, quickly switch different rainfall scenes, do not need to frequently manually operate, greatly improve the experimental efficiency, and can better meet the splash erosion simulation demand under different soil types and different climate conditions. Finally, the rainfall simulator 1 can stably output controllable rainfall parameters, and in combination with other devices in the system, a complete data chain of “rainfall parameter input-soil splash erosion response” can be formed. The simulated rainfall scene can accurately reflect the effect of different rainfall conditions on the soil, thereby providing reliable experimental data support for the research on the correlation mechanism of the rainfall droplet splash and the soil splash erosion and the establishment of a splash erosion prediction model.

[0083] In an optional embodiment of the present application, as shown in Figure 3 The wind simulator 5 includes a rotatable base 51, a second liftable support frame 52, a mixed-flow fan set 53, a honeycomb-shaped flow guide device 54 installed at an air outlet of the mixed-flow fan set 53, and the rotatable base 51, the second liftable support frame 52 and the mixed-flow fan set 53 are in communication connection with the electronic device 6, wherein:

[0084] The rotatable base 51 is used for adjusting the wind direction by rotating the direction under the control of the electronic device 6;

[0085] The second liftable support frame 52 is used for adjusting the height under the control of the electronic device 6, so as to match the height of the rainfall nozzle 13 in the rainfall simulator 1;

[0086] The mixed-flow fan set 53 is used for adjusting the rotating speed frequency under the control of the electronic device 6, so as to adjust the wind speed;

[0087] The honeycomb-shaped flow guide device 54 is used to effectively reduce the turbulence phenomenon generated when the fan blows out.

[0088] Specifically, the rotatable base 51 can realize accurate control of wind direction. The rotatable base 51 is internally provided with a stepping motor 15 (controlling rotation) and an angle encoder (accuracy ±0.5°), and the base carries the entire main body of the wind simulator 5 and can rotate 360° without dead angle. Under the control of the electronic device 6, the orientation of the fan outlet is adjusted to simulate different wind directions (such as direct wind, direct counter-wind, 45° crosswind, etc.). For example, if it is necessary to simulate “raindrops impacting the soil at an angle”, the electronic device 6 sends a “30°” instruction, and the base rotates to 30°, so that the wind force acts on the raindrops in the 30° direction.

[0089] The second liftable support frame 52 can realize wind height adaptation. The second liftable support frame 52 includes an electric telescopic rod (stroke 0.5-3m) and is matched with a displacement sensor (accuracy ±0.01m) to support the mixed-flow fan set 53 and adjust the height thereof. According to the instruction of the electronic device 6, the telescopic rod is extended or retracted to make the height of the fan outlet consistent with the height of the rain nozzle (for example, when the height of the rain nozzle 13 is 2.2m, the support frame is synchronously adjusted to 2.2m).

[0090] The mixed-flow fan set 53 can realize continuous adjustment of wind speed. The mixed-flow fan set 53 includes a variable frequency motor 15 (controlling rotating speed), an impeller (generating airflow) and a wind speed sensor (range 0-12m / s, accuracy ±0.2m / s) to support continuous adjustment of wind speed of 0.5-10m / s. The rotating speed frequency is controlled by the electronic device 6 (for example, 20Hz corresponds to 2m / s, and 50Hz corresponds to 8m / s) to accurately output the target wind speed. For example, when simulating a light wind environment, 15Hz (1.5m / s) is set, and when simulating a strong wind environment, 60Hz (10m / s) is set. The mixed-flow fan has the characteristics of large flow of an axial-flow fan and high wind pressure of a centrifugal fan, which can not only output stable wind speed, but also ensure that the wind force covers the surface of the soil of the splash erosion simulation device 2 (covering diameter ≥1m).

[0091] The honeycomb-shaped flow guide device 54 can realize stable output of airflow. The honeycomb-shaped flow guide device 54 is composed of a honeycomb-shaped grid (made of ABS plastic, single honeycomb aperture 8mm, length 5cm) arranged in a regular hexagonal array and installed at the fan outlet. When the fan directly blows out, there is turbulence (turbulent airflow, large wind speed fluctuation), and the honeycomb grid can “comb” the turbulent airflow into laminar flow (parallel airflow, uniform wind speed), so that the turbulence degree of the outlet is reduced from 30% to below 5%. Turbulence will cause uneven force on the raindrops (some raindrops deviate too much, and some raindrops do not deviate), and laminar flow can ensure that the raindrops at the same height are subjected to consistent wind force, so that the subsequent “raindrop kinetic energy and trajectory” measurement data is more reliable (avoiding errors caused by turbulence).

[0092] After the electronic device 6 receives the parameters of the rainfall nozzle 13 height, rainfall intensity, etc., it automatically adjusts the height of the second liftable support frame 52 and the fan speed (such as matching a higher wind speed in heavy rainfall to simulate severe weather). After the wind acts on the raindrops, the raindrop characteristic measuring device 3 will capture the attitude changes (such as the inclination angle) of the raindrops due to the wind, and the data will be fed back to the electronic device 6 for verifying whether the correlation between “wind force-raindrop trajectory” meets the expectation (if the deviation is too large, the electronic device 6 will fine-tune the fan parameters).

[0093] The soil splash erosion simulation analysis system provided by the embodiments of the present application can accurately control the wind speed output through the adjustment of the rotating speed frequency of the mixed-flow fan set 53, and meet the simulation needs of different wind speed scenes. The rotatable base 51 can flexibly adjust the wind direction under the control of the electronic device 6, and in combination with the wind speed adjustment, different wind force conditions can be reproduced. This design breaks through the limitation of single parameter in traditional wind simulation, can accurately simulate the complex wind state in the natural environment, and provides reliable wind force parameter input for the research on the influence of wind force on raindrop movement and soil splash erosion. The second liftable support frame 52 can adjust the height to match the height of the rainfall nozzle 13, ensure that the wind force action range and the rainfall range are effectively overlapped, and avoid the problem that the wind force is not sufficient for the raindrops due to the height difference; the honeycomb-shaped flow guide device 54 can reduce the turbulence phenomenon, make the output wind force more stable, ensure the uniformity of the wind force acting on the raindrops, and thus improve the authenticity of the “rainfall-wind force” coordinated simulation, and more closely match the actual scene in which wind and rain jointly act on the soil in the natural environment. The rotatable base 51, the second liftable support frame 52 and the mixed-flow fan set 53 are in communication connection with the electronic device 6, and the automation of the wind direction, height and wind speed adjustment is realized. The researchers can quickly set the wind force parameters through the electronic device 6 without manual adjustment, which greatly shortens the switching time of different wind force scenes, improves the experimental efficiency, and is especially suitable for multiple comparison experiments which need to frequently change the wind force conditions. The stable and controllable wind force simulation combined with the rainfall simulation can accurately capture the influence of the wind force on the kinetic energy of the raindrops, the movement trajectory of the raindrops and the splash migration of the soil particles, provide high-quality experimental data for revealing the action mechanism of the wind force in the soil splash erosion process, and help deeper soil erosion research.

[0094] In an optional embodiment of the present application, as shown in Figure 4 The splash erosion simulation device 2 includes a splash erosion disc 21, a collection drawer 22 and a waterproof cover 23. The splash erosion disc 21 is composed of a central source area 211 and a collection area 212. The central source area 211 is used to place a soil disc 213 with a drainage hole at the bottom. The soil disc 213 is used to hold the target soil. The collection area 212 is composed of multiple concentric circles. The circles are separated by steel plates 214. The ring area of each circle includes two symmetric small holes for collecting splash erosion materials. The small hole positions of adjacent ring areas are staggered, and the ring areas are kept sealed and waterproof.

[0095] A collection drawer 22 is installed below the sputtering disc 21 for placing the sputtering material sample collection basin, improving the collection efficiency of the sputtering material;

[0096] A waterproof cover 23 is installed on the periphery of the sputtering disc 21 to ensure that raindrops only act on the soil disc 213.

[0097] Specifically, the sputtering disc 21 is the main body of the device, which is divided into a central source area 211 and a collection area 212, and is made of stainless steel (corrosion resistant and easy to clean): the central source area 211 (diameter 20 cm): places the soil disc 213 (diameter 15 cm, height 5 cm), which is the "origin" where raindrops directly splash soil. The bottom of the soil disc 213 is provided with 3-5 drainage holes with a diameter of 2 mm, which allows excess water to drain after rainfall (to avoid water accumulation in the soil affecting the sputtering process), but the hole diameter is smaller than the soil particles (to prevent soil from leaking out). The target soil is loaded into the soil disc 213 (compacted to a bulk density of 1.3 g / cm³ to simulate the natural soil state), and raindrops fall from above and directly splash the soil surface, producing sputtering particles (soil clumps or single particles) that fly in all directions.

[0098] The collection area 212 (surrounding the central source area 211, width 80 cm) is composed of 5 concentric circles (from inside to outside, 5 cm, 10 cm, 20 cm, 30 cm, 40 cm ring area, the distance from the edge of the central source area 211 is the particle migration distance), and the circles are separated by 2 mm thick steel plates 214 (height 3 cm, to prevent mixing of particles in adjacent ring areas). The bottom of each circular ring is provided with 2 symmetrical small holes (diameter 1 cm) for guiding the sputtering material to fall into the collection drawer 22 below. The positions of the small holes of adjacent circular rings are staggered (for example, the small holes of the 5 cm ring area are at 0° and 180°, and the small holes of the 10 cm ring area are at 90° and 270°), to avoid particles in the outer ring area being "incorrectly collected" by the inner small holes. The gaps between the ring areas are sealed with silica gel to prevent rainwater or sputtering material from seeping into the gaps (to ensure that the collection amount of each ring area corresponds to its migration distance).

[0099] The collection drawer 22 is an auxiliary structure for improving the collection efficiency of the sputtering material. It is horizontally embedded directly below the sputtering disc 21 (vertically corresponding to the small holes of the circular rings) and can be pulled out (similar to a drawer design). Inside the drawer, "sample collection basins" are placed, which match the number of circular rings (each basin corresponds to a small hole in a ring area), and the sputtering material falling from the small holes of the circular rings falls directly into the corresponding basins. This avoids the material residue caused by traditional "direct collection at the bottom of the ring area" (the drawer can be completely pulled out, and the collection basins are easy to pour); the collection basins are labeled with ring area numbers (such as "5 cm" "10 cm"), which can directly correspond to the migration distance when weighing later, reducing data recording errors.

[0100] The waterproof cover 23 is a protective structure designed to eliminate secondary splash interference. Made of transparent acrylic (without obstructing observation), it is cylindrical and covers the periphery of the splash plate 21 (1.2m high and 1.2m in diameter). The bottom is sealed to the edge of the splash plate 21. This ensures that raindrops falling from the rainfall simulator 1 fall vertically only onto the soil plate 213 in the central source area 211 (preventing raindrops from splashing onto the collection area 212 or outside the device). If splash particles splash onto the inner wall of the waterproof cover 23, they slide along the wall to the corresponding annular area (rather than being dispersed by raindrops again), ensuring that the collected material is a product of "primary splash" (truly reflecting the initial splash state of the soil).

[0101] Raindrops impact the target soil surface, shattering sand particles and scattering them in all directions—small particles have a long range (falling into a 40cm ring), while large particles have a short range (falling into a 5cm ring). The splashed material from each ring is collected along the bottom slope to the small holes and falls into the corresponding collection basin in the collection drawer 22 below. A waterproof cover 23 prevents raindrops from escaping and protects the splashed particles from being disturbed by external airflow.

[0102] After the experiment, the collection drawer 22 was pulled out, and the materials in each basin were dried (105°C, 24 hours). After weighing, the “splashing amount at different migration distances” was obtained (e.g., 15 g for a 5 cm ring area and 3 g for a 40 cm ring area).

[0103] In the soil splash erosion simulation and analysis system provided in the embodiment of the present application, a central source area 211 is specifically used to place a soil pan 213 containing target soil, and a drainage hole is provided at the bottom of the soil pan 213. This ensures that raindrops directly splash the soil while also allowing excess water to be drained away in a timely manner. This design makes the initial conditions for splash erosion closer to nature, reducing the deviation between the experimental scenario and the actual environment. The collection area 212 is composed of multiple concentric rings with fixed spacing between the rings. The particle migration distance can be directly determined by the "ring area where the splash material is located" - small particles (easy to splash) generally fall into the far ring area, while large particles (resistant to breakage) fall into the near ring area. This hierarchical collection method can accurately capture the relationship between "soil particle size and migration distance", providing direct data support for analyzing soil erosion resistance (for example, soil with strong aggregate stability has a high proportion of large particles and a high amount of splash erosion in the near ring area). The circular rings are separated by steel plates 214 and sealed to prevent splashing materials from one ring zone from mixing into other ring zones due to water flow or vibration. The positions of the small holes in adjacent ring zones are staggered (for example, the small holes in the 5cm ring zone are at 0° and 180°, and those in the 10cm ring zone are at 90° and 270°) to prevent splashing materials from "crossing ring zones" and falling into the wrong collection holes. These two designs physically ensure that "the materials collected in each ring zone only come from splashing within the range of that ring zone," making the data reliable and reducing misjudgment of results due to collection errors. The collection drawer 22 is installed below the splash tray 21, and can directly hold the sample collection basin of the corresponding ring zone. The splashing materials naturally fall into the basin through the small holes in the ring zone, eliminating the need for manual collection. After the experiment, the drawer can be pulled out as a whole, and the samples can be quickly transferred to the drying and weighing stages, greatly improving operational efficiency. It is particularly suitable for multiple comparative experiments. The core function of the waterproof cover 23 is to block "secondary interference." Firstly, it confines raindrops to the soil pan 213 in the central source area 211, preventing them from splashing into the collection area 212 or outside the device. Secondly, it prevents splashed material already in the collection area 212 from being re-splashed by subsequent raindrops (i.e., "secondary splash"). Without the waterproof cover 23, material in the collection area 212 might be dispersed by raindrops and re-splashed, distorting the collected volume. The waterproof cover 23 ensures that the collected material is a direct product of "primary splash," accurately reflecting the soil's splash state under the initial impact. Data collected by the splash simulation device 2, such as "splash volume per annular zone," "total splash volume," and "splash material particle size (subsequently screened to obtain the average weight diameter of the splash material)," can be directly correlated with parameters of the rainfall simulator 1 (raindrop diameter and kinetic energy) and the wind simulator 5 (wind speed and direction), providing the foundational data for the electronic device 6 to establish an "external driver-splash erosion result" model.

[0104] In an alternative embodiment of the present application, the "erosion prediction model between raindrop data and soil erosion data is established based on raindrop pH value, raindrop data, wind speed, wind direction, initial pH value, initial conductivity, target pH value, target conductivity, target erosion depth, and target sputtering material average weight diameter and target total sputtering amount", as shown in Figure 5 may include the following steps:

[0105] Step S101, based on the initial conductivity and the target conductivity, calculate the conductivity difference; based on the initial pH value and the target pH value, calculate the pH value difference.

[0106] Specifically, the electronic device 6 can calculate the conductivity difference by subtracting the initial conductivity from the target conductivity. Then, calculate the pH value difference by subtracting the initial pH value from the target pH value.

[0107] Step S102, input the raindrop pH value, raindrop data, wind speed and wind direction into the first branch of the initial erosion prediction network, and perform feature extraction on the raindrop pH value and raindrop data to obtain external driving features.

[0108] Specifically, the above step S102 can include the following steps:

[0109] Step a1, based on the wind speed, correct the raindrop diameter to obtain a diameter correction coefficient.

[0110] Specifically, the electronic device 6 can compare the wind speed with the preset wind speed threshold, and if the wind speed is greater than the preset wind speed threshold, multiply the preset diameter coefficient by the size of the raindrop diameter to obtain the diameter correction coefficient.

[0111] For example, when the wind speed is greater than 3 m / s, the large-diameter raindrops (greater than 3 mm) are affected by wind shear force, and the flat deformation is intensified, so the preset diameter coefficient of the large-diameter raindrops is increased to 1.1; The small-diameter raindrops (less than 1 mm) are easily deviated from the trajectory under the influence of the wind, and the actual impact area is reduced by 5% compared with no wind, so the preset diameter coefficient is 0.95. After correction, the error between the impact range characteristics under different wind speeds and the actual measured erosion distribution radius is reduced to within 5%.

[0112] Step a2, calculate the raindrop falling time based on the raindrop terminal velocity.

[0113] Specifically, the electronic device 6 can calculate the raindrop falling time based on the raindrop terminal velocity.

[0114] Step a3, based on the soil type corresponding to the target soil, correct the raindrop falling time to obtain a time correction coefficient.

[0115] Specifically, the electronic device 6 can determine a preset time coefficient according to the soil type corresponding to the target soil and the raindrop falling time, and then multiply the raindrop falling time by the preset time coefficient to obtain the time correction coefficient.

[0116] For example, sandy soil: if the raindrop falling time > 2s, the preset time coefficient is 0.95; if the raindrop falling time < 0.5s, the preset time coefficient is 1.15 (sandy soil particles are loose, and even if the falling time is short, they are easy to spread); clay soil: if the raindrop falling time > 2s, the preset time coefficient is 0.85; if the raindrop falling time < 0.5s, the preset time coefficient is 1.05 (clay soil particles are cohesive, and the range is more obvious when the falling time is long).

[0117] Step a4, correct the raindrop terminal velocity based on the wind speed to obtain a raindrop speed correction coefficient.

[0118] Specifically, the electronic device 6 corrects the raindrop terminal velocity based on the following formula to obtain the raindrop speed correction coefficient.

[0119] Raindrop speed correction coefficient = raindrop terminal velocity x (1 + 0.02 x min(wind speed, 5)) (Note: the unit of wind speed is m / s, and when the wind speed > 5m / s, it is calculated as 5m / s to avoid excessive amplification of extreme wind speed).

[0120] It should be noted that for every 1m / s increase in wind speed, the effective impact speed of the raindrop on the soil is increased by about 2% (for example, a medium-speed raindrop at 3m / s wind speed, the actual speed coefficient = 1.2 x (1 + 0.02 x 3) = 1.272). This correction makes the speed coefficient no longer fixed, but dynamically adapts to the wind conditions.

[0121] Step a5, calculate the uniformity coefficient based on the wind speed and the raindrop attitude.

[0122] Wherein, the raindrop attitude is the angle α between the raindrop motion direction and the vertical direction of gravity. The electronic device 6 can calculate the uniformity coefficient based on the following formula.

[0123] Uniformity coefficient = 1.0 - 0.01 x wind speed - 0.02 x α. Wherein, the unit of wind speed is m / s, the unit of α is °, and the result of uniformity coefficient is controlled between 0.6-1.0.

[0124] For example, wind speed 3m / s, α = 10° → uniformity coefficient = 1.0 - 0.01 x 3 - 0.02 x 10 = 0.77 (between dispersion and concentration); wind speed 5m / s, α = 25° → coefficient = 1.0 - 0.01 x 5 - 0.02 x 25 = 0.45 (forced to 0.6 to avoid being too low).

[0125] Step a6, the basic spatial range feature is calculated based on the diameter correction coefficient, the time correction coefficient and the raindrop speed correction coefficient.

[0126] Specifically, the electronic device 6 multiplies the diameter correction coefficient, the time correction coefficient and the raindrop speed correction coefficient to obtain the basic spatial range feature.

[0127] Step a7, the impact range feature is calculated based on the basic spatial range feature and the uniformity coefficient.

[0128] Specifically, the electronic device 6 can calculate the impact range feature by multiplying the basic spatial range feature by the uniformity coefficient.

[0129] Step a8, the raindrop kinetic energy density is obtained based on the raindrop kinetic energy divided by the diameter correction coefficient.

[0130] Specifically, the electronic device 6 obtains the raindrop kinetic energy density based on the raindrop kinetic energy divided by the diameter correction coefficient.

[0131] Step a9, the kinetic energy correction coefficient is obtained by correcting the raindrop kinetic energy based on the wind direction, the wind speed and the raindrop running posture.

[0132] Specifically, the electronic device 6 can calculate the angle between the wind direction and the raindrop running direction based on the wind direction and the raindrop running posture. Then, the angle between the wind direction and the raindrop running direction is compared with the first preset angle threshold and the second preset angle threshold. If the angle between the wind direction and the raindrop running direction is less than the first preset angle threshold, the raindrop terminal speed is multiplied by the first preset speed coefficient, so as to correct the raindrop terminal speed and obtain the raindrop speed correction coefficient. Then, the raindrop kinetic energy is corrected based on the raindrop speed correction coefficient to obtain the initial kinetic energy coefficient.

[0133] If the angle between the wind direction and the raindrop running direction is greater than the second preset angle threshold, the raindrop terminal speed is multiplied by the second preset speed coefficient, so as to correct the raindrop terminal speed and obtain the raindrop speed correction coefficient. Then, the raindrop kinetic energy is corrected based on the raindrop speed correction coefficient to obtain the initial kinetic energy coefficient.

[0134] For example, when the angle between the wind direction and the raindrop running direction is < 30°, the raindrop is boosted by the wind force, the raindrop terminal speed is increased by 5% (raindrop speed correction coefficient = raindrop terminal speed x 1.05), and then the raindrop kinetic energy is corrected based on the raindrop speed correction coefficient to obtain the initial kinetic energy correction coefficient. If the angle between the wind direction and the raindrop running direction is > 60°, the raindrop terminal speed is reduced by 8% (raindrop speed correction coefficient = raindrop terminal speed x 0.92), and then the raindrop kinetic energy is corrected based on the raindrop speed correction coefficient to obtain the initial kinetic energy coefficient.

[0135] Then, the electronic device 6 calculates the kinetic energy correction coefficient based on the initial kinetic energy coefficient. The specific formula is as follows:

[0136] Kinetic energy correction coefficient = initial kinetic energy coefficient x (1 + 0.03 x min (wind speed, 6)). Wherein, the unit of wind speed is m / s, and when the wind speed is greater than 6 m / s, it is calculated according to 6 m / s to avoid extreme value interference.

[0137] Step a10, according to the soil type corresponding to the target soil and the pH value of the raindrop, the synergistic factor is calculated;

[0138] Specifically, the electronic device 6 can determine the soil buffer coefficient according to the soil type corresponding to the target soil. According to the corresponding relationship between the pH value of the raindrop and the pH value enhancement factor, the pH value enhancement factor is determined. Then, the synergistic factor is calculated by multiplying the pH value enhancement factor by the soil buffer coefficient.

[0139] Wherein, when the target soil is sandy soil, the soil buffer coefficient is 1.2, when the target soil is loamy soil, the soil buffer coefficient is 1.0; and when the target soil is clayey soil, the soil buffer coefficient is 0.8.

[0140] Example: acidic raindrop (enhancement factor 1.3) with pH = 4.0 acting on sandy soil (buffer coefficient 1.2), synergistic factor = 1.3 x 1.2 = 1.5.

[0141] Step a11, based on the raindrop running posture and the angle between the wind direction and the impact direction, the direction intensity coefficient is calculated.

[0142] Specifically, wherein the raindrop running posture is used to represent the angle α between the raindrop motion direction and the vertical direction of gravity. The electronic device 6 can calculate the vertical direction intensity coefficient according to the raindrop running posture. The calculation formula is as follows: vertical direction intensity coefficient = cos (α) x (1-0.005 x α); wherein, α is the raindrop running posture, unit °, 0°≤α≤60°, > 60° is forced to take 0.5, to avoid excessive reduction.

[0143] The electronic device 6 can calculate the angle between the wind direction and the impact direction according to the raindrop running posture and the wind direction. Wherein, the impact direction refers to the impact direction of raindrops (or other impact objects) on the soil in the splash erosion scene (if it is natural rainfall, it is usually approximately vertical downward; if it is a simulation experiment, there may be inclination due to device angle adjustment).

[0144] Then, the electronic device 6 calculates the horizontal direction intensity coefficient based on the wind direction and the impact direction angle. The calculation formula is as follows: horizontal direction intensity coefficient = 1.0 + 0.003 x θ (θ is the wind direction and the impact direction angle, 0°≤θ≤90°). When θ = 0° (downwind): horizontal direction intensity coefficient = 1.1 (wind power helps the particles to migrate, and the horizontal intensity increases by 10%);

[0145] When θ = 90° (crosswind): horizontal direction intensity coefficient = 1.27 (crosswind makes the particle diffusion range expand, and indirectly enhances the horizontal direction damage).

[0146] Then, the electronic device 6 adds the vertical direction intensity coefficient and the horizontal direction intensity coefficient after multiplying the corresponding weight coefficients, to obtain the direction intensity coefficient.

[0147] Step a12, the energy intensity is calculated based on the kinetic energy correction coefficient, the raindrop kinetic energy density and the synergy factor.

[0148] Specifically, the electronic device 6 multiplies the kinetic energy correction coefficient, the raindrop kinetic energy density and the synergy factor to obtain the energy intensity.

[0149] Step a13, the impact intensity feature is calculated based on the energy intensity and the direction intensity coefficient.

[0150] Specifically, the electronic device 6 multiplies the energy intensity and the direction intensity coefficient to obtain the impact intensity feature.

[0151] Step a14, the external driving feature is generated by fusing the impact range feature and the impact intensity feature.

[0152] Specifically, the electronic device 6 can compare the impact intensity feature with the preset impact intensity threshold value, and if the impact intensity feature is greater than the preset impact intensity threshold value, it is determined that the weight corresponding to the impact intensity feature is greater than the weight corresponding to the impact range feature. Then, according to the weights corresponding to the impact range feature and the impact intensity feature respectively, the impact range feature and the impact intensity feature are fused to generate the external driving feature.

[0153] Step S103, input the conductivity difference value and the pH value difference value into the second branch of the initial sputtering prediction network, and perform feature extraction on the conductivity difference value and the pH value difference value to obtain the soil response feature.

[0154] Specifically, the above step S103 can include the following steps:

[0155] Step b1, based on the relationship between the conductivity difference value and the pH value difference value, the conductivity difference value and the pH value difference value are corrected to obtain a corrected conductivity coefficient and a corrected pH value coefficient.

[0156] Specifically, the electronic device 6 can correct the conductivity difference value and the pH value difference value based on the positive and negative relationship between the conductivity difference value and the pH value difference value to obtain a corrected conductivity coefficient and a corrected pH value coefficient.

[0157] For example, if the pH value difference value is negative (soil acidification) and the conductivity difference value is positive (salt loss increase), it indicates that the acidic condition aggravates the soil structure damage, at this time, the conductivity difference value is multiplied by the conductivity coefficient to obtain the corrected conductivity coefficient (amplifying the influence of salt loss), and the pH value difference value is multiplied by the pH value coefficient to obtain the corrected pH value coefficient (strengthening the associated effect of acidification). The conductivity coefficient can be 1.1, 1.2 or other values. The pH value coefficient can be 1.05, 1.06 or other values. The embodiments of the present application do not make specific limitations on the conductivity coefficient and the pH value coefficient.

[0158] If the change direction of the two is not related (for example, the pH value is unchanged but the conductivity increases), it is determined that it is not a splash erosion factor (for example, natural soil salt migration), and the correction coefficient is multiplied by 0.8 (weakening the interference).

[0159] Effect: Eliminate irrelevant interference, so that the corrected parameters more accurately reflect the soil physicochemical changes caused by splash erosion, and avoid single parameter misjudgment.

[0160] Step b2, calculating the soil buffer index based on the first relationship between the corrected conductivity coefficient and the corrected pH value coefficient.

[0161] Specifically, the electronic device 6 can calculate the soil buffer index based on the synergistic relationship (the first relationship) between the corrected conductivity coefficient and the corrected pH value coefficient:

[0162] For example, the soil buffer index = 1 / (corrected pH value coefficient + corrected conductivity coefficient / 2), and the larger the value, the stronger the soil's ability to resist changes in acid, alkali and salt (for example, when the buffer index is greater than 2, it indicates that the soil can effectively maintain the physicochemical balance).

[0163] Thus, the two dispersed physicochemical parameters are integrated into a single index, which directly reflects the soil's ability to withstand physicochemical disturbance caused by splash erosion.

[0164] Step b3, calculating the structure integrity feature based on the second relationship between the corrected conductivity coefficient and the corrected pH value coefficient.

[0165] Specifically, the electronic device 6 can calculate the structure integrity feature based on the reverse correlation (the second relationship) between the corrected conductivity coefficient and the corrected pH value coefficient:

[0166] For example, the structure integrity feature = 1-(corrected pH value coefficient x 0.6 + corrected conductivity coefficient x 0.4) x soil texture coefficient.

[0167] wherein, for sandy soil, the soil texture coefficient is 1.2; for clayey soil, the soil texture coefficient is 0.7, and the greater the value, the more complete the soil structure (e.g., when the structural integrity > 0.7, the degree of aggregation breakage is low).

[0168] Step b4, fusing the soil buffer index and the structural integrity feature to generate a soil response feature.

[0169] Specifically, the electronic device 6 can fuse the soil buffer index (reflecting the physicochemical stability) with the structural integrity feature (reflecting the structure's resistance to damage).

[0170] Fusion logic: dynamically assign weights according to the buffer index - when the buffer index > 2 (strong buffer), the physicochemical stability weight is 60%; when the buffer index < 1 (weak buffer), the structural integrity weight is 70% (structure damage dominates the resistance).

[0171] Thus, the output soil response feature not only covers the anti-interference ability on the physicochemical level, but also includes the anti-breakage ability on the structure level, and quantifies the internal resistance potential of the soil to splash erosion, providing reliable "soil internal properties" input for subsequent fusion with external driving features.

[0172] Step S104, fusing the external driving features and the soil response features to obtain target features.

[0173] Specifically, the external driving features include impact range features and impact intensity features; the soil response features include soil buffer index and structural integrity features; and the above step S104 can include the following steps:

[0174] Step c1, assigning initial weights to the external driving features and the soil response features, respectively.

[0175] Specifically, the electronic device 6 can assign initial weights to the external driving features and the soil response features, respectively.

[0176] For example, the initial weights of the external driving features and the soil response features are assigned (e.g., each accounting for 50%), ensuring that both are considered in the initial fusion stage:

[0177] This allocation avoids the dominance of a single feature in the fusion result, lays the foundation for subsequent weight adjustment based on data correlation, and makes the fusion process start from an equal starting point, taking into account the dual attributes of "external impact" and "soil resistance".

[0178] Step c2, constructing a correlation coefficient matrix of the external driving features and the soil response features.

[0179] Specifically, the electronic device 6 can determine the correlation strength of the two by calculating the correlation coefficient of the external driving features (such as impact range, impact strength) and the soil response features (such as soil buffer index, structural integrity), and construct a correlation coefficient matrix of the external driving features and the soil response features according to the calculated correlation coefficient.

[0180] For example: the impact strength feature and the structural integrity feature may have a significant negative correlation (correlation coefficient -0.8), indicating that the stronger the impact, the easier the soil structure is damaged; the impact range feature and the soil buffer index may have a negative correlation (correlation coefficient -0.7), indicating that the wider the impact range, the easier the soil physicochemical balance is broken.

[0181] Effect: The internal relationship between the features is intuitively presented by the matrix, providing data basis for subsequent focusing on key associations and avoiding invalid fusion of unrelated features.

[0182] Step c3, based on the correlation coefficient matrix, determine the target association features with a correlation coefficient greater than a preset correlation threshold.

[0183] Specifically, the electronic device 6 can select target association feature pairs (such as "impact strength-structural integrity" and "impact range-soil buffer index") with a correlation coefficient greater than a preset threshold (such as 0.7) from the correlation coefficient matrix. Only high correlation feature pairs are retained for subsequent operations, and weakly correlated or unrelated features (such as impact range and structural integrity with a correlation coefficient <0.3) are excluded to reduce redundant information interference. The fusion process focuses on the association relationship that has a core impact on the sputtering result, improving the relevance and efficiency of feature fusion.

[0184] Step c4, cross operation on the target association features to generate interaction features;

[0185] Specifically, the target association features are cross-operated to generate interaction features.

[0186] For example, the interaction feature is calculated as follows:

[0187] Structural excess impact = impact strength feature x (1-structural integrity feature); Physical meaning: when the structural integrity feature = 0.5, if the impact strength feature = 0.8, then the excess impact = 0.8 x (1-0.5) = 0.4 - indicating that 40% of the impact strength exceeds the soil structure bearing capacity.

[0188] Physicochemical excess impact = impact range feature x (1-soil buffer index); Physical meaning: when the soil buffer index = 0.6, if the impact range feature = 0.7, then the excess impact = 0.7 x (1-0.6) = 0.28 - indicating that 28% of the impact range exceeds the soil physicochemical buffer capacity, corresponding to "local soil appears acid-base imbalance".

[0189] The interaction feature directly quantifies the "potential driving force of splash erosion", and the higher the value (e.g. > 0.5), the more the impact on the soil has entered the "irreversible" stage of damage.

[0190] Step c5, based on the interaction feature, the initial weight is corrected to obtain the target weight corresponding to the external driving feature and the soil response feature respectively.

[0191] Specifically, the electronic device 6 can correct the initial weight based on the interaction feature to obtain the target weight corresponding to the external driving feature and the soil response feature respectively.

[0192] For example, if the interaction feature value is > 0.5 (the impact significantly exceeds the bearing capacity of the soil), the weight of the external driving feature is increased (e.g. from 50% to 60%); if the interaction feature value is < 0.3 (the impact is effectively resisted), the weight of the soil response feature is increased (e.g. from 50% to 60%). The weight distribution is no longer fixed, but dynamically adapts to the actual strength of "drive-response", ensuring that the fusion result fits the real mechanism of splash erosion.

[0193] Step c6, based on the target weight, the external driving feature and the soil response feature are fused to generate a target feature.

[0194] Specifically, the electronic device 6 fuses the external driving feature and the soil response feature based on the target weight. The fusion result contains both the destructive power information of the external impact and the erosion resistance information of the soil itself, for example: high external driving + low soil response, the target feature value is high, corresponding to high splash erosion risk. Thus, the target feature completely depicts the comprehensive conditions of splash erosion, providing comprehensive and accurate input for subsequent splash erosion prediction, avoiding the one-sidedness of single-dimensional evaluation.

[0195] Step S105, the target feature is transmitted into the output layer to output the virtual erosion depth, the virtual splash erosion material average weight diameter and the virtual total splash erosion amount.

[0196] Specifically, the electronic device 6 can transmit the target feature into the output layer to output the virtual erosion depth, the virtual splash erosion material average weight diameter and the virtual total splash erosion amount.

[0197] Step S106, based on the relationship between the virtual erosion depth, the virtual splash erosion material average weight diameter and the virtual total splash erosion amount and the target erosion depth, the target splash erosion material average weight diameter and the target total splash erosion amount, the initial splash erosion prediction network is trained to obtain a splash erosion prediction model.

[0198] Specifically, the above step S106 can include the following steps:

[0199] Step d1, based on the virtual erosion depth and the target erosion depth, calculate the erosion relative error and the erosion absolute error; according to the soil type corresponding to the target soil, determine the first weight information corresponding to the erosion relative error and the erosion absolute error respectively; based on the first weight information, the erosion relative error and the erosion absolute error, calculate the erosion error;

[0200] Specifically, the electronic device 6 can calculate the erosion relative error and the erosion absolute error based on the virtual erosion depth and the target erosion depth, the erosion relative error = |virtual erosion depth-target erosion depth|; the erosion absolute error = |(virtual erosion depth-target erosion depth) / target erosion depth|.

[0201] Because there are differences in erosion sensitivity of different soil types (such as sandy soil is easy to erode, and clay soil is difficult to erode), therefore, the electronic device 6 can determine the first weight information corresponding to the erosion relative error and the erosion absolute error according to the soil type corresponding to the target soil.

[0202] For example, the soil type is sandy soil: when the target erosion depth > 1 cm, the first weight corresponding to the erosion absolute error is increased to 0.7 (because deep erosion has a greater impact on sandy soil productivity), and the first weight corresponding to the erosion relative error is 0.3; the soil type is clay soil: when the target erosion depth < 0.5 cm, the weight corresponding to the erosion relative error is increased to 0.5 (more sensitive to shallow erosion, which needs to be accurately captured), and the first weight corresponding to the erosion absolute error is 0.5.

[0203] Then, based on the first weight information, the erosion relative error and the erosion absolute error, the error is calculated. Thus, the erosion error calculation is adapted to the soil type, avoiding "one-size-fits-all" evaluation of different types of soil.

[0204] Example of the modified formula (sandy soil):

[0205] Erosion error = 0.7 x |virtual erosion depth-target erosion depth| + 0.3 x |(virtual erosion depth-target erosion depth) / target erosion depth| (when the target erosion depth > 1 cm).

[0206] Step d2, based on the virtual splash erosion material average weight diameter and the target splash erosion material average weight diameter, calculate the splash erosion material average weight diameter square error.

[0207] Specifically, the electronic device 6 can calculate the splash erosion material average weight diameter square error based on the virtual splash erosion material average weight diameter and the target splash erosion material average weight diameter.

[0208] Splash erosion material average weight diameter square error = (virtual splash erosion material average weight diameter-target splash erosion material average weight diameter)².

[0209] Step d3, calculating the proportion of first particles having a diameter greater than a preset diameter in the average weight diameter of the virtual sputtered material and the proportion of second particles having a diameter greater than a preset diameter in the average weight diameter of the target sputtered material.

[0210] Logically speaking, the average weight diameter of the spattered material is a key indicator for measuring the water stability of the agglomerates (the larger the average weight diameter of the spattered material, the stronger the water stability of the agglomerates), and the proportion of particles with a diameter greater than the preset diameter threshold is an important auxiliary feature reflecting the integrity of the agglomerates - a high proportion of large particles usually corresponds to a larger average weight diameter of the spattered material and more stable agglomerates.

[0211] Therefore, the electronic device 6 can calculate the proportion of first particles having a diameter greater than the preset diameter in the virtual average weight diameter of the sputtered material and the proportion of second particles having a diameter greater than the preset diameter in the target average weight diameter of the sputtered material.

[0212] Among them, the preset diameter can be 0.25mm or 0.26mm. The implementation of this application does not make any specific limitation on the preset diameter.

[0213] Step d4, calculating the average weight diameter error of the sputtered material according to the deviation between the first particle ratio and the second particle ratio and the square error of the average weight diameter of the sputtered material.

[0214] Specifically, the electronic device 6 calculates the deviation between the first particle ratio and the second particle ratio, and then calculates the average weight diameter error of the sputtered material based on the formula.

[0215] The average weight diameter error of the spattered material is [(the average weight diameter of the virtual spattered material - the average weight diameter of the target spattered material)² × 1.2] × [1 + a × I (the deviation of the particle ratio > 15%)]. Where I is the indicator function, which takes the value of 1 if the condition is met and 0 otherwise, and a is the first penalty coefficient.

[0216] For example, if the proportion of particles >0.25 mm calculated from the virtual average weight diameter of the spattered material deviates by more than 15% from the proportion of particles corresponding to the actual target average weight diameter of the spattered material, it means that although the virtual value may be close to the target in terms of diameter value, it does not truly reflect the actual state of the particle composition (for example, the virtual average weight diameter of the spattered material meets the standard but the proportion of large particles is insufficient, which may underestimate the degree of agglomerate breakup).

[0217] Therefore, on the basis of the original square error ((virtual sputtering material average weight diameter-target sputtering material average weight diameter) 2*1.2), an additional impact penalty (such as 0.2) is added, which is essentially to force the virtual value to match the target value of the particle distribution characteristics through quantitative means, avoid only pursuing the diameter value matching and ignoring the internal soil structure consistency, and make the error calculation more consistent with the actual mechanism of "sputtering material average weight diameter reflecting the stability of the aggregate". Thus, the sputtering material average weight diameter error is associated with the actual particle distribution, avoiding the one-sidedness of only focusing on the diameter and ignoring the particle composition.

[0218] Step d5, according to the virtual total sputtering amount and the target total sputtering amount, the total sputtering amount error is calculated.

[0219] Specifically, the electronic device 6 calculates the absolute value of the difference between the virtual total sputtering amount and the target total sputtering amount as the basic error.

[0220] Then, based on the virtual total sputtering amount, the virtual proportion of the virtual ring zone sputtering amount corresponding to each ring zone in the collection area 212 to the virtual total sputtering amount is calculated; and then based on the target total sputtering amount, the target proportion of the target ring zone sputtering amount corresponding to each ring zone in the collection area 212 to the target total sputtering amount is calculated.

[0221] For each ring zone, the ring zone deviation between the virtual proportion and the target proportion corresponding to each ring zone is calculated, and if the ring zone deviation corresponding to at least one ring zone is greater than a preset ring zone deviation threshold, a second penalty coefficient is determined.

[0222] Then, based on the second penalty coefficient and the basic error, the total sputtering amount error is calculated.

[0223] For example, the calculation formula is: total sputtering amount error= basic error* [1+b*I(ring zone proportion deviation>10%)].

[0224] Where I is an indicator function, taking 1 when the condition is met, otherwise taking 0, and b is the first penalty coefficient.

[0225] Step d6, based on the erosion error, the sputtering material average weight diameter square error and the total sputtering amount error, the target loss function is generated.

[0226] Specifically, the electronic device 6 can generate the target loss function based on the erosion error, the sputtering material average weight diameter square error and the total sputtering amount error. The formula is as follows:

[0227] Target loss function=(erosion error*c+sputtering material average weight diameter square error*d+total sputtering amount error*f)*risk coefficient*dynamic balance coefficient.

[0228] Wherein c, d, f are weight data, risk coefficient: 1.2 when target total sputtering loss is greater than 200g / m2; 1.0 when 100-200g / m2; 0.8 when less than 100g / m2; dynamic balance coefficient: after every 20 rounds of training, if the contribution of a certain error term is greater than 50%, the corresponding weight is reduced by 10% (such as the average weight diameter error weight of sputtering material from 0.4 to 0.36), and the balance coefficient is adjusted synchronously to maintain the total weight logic.

[0229] Step d7, training the initial sputtering prediction network based on the target loss function to obtain a sputtering prediction model.

[0230] Specifically, the electronic device 6 trains the initial sputtering prediction network based on the target loss function until the function value corresponding to the target loss function stabilizes at a preset function value or less, stops training, and obtains a sputtering prediction model.

[0231] The soil splash erosion simulation analysis system provided by the embodiments of the present application can directly quantify the changes of the soil physical and chemical properties before and after rainfall by calculating the conductivity difference and the pH value difference. The conductivity difference reflects the degree of soil salt loss or migration, and the pH value difference reflects the change of the soil acid-base balance. Both of them serve as the "physical and chemical signals" of the soil affected by splash erosion, and provide quantitative basis for subsequent evaluation of soil erosion resistance. The diameter correction coefficient is obtained by correcting the raindrop diameter based on the wind speed, so that the corrected diameter correction coefficient is more consistent with the actual impact area. The raindrop falling time is calculated based on the terminal velocity of the raindrop, and the raindrop falling time is corrected based on the soil type corresponding to the target soil, so that the time correction coefficient is obtained, so that the influence of the time factor on the impact range is matched with the actual characteristics of the soil, and the rationality of the basic spatial range is improved. The raindrop velocity correction coefficient is obtained by correcting the terminal velocity of the raindrop based on the wind speed. The uniformity coefficient is calculated based on the wind speed and the raindrop posture. The basic spatial range characteristics are calculated based on the diameter correction coefficient, the time correction coefficient and the raindrop velocity correction coefficient, so as to avoid the one-sidedness of single parameter dominance. The impact range characteristics are calculated based on the basic spatial range characteristics and the uniformity coefficient, which not only reflect the spatial breadth of raindrop coverage, but also reflect the distribution concentration through the uniformity, which is consistent with the actual "wind causes uneven distribution of raindrops" in natural rainfall. The raindrop kinetic energy density is obtained by dividing the raindrop kinetic energy by the diameter correction coefficient, which standardizes the energy of raindrops per unit volume and eliminates the apparent advantage of large-diameter raindrop kinetic energy. The kinetic energy correction coefficient is obtained by correcting the raindrop kinetic energy based on the wind direction, the wind speed and the raindrop running posture, so that the kinetic energy correction coefficient is closer to the actual impact energy. The synergy factor is calculated according to the soil type corresponding to the target soil and the raindrop pH value. The direction intensity coefficient is calculated based on the raindrop running posture and the angle between the wind direction and the impact direction, so as to accurately capture the effective action of energy in the vertical and horizontal directions and avoid ignoring the influence of direction on the damage degree. The energy intensity is calculated based on the kinetic energy correction coefficient, the raindrop kinetic energy density and the synergy factor. The impact intensity characteristics are calculated based on the energy intensity and the direction intensity coefficient. The external driving characteristics are generated by fusing the impact range characteristics and the impact intensity characteristics. The output external driving characteristics not only cover the spatial distribution, but also contain the energy intensity, which completely depict the external action of raindrops on the soil and provide comprehensive input for subsequent splash erosion prediction.

[0232] The second branch corrects the conductivity difference and the pH difference based on a relationship between the conductivity difference and the pH difference, to obtain a corrected conductivity coefficient and a corrected pH coefficient, to eliminate irrelevant interference, so that the corrected parameters more accurately reflect the physicochemical changes of the soil caused by sputtering and erosion, and to avoid misjudgment of a single parameter. Based on a first relationship between the corrected conductivity coefficient and the corrected pH coefficient, a soil buffer index is calculated; the two dispersed corrected conductivity coefficients and corrected pH coefficients are integrated into a single index to intuitively reflect the bearing capacity of the soil to the physicochemical disturbance caused by sputtering and erosion. Based on a second relationship between the corrected conductivity coefficient and the corrected pH coefficient, a structural integrity feature is calculated, the soil structure state is inferred through physicochemical parameters, and the complexity of direct measurement of the structure is solved. The soil buffer index and the structural integrity feature are fused to generate a soil response feature. The output soil response feature not only covers the anti-interference ability in the physicochemical aspect, but also includes the anti-crushing ability in the structure aspect, and quantifies the internal resistance potential of the soil to sputtering and erosion, providing reliable "soil internal attribute" input for subsequent fusion with external driving features.

[0233] Initial weights are respectively assigned to the external driving features and the soil response features to avoid the dominance of a single feature in the fusion result, to lay a foundation for subsequent weight adjustment based on data correlation, to start the fusion process from a balanced starting point, and to take into account the dual attributes of "external impact" and "soil resistance". A correlation coefficient matrix of the external driving features and the soil response features is constructed to intuitively present the internal relationship between the features through the correlation coefficient matrix, to provide data basis for subsequent focus on key correlations, and to avoid ineffective fusion of irrelevant features. Based on the correlation coefficient matrix, target correlation features with a correlation coefficient greater than a preset correlation threshold are determined, so that the fusion process focuses on the correlation relationship that has a core influence on the sputtering result, and the pertinence and efficiency of feature fusion are improved. The target correlation features are cross-operated to generate interaction features, to convert the abstract correlation relationship into quantifiable features, to accurately reflect the dynamic game between "external impact and soil resistance", and to provide a key basis for weight correction. Based on the interaction features, the initial weights are corrected to obtain target weights corresponding to the external driving features and the soil response features, so that the weight distribution is no longer fixed, but dynamically adapts to the actual strength contrast of "driving-response", and ensures that the fusion result fits the real mechanism of sputtering and erosion. Based on the target weights, the external driving features and the soil response features are fused to generate target features. The target features completely depict the comprehensive conditions of sputtering and erosion, provide comprehensive and accurate input for subsequent sputtering and erosion prediction, and avoid one-sidedness of single-dimensional evaluation. The output layer outputs a virtual erosion depth, a virtual sputtering material average weight diameter, and a virtual total sputtering amount, which correspond to the soil physical damage depth, the aggregate stability change, and the total material loss amount, respectively, covering the core dimensions of sputtering and erosion evaluation.

[0234] Then, for the "erosion depth", the absolute error and the relative error are combined, and different weights are given according to the soil type to avoid the limitations of a single error indicator, making the error calculation related to erosion more in line with the characteristics of the soil. For the "splashed material particles", not only the "average weight diameter square error" is used to quantify the overall size deviation of the particles, but also the "percentage deviation of particles with a diameter greater than a preset value" is used to focus on the splashing rules of coarse particles. The combination of these two indicators can more comprehensively reflect the particle distribution characteristics of the splashed material, avoiding the loss of details caused by focusing only on the average diameter. For the "total splashing amount", the total amount deviation is directly quantified to ensure the prediction accuracy of the model for the overall scale of the splashing. The coverage of multi-dimensional error allows the model to learn the core indicators such as erosion depth, particle characteristics, and total amount at the same time, making the prediction results closer to the real splashing process. The target loss function combines erosion error, particle diameter error, and total splashing amount error, rather than relying on a single indicator. This "multi-target fusion" design can avoid the model sacrificing other indicators to optimize a single indicator, ensuring the model's generalization ability in different scenarios - even if the input conditions change, the model can maintain stable prediction results in multiple key dimensions. The trained splashing prediction model can accurately output key parameters such as erosion depth, particle distribution, and total splashing amount, providing data support for soil erosion prevention - for example, by predicting the splashing risk of a specific soil under rainfall, measures such as vegetation cover and engineering protection can be taken in advance.

[0235] In an optional embodiment of the present application, as shown in Figure 6 The electronic device 6 is also configured to perform the following steps:

[0236] In step S201, an anti-erosion ability index is calculated based on the target erosion depth.

[0237] Specifically, the electronic device 6 can obtain the soil type corresponding to the target soil, and then correct the target erosion depth according to the soil type to obtain the anti-erosion ability index.

[0238] For example, the anti-erosion basic ability of different soil types differs, and a texture correction coefficient is introduced (0.8 for sandy soil, 1.0 for loamy soil, and 1.2 for clay soil). The corrected formula is: anti-erosion ability index = [1 / (1+target erosion depth)] x texture correction coefficient. The specific logic is: clay has strong cohesion, so it has higher actual erosion resistance at the same erosion depth; sandy soil is easy to erode, so the score needs to be reduced. For example, for clay with a target erosion depth of 0.3 cm, the anti-erosion ability index is [1 / (1+0.3)] x 1.2 ≈ 0.92, which is more in line with the actual anti-erosion characteristics of clay.

[0239] Optionally, if the target erosion depth is greater than a preset depth threshold, the anti-erosion ability index is additionally multiplied by a corresponding penalty coefficient.

[0240] For example, if the target erosion depth is greater than 2 cm (exceeding 40% of the height of the soil pan 213 , assuming the soil pan 213 is 5 cm high in this application), an additional penalty factor of 0.8 is multiplied because deep erosion causes more irreversible damage to the soil structure.

[0241] Step S202: calculating the anti-breakage capability index of the agglomerates based on the average weight diameter of the target sputtered material.

[0242] Specifically, the electronic device 6 can determine the average weight diameter threshold of the spattered material corresponding to the target soil based on the soil type corresponding to the target soil, and then divide the target average weight diameter of the spattered material by the average weight diameter threshold of the spattered material to obtain the initial aggregate anti-crushing ability. The electronic device 6 then calculates the proportion of second particles with a diameter greater than a preset diameter in the target average weight diameter of the spattered material. If the proportion of second particles is greater than the first preset proportion threshold, the initial aggregate anti-crushing ability is multiplied by a first coefficient to obtain an aggregate anti-crushing ability index. If the proportion of second particles is less than the second preset proportion threshold, the initial aggregate anti-crushing ability is multiplied by a second coefficient to obtain an aggregate anti-crushing ability index. Wherein, the first preset proportion threshold is greater than the second preset proportion threshold, the first coefficient is greater than 1, and the second coefficient is less than 1. If the proportion of second particles is greater than or equal to the second preset proportion threshold and less than or equal to the first preset proportion threshold, the initial aggregate anti-crushing ability is determined as the aggregate anti-crushing ability index.

[0243] For example, a threshold for the average weight diameter of splash material is set based on soil type (2mm for sandy soil, 3mm for loamy soil, and 4mm for clay soil). The formula is adjusted to: Initial aggregate resistance to breakage = Target average weight diameter of splash material divided by the average weight diameter threshold of splash material. If the proportion of particles >0.25mm in the splash material is >60% (the first preset proportion threshold) (predominantly coarse particles), the aggregate resistance to breakage is multiplied by 1.1 (coarse particle aggregates are more stable); if it is <20% (the second preset proportion threshold) (predominantly fine particles), the resistance is multiplied by 0.9 (fine particles are more easily dispersed).

[0244] Example optimization: For sandy soil with an average weight diameter of 2.1 mm for target splash material (threshold 2 mm), this indicator is 2.1 / 2 = 1.0 (capped at 1.0). Since coarse particles account for 70%, the final indicator is 1.0 × 1.1 = 1.1 (normalized to 1.0), which more accurately reflects the actual stability of sandy soil aggregates.

[0245] Step S203: Calculate the overall anti-loss capability index based on the target total sputtering amount.

[0246] Specifically, the electronic device 6 may calculate the initial anti-loss capability based on the formula: initial anti-loss capability=1 / (1+target total sputtering amount / 100).

[0247] For example, the target total sputtering amount is 80 g / m2, and the initial anti-washout capacity is 1 / (1+80 / 100)≈0.56.

[0248] Then, the electronic device 6 calculates the proportion of the sputtering amount in the farthest ring zone in the target total sputtering amount and the proportion of the sputtering amount in the nearest ring zone in the target total sputtering amount. If the proportion of the sputtering amount in the farthest ring zone in the target total sputtering amount is greater than a third preset proportion threshold, the proportion is multiplied by a third coefficient; and if the proportion of the sputtering amount in the nearest ring zone in the target total sputtering amount is greater than a fourth preset proportion threshold, the proportion is multiplied by a fourth coefficient.

[0249] The third preset proportion threshold can be 15% or 18%, and the fourth preset proportion threshold can be 60% or 62%. The third preset proportion threshold and the fourth preset proportion threshold are not limited in the embodiments of the present application. The third coefficient is less than 1, and the fourth coefficient is greater than 1.

[0250] For example, if the sputtering amount proportion of the 40 cm ring zone is >15% (the particle migration distance is far), the anti-washout capacity is multiplied by 0.9 (the washout range is wide, and the risk is higher); and if the 0 cm ring zone proportion is >60% (the particle has not diffused), the anti-washout capacity is multiplied by 1.1 (the washout is concentrated, and it is easy to control). The target total sputtering amount is 80 g / m2, and the 40 cm ring zone proportion is 20%. Therefore, the overall anti-washout capacity index is 1 / (1+80 / 100)×0.9≈0.50, which is more consistent with the influence of the washout range on the risk.

[0251] Optionally, when the target total sputtering amount is greater than a preset total sputtering amount threshold (such as 200 g / m2) (high risk), no matter the calculation result, the anti-washout capacity is the preset anti-washout capacity threshold (such as 0.6) at most – to strengthen the risk warning for the high washout scenario.

[0252] In step S204, the target anti-erosion capacity of the target soil is obtained based on the anti-erosion capacity index, the aggregate anti-breaking capacity index, and the overall anti-washout capacity index.

[0253] Specifically, the electronic device 6 can assign weights according to the physical meanings of the anti-erosion capacity index, the aggregate anti-breaking capacity index, and the overall anti-washout capacity index, and fuse to obtain the target anti-erosion capacity.

[0254] For example, the weight distribution is: the aggregate anti-breaking capacity index (40%) > the anti-erosion capacity index (30%) > the overall anti-washout capacity index (30%) (because the aggregate stability is the core of the anti-erosion).

[0255] The calculation formula is: target anti-erosion capacity=(aggregate anti-breaking capacity index×0.4)+(anti-erosion capacity index×0.3)+(overall anti-washout capacity index×0.3).

[0256] Then, the electronic device 6 maps the target anti-erosion ability to the [0, 1] interval, divides it into 5 levels, and clearly defines the strength of the anti-erosion: very strong (0.8-1.0): the aggregate is complete, the erosion is slight, and almost no particles are lost; strong (0.6-0.8): the aggregate is relatively stable, the erosion is shallow, and the loss is small (for example, 0.67 in the example corresponds to “strong”); medium (0.4-0.6): the aggregate is partially broken, the erosion is moderate, and there is a certain loss; weak (0.2-0.4): the aggregate is largely broken, the erosion is deep, and the loss is obvious; very weak (0-0.2): the aggregate is completely broken, the erosion is serious, and the loss is large.

[0257] The soil splash erosion simulation analysis system provided by the embodiments of the present application quantifies the anti-erosion related abilities from three key links of erosion, i.e., erosion, aggregate breaking, and loss, avoids one-sided judgment of the soil anti-erosion performance by a single index, integrates the indexes of the three stages, completely covers the whole process of anti-erosion performance of the soil from being splashed and stripped to particle breaking and finally to loss, and reflects the comprehensive anti-erosion level. The weak links of the soil anti-erosion (such as weak anti-erosion and easy breaking of the aggregate) can be directly located, and clear basis is provided for targeted improvement (such as increasing organic fertilizer to improve the stability of the aggregate).

[0258] Although the embodiments of the present application are described in combination with the drawings, various modifications and variations can be made by those skilled in the art without departing from the spirit and scope of the present application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A soil splash erosion simulation and analysis system, characterized in that: The system includes: a rainfall simulator, a splash erosion simulation device, a raindrop characteristic measurement device, a data acquisition device, a wind simulator, and an electronic device, wherein the rainfall simulator is installed above the raindrop characteristic measurement device, the raindrop characteristic measurement device is installed above the splash erosion simulation device, and the wind simulator is installed on one side of the raindrop characteristic measurement device; the raindrop characteristic measurement device, the wind simulator, and the data acquisition device are all communicatively connected to the electronic device, wherein: The rainfall simulator is used to simulate rainfall of different intensities and different pH values; The wind simulator is used to adjust the wind speed and direction under the control of the electronic device; The raindrop characteristic measuring device is used to measure raindrop data, wherein the raindrop data includes raindrop diameter, raindrop terminal velocity, raindrop kinetic energy and raindrop movement posture; The splash erosion simulation device is used to simulate the process of raindrops landing on the target soil, causing splash erosion and generating splash erosion materials; The data acquisition device is used to collect the initial pH value and initial conductivity of the target soil before rainfall; and to collect the target pH value, target conductivity, target erosion depth of the target soil after rainfall, and the target average weight diameter of the splash material and the target total splash amount corresponding to the splash material; The electronic device is configured to establish a splash erosion prediction model between the raindrop data and the soil splash erosion data based on the raindrop pH value, the raindrop data, the wind speed, the wind direction, the initial pH value, the initial conductivity, the target pH value, the target conductivity, the target erosion depth, the target average weight diameter of the splash erosion material, and the target total splash erosion amount; and to predict the predicted erosion depth, the predicted average weight diameter of the splash erosion material, and the predicted total splash erosion amount corresponding to other soils to be tested based on the splash erosion prediction model; The splash erosion simulation device includes a splash erosion tray, a collection drawer, and a waterproof cover. The splash erosion tray consists of a central source area and a collection area. The central source area is used to place a soil tray with drainage holes at the bottom. The soil tray is used to hold the target soil. The collection area consists of multiple concentric rings, each separated by a steel plate. Each ring area includes two symmetrical small holes for collecting splash erosion materials. The small holes in the ring areas of adjacent rings are staggered to maintain a sealed and waterproof environment. The collection drawer is installed below the spatter plate and is used to place a spatter material sample collection basin to improve the collection efficiency of the spatter material; The waterproof cover is installed on the periphery of the splash plate to ensure that raindrops only act on the soil plate; The electronic device is configured to calculate a conductivity difference based on the initial conductivity and the target conductivity; and to calculate a pH difference based on the initial pH value and the target pH value. Inputting the raindrop pH value, the raindrop data, the wind speed, and the wind direction into a first branch of an initial splash erosion prediction network, performing feature extraction on the raindrop pH value and the raindrop data to obtain external driving features; Inputting the conductivity difference and the pH value difference into the second branch of the initial splash prediction network, performing feature extraction on the conductivity difference and the pH value difference to obtain soil response features; fusing the external driving feature and the soil response feature to obtain a target feature; The target features are transferred to the output layer, and the virtual ablation depth, the virtual average weight diameter of the spattered material and the virtual total spattered amount are output; Based on the relationship between the virtual erosion depth, the virtual average weight diameter of the sputtering material, and the virtual total sputtering amount and the target erosion depth, the target average weight diameter of the sputtering material, and the target total sputtering amount, the initial sputtering prediction network is trained to obtain the sputtering prediction model.

2. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The rainfall simulator includes: a rain barrel, a regulating pump, a rainfall nozzle, a first liftable support frame, and a motor; wherein the regulating pump is connected to the rain barrel and the rainfall nozzle, the motor is installed on the rainfall nozzle, the first liftable support frame is installed below the rainfall nozzle, and the regulating pump, the rainfall nozzle, the first liftable support frame, and the motor are all communicatively connected to the electronic device, wherein: The rainwater barrel is used to hold rainwater with different pH values; The regulating pump is used to control the flow of water pumped to the rainfall nozzle by rotating speed under the control of the electronic device to simulate rainfall of different intensities; The motor is used to drive the rainfall nozzle to rotate under the control of the electronic device to simulate the effect of natural rainfall; The rainfall nozzle is used to adjust different calibers under the control of the electronic device to adjust the diameter of raindrops, thereby simulating raindrops of different sizes; The first liftable support frame is used to adjust the height of the rainfall nozzle under the control of the electronic device, thereby controlling the height of the rainfall.

3. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The wind simulator includes a rotatable base, a second liftable support frame, a mixed flow fan unit, and a honeycomb guide device installed at the air outlet of the mixed flow fan unit. The rotatable base, the second liftable support frame, and the mixed flow fan unit are all communicatively connected to the electronic device, wherein: The rotatable base is used to rotate under the control of the electronic device to adjust the wind direction; The second elevating support frame is used to adjust its height under the control of the electronic device to match the height of the rainfall nozzle in the rainfall simulator; The mixed flow fan unit is used to adjust the rotation speed and frequency under the control of the electronic equipment, thereby adjusting the wind speed; The honeycomb-shaped flow guide device is used to effectively reduce the turbulence phenomenon generated when the fan discharges air.

4. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The electronic device is configured to correct the raindrop diameter based on the wind speed to obtain a diameter correction coefficient; Calculating raindrop falling time based on the raindrop terminal velocity; Correcting the raindrop falling time based on the soil type corresponding to the target soil to obtain a time correction coefficient; Correcting the raindrop terminal velocity based on the wind speed to obtain a raindrop velocity correction coefficient; Calculating a uniformity coefficient based on the wind speed and the raindrop posture; Calculating a basic spatial range feature based on the diameter correction coefficient, the time correction coefficient, and the raindrop velocity correction coefficient; Calculating an impact range characteristic based on the basic spatial range characteristic and the uniformity coefficient; Obtaining a raindrop kinetic energy density based on the raindrop kinetic energy divided by the diameter correction coefficient; Correcting the kinetic energy of the raindrops based on the wind direction, the wind speed, and the raindrop movement posture to obtain a kinetic energy correction coefficient; Calculating a synergistic factor according to the soil type corresponding to the target soil and the pH value of the raindrop; Calculating a directional intensity coefficient based on the raindrop movement posture and the angle between the wind direction and the impact direction; Calculating energy intensity based on the kinetic energy correction coefficient, the raindrop kinetic energy density, and the synergy factor; Calculating an impact strength characteristic based on the energy intensity and the directional strength coefficient; The impact range feature and the impact intensity feature are fused to generate the external driving feature.

5. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The electronic device is configured to correct the conductivity difference and the pH difference based on a relationship between the conductivity difference and the pH difference to obtain a corrected conductivity coefficient and a corrected pH coefficient; Calculating a soil buffer index based on a first relationship between the corrected conductivity coefficient and the corrected pH coefficient; calculating a structural integrity characteristic based on a second relationship between the corrected conductivity coefficient and the corrected pH coefficient; The soil buffer index and the structural integrity characteristic are fused to generate the soil response characteristic.

6. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The external driving characteristics include an impact range characteristic and an impact intensity characteristic; the soil response characteristics include a soil buffer index and a structural integrity characteristic; the electronic device is used to assign initial weights to the external driving characteristics and the soil response characteristics respectively; Constructing a correlation coefficient matrix between the external driving characteristics and the soil response characteristics; Determining, based on the correlation coefficient matrix, target correlation features having a correlation coefficient greater than a preset correlation threshold; Performing cross operations on the target-related features to generate interactive features; Based on the interaction characteristics, the initial weights are modified to obtain target weights corresponding to the external driving characteristics and the soil response characteristics respectively; The external driving feature and the soil response feature are fused based on the target weight to generate the target feature.

7. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The electronic device is configured to calculate an erosion relative error and an erosion absolute error based on the virtual erosion depth and the target erosion depth; determine first weight information corresponding to the erosion relative error and the erosion absolute error, respectively, based on the soil type corresponding to the target soil; and calculate an erosion error based on the first weight information, the erosion relative error, and the erosion absolute error; Calculating a square error of the average weight diameter of the sputtered material based on the virtual average weight diameter of the sputtered material and the target average weight diameter of the sputtered material; Calculating the proportion of first particles with a diameter greater than a preset diameter in the virtual average weight diameter of the sputtered material and the proportion of second particles with a diameter greater than the preset diameter in the target average weight diameter of the sputtered material; Calculating an average weight diameter error of the sputtered material according to a deviation between the first particle ratio and the second particle ratio and a square error of an average weight diameter of the sputtered material; Calculating a total sputtering amount error according to the virtual total sputtering amount and the target total sputtering amount; generating a target loss function based on the ablation error, the square error of the average weight diameter of the sputtered material, and the total sputtering amount error; The initial sputtering prediction network is trained based on the target loss function to obtain the sputtering prediction model.

8. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The electronic device is further configured to calculate an anti-erosion capability index based on the target erosion depth; Calculating the anti-crushing ability index of the agglomerate based on the average weight diameter of the target spattered material; Calculating an overall anti-loss capability index based on the target total spattering amount; Based on the anti-erosion capacity index, the aggregate anti-breakage capacity index, and the overall anti-loss capacity index, the target anti-erosion capacity corresponding to the target soil is obtained.

Images