Soil heavy metal content prediction method and soil sampler
By designing an adjustable soil sampler and a fully connected neural network model, the problem of the soil sampler adapting to different soil textures was solved, and efficient sampling and accurate prediction of heavy metal content were achieved.
Patent Information
- Application Number
- CN202510826143.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-19
- Publication Date
- 2025-09-16
AI Technical Summary
Existing soil samplers are difficult to adapt to different soil textures, and traditional structures cannot meet the sampling requirements of clay and sand. In addition, existing soil heavy metal content prediction methods ignore dynamic factors such as climate conditions, vegetation cover and soil properties.
A soil sampler was designed. The position of the second curved plate was adjusted by adjusting the component so that it formed a circle or semicircle with the first curved plate. It is suitable for sampling clay and sandy soils. In combination with a fully connected neural network model, the soil heavy metal content was predicted using multi-source data, including satellite remote sensing data and environmental factors.
Flexible sampling of different soil textures was achieved, which improved sampling efficiency and accuracy. At the same time, the prediction accuracy and comprehensiveness of soil heavy metal content were improved through the neural network model.
Smart Images
Figure CN120651568A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of soil samplers, and in particular to a method for predicting soil heavy metal content and a soil sampler. Background Art
[0002] Soil heavy metal pollution is an important part of global environmental problems. The heavy metal content in soil not only affects plant growth and ecosystem health, but is also closely related to human health. Accurately predicting the content of heavy metals in soil is of great significance for environmental risk assessment, agricultural management and pollution control. The distribution of heavy metal content in soil is affected by a variety of environmental factors, including climatic conditions (such as temperature and precipitation), vegetation cover (such as normalized vegetation index), and soil properties (such as sand content and clay content). However, existing methods for predicting the content of heavy metals in soil often only focus on the proportion of heavy metals in some soil samples, ignoring the importance of dynamic factors such as climatic conditions, vegetation cover, and soil properties.
[0003] Soil samples require a soil sampler for soil sampling. Traditional soil samplers mostly adopt a fixed cylindrical structure or a shovel structure. However, different regions require different soil sampler structures, and existing soil samplers are difficult to adapt to different soil textures (such as clay, sand, etc.). Summary of the Invention
[0004] In view of this, the purpose of the present invention is to provide a soil heavy metal content prediction method and a soil sampler to solve the problem in the prior art that different regions require different soil sampler structures and the existing soil samplers are difficult to adapt to different soil textures.
[0005] The present invention is achieved through the following technical solutions:
[0006] A soil sampler includes a handle and a drill tool, the drill tool includes a first curved plate and a second curved plate, one end of the handle is fixedly connected to a mounting plate, the first curved plate is fixedly connected to the mounting plate, an adjustment component is installed between the second curved plate and the mounting plate, and the second curved plate can be adjusted to form a circle with the first curved plate under the adjustment of the adjustment component.
[0007] Further, the adjustment assembly includes a first sleeve sleeved on the handle, the first sleeve is slidably connected to the handle, the mounting plate is provided with a groove adapted to the second arc plate, one end of the second arc plate is fixedly connected to the first sleeve through the groove, the first sleeve is provided with a plurality of notches, and the plurality of notches extend along the length direction of the first sleeve, an external thread is cut on the outer wall of the first sleeve, a second sleeve is sleeved on the first sleeve, an internal thread adapted to the external thread is cut on the inner wall of the second sleeve, the second sleeve is threadedly connected to the first sleeve, an anti-slip strip is fixedly connected to the outer wall of the second sleeve, and a plurality of the anti-slip strips are circumferentially evenly distributed around the central axis of the second sleeve.
[0008] Furthermore, a push rod is slidably connected to the handle, a first through hole adapted to the push rod is provided on the handle, a second through hole adapted to the push rod is provided on the mounting plate, the push rod passes through the first through hole and the second through hole, and a piston block is fixedly connected to one end of the push rod close to the mounting plate, the piston block is adapted to the first arc plate and the second arc plate, and a handle is fixedly connected to one end of the push rod away from the mounting plate.
[0009] Furthermore, a first bolt is provided at one end of the handle away from the mounting plate, a third through hole adapted to the first bolt is opened on the handle, a fourth through hole adapted to the first bolt is opened on the push rod, and the first bolt passes through the third through hole and the fourth through hole.
[0010] Furthermore, a plurality of protruding teeth are fixedly connected to one side of the first arc plate and the second arc plate away from the mounting plate.
[0011] Furthermore, a ring-shaped protrusion is fixedly connected to the second curved plate, and a limiting groove adapted to the protrusion is provided on the mounting plate. The limiting groove is ring-shaped, and the protrusion is rotatably connected in the limiting groove. The adjustment assembly includes a second bolt, and a fifth through hole adapted to the second bolt is provided on the side of the first curved plate close to the mounting plate, and a sixth through hole adapted to the second bolt is provided on the second curved plate, and the second bolt passes through the fifth through hole and the sixth through hole.
[0012] A method for predicting heavy metal content in soil, comprising a soil sampler,
[0013] Step 1: Collect soil using a soil sampler based on the existing sampling point location information in the target area;
[0014] Step 2: Collect soil heavy metal data and, through satellite remote sensing spectral inversion, obtain data on the normalized vegetation index, saturated water vapor pressure deficit, temperature, precipitation, relative humidity, light absorption scale factor, carbon dioxide concentration, soil sand content, soil clay content, vapor pressure, maximum root depth, drought index, wind speed at 10 meters, and soil moisture, and perform preprocessing.
[0015] Step 3: Build a fully connected neural network model, and use the mean-standard deviation standardization method to standardize the data collected in step 2. The normalized data is used as the input variable; the normalized data is input into the fully connected neural network, and the model is trained using the training set data. The loss function is calculated based on the output value and the true value, and the model parameters are updated using the back propagation algorithm until the model performance converges; during the training process, the model performance is regularly evaluated using the validation set, and the determination coefficient R of the training set and validation set is 0. 2 When the set value is reached and it is confirmed that the model has no significant overfitting or underfitting phenomenon, the heavy metal content prediction model is obtained, and the independent variables are input into the fully connected neural network model for calculation. The output result is the predicted value of the heavy metal content in the soil to be tested.
[0016] Furthermore, the fully connected neural network model includes a four-layer network structure, the first layer of the network structure is a fully connected layer, the number of neurons is 100, and the activation function is sigmoid; the second layer of the network structure is a fully connected layer, the number of neurons is 100, and the activation function is sigmoid; the third layer of the network structure is a fully connected layer, the number of neurons is 100, and the activation function is sigmoid; the fourth layer of the network structure is a fully connected layer, the number of neurons is 1, and there is no activation function.
[0017] Furthermore, the preprocessing includes filling missing values and removing outliers for the collected normalized vegetation index, saturated water vapor pressure difference, temperature, precipitation, relative humidity, light absorption scaling factor, carbon dioxide concentration, soil sand content, soil clay content, vapor pressure, maximum root depth, drought index, ten-meter high wind speed, soil moisture and soil heavy metal content.
[0018] Furthermore, the coefficient of determination R 2 The calculation formula is
[0019]
[0020] R 2 represents the coefficient of determination, y i represents the true value of sample i, represents the predicted value of sample i, represents the average value of sample i.
[0021] The beneficial effects of the present invention are:
[0022] The method for predicting heavy metal content in soil and the soil sampler can adjust the position of the second curved plate by adjusting the component, so that the first curved plate and the second curved plate can be enclosed into a circle for sampling clay soil, or only the semicircular first curved plate can be retained for sampling sandy soil.
[0023] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] Figure 1 The three-dimensional embodiment of the present invention Figure 1 ;
[0025] Figure 2 The three-dimensional embodiment of the present invention Figure 2 ;
[0026] Figure 3 This is a schematic structural diagram of embodiment 1 of the present invention;
[0027] Figure 4 This is a schematic diagram of the connection between the handle and the mounting plate according to an embodiment of the present invention;
[0028] Figure 5 This is a schematic diagram of the connection between the push rod and the piston block in accordance with the first embodiment of the present invention;
[0029] Figure 6 The three-dimensional embodiment of the present invention Figure 1 ;
[0030] Figure 7 The three-dimensional embodiment of the present invention Figure 2 ;
[0031] Figure 8 This is a structural diagram of embodiment 2 of the present invention;
[0032] Figure 9 This is a schematic diagram of the connection between the handle and the mounting plate according to the second embodiment of the present invention;
[0033] Figure 10 Flowchart for constructing the soil zinc content model of the present invention
[0034] Figure 11 It is a scatter plot of the measured value and predicted value of soil zinc content in the present invention.
[0035] In the picture:
[0036] 1. Handle; 2. Drilling tool; 3. First curved plate; 4. Second curved plate; 5. Mounting plate; 6. Adjustment assembly; 7. First sleeve; 8. Groove; 9. Notch; 10. External thread; 11. Second sleeve; 12. Internal thread; 13. Anti-slip strip; 14. Push rod; 15. First through hole; 16. Second through hole; 17. Piston block; 18. Handle; 19. First bolt; 20. Third through hole; 21. Fourth through hole; 22. Protruding tooth; 23. Bump; 24. Groove; 25. Second bolt; 26. Fifth through hole; 27. Sixth through hole. DETAILED DESCRIPTION
[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.
[0038] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.
[0039] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not require further definition or explanation in subsequent drawings.
[0040] In the above description of the present invention, it should be noted that the terms "one side," "the other side," and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, or the orientations or positional relationships in which the inventive product is typically placed when in use. These terms are intended solely to facilitate the description of the present invention and simplify the description, and are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and the like are used solely for distinction and should not be construed as indicating or implying relative importance.
[0041] Furthermore, the term "identical" and similar terms do not necessarily require that the components be absolutely identical; slight variations are permitted. The term "perpendicular" simply refers to the positional relationship between components being more perpendicular than "parallel," not that the structure must be perfectly vertical; rather, it can be slightly tilted.
[0042] Example 1:
[0043] See also Figure 1-5 The present invention provides a technical solution: a soil sampler, comprising a handle 1 and a drill tool 2, the drill tool 2 comprising a first curved plate 3 and a second curved plate 4, one end of the handle 1 is fixedly connected to a mounting plate 5, the first curved plate 3 is fixedly connected to the mounting plate 5, an adjustment component 6 is installed between the second curved plate 4 and the mounting plate 5, and the second curved plate 4 can be adjusted by the adjustment component 6 to enclose a circle with the first curved plate 3.
[0044] The adjusting assembly 6 includes a first sleeve 7 sleeved on the handle 1, the first sleeve 7 is slidably connected to the handle 1, the mounting plate 5 is provided with a groove 8 adapted to the second arc-shaped plate 4, one end of the second arc-shaped plate 4 is fixedly connected to the first sleeve 7 through the groove 8, the first sleeve 7 is provided with a plurality of notches 9, and the plurality of notches 9 extend along the length direction of the first sleeve 7, an external thread 10 is cut on the outer wall of the first sleeve 7, a second sleeve 11 is sleeved on the first sleeve 7, an internal thread 12 adapted to the external thread 10 is cut on the inner wall of the second sleeve 11, the second sleeve 11 is threadedly connected to the first sleeve 7, an anti-slip strip 13 is fixedly connected to the outer wall of the second sleeve 11, and a plurality of the anti-slip strips 13 are circumferentially evenly distributed around the central axis of the second sleeve 11.
[0045] In this solution, the handle 1 is T-shaped, which provides a better grip and rotation force point for the user to drive the drill 2 to cut into the soil.
[0046] The mounting plate 5 is fixedly connected to the lower end of the handle 1 for fixing the first arc plate 3 , and the second arc plate 4 is moved along the length direction of the handle 1 by providing a groove 8 on the mounting plate 5 .
[0047] The first curved plate 3 and the second curved plate 4 are both semicircular. The first curved plate 3 and the second curved plate 4 can enclose a circle to sample clay soil, and after the second curved plate 4 moves, the first curved plate 3 can sample sand.
[0048] The first sleeve 7 is slidably connected to the handle 1 and fixedly connected to the second curved plate 4, facilitating the movement of the second curved plate 4. A notch 9 and external threads 10 are provided along the first sleeve 7, allowing the second sleeve 11 to be threadedly connected to the second sleeve 11 via internal threads 12. The friction between the first sleeve 7 and the handle 1 secures the second curved plate 4 to the handle 1. A plurality of anti-slip strips 13 are provided on the second sleeve 11 to increase friction, making it easier for the user to rotate the second sleeve 11 to secure it to the first sleeve 7.
[0049] Working principle and usage:
[0050] Step 1: The user determines whether the soil at the sampling point is clay or sand by visual observation.
[0051] Step 2: If the soil at the sampling point is clay, rotate the second sleeve 11 clockwise to move it along the external thread 10 of the first sleeve 7, unlock the second curved plate 4 and the first sleeve 7 from the handle 1, move the first sleeve 7 and the second curved plate 4 toward the lower end of the handle 1, so that the first curved plate 3 and the second curved plate 4 enclose a circle, move the second sleeve 11 toward the first sleeve 7, rotate the second sleeve 11 counterclockwise to move it along the external thread 10 of the first sleeve 7, and lock the second curved plate 4 and the first sleeve 7 from the handle 1.
[0052] If the soil at the sampling point is sandy, rotate the second sleeve 11 clockwise to move it along the external thread 10 of the first sleeve 7, unlock the second curved plate 4 and the first sleeve 7 from the handle 1, move the first sleeve 7, the second sleeve 11 and the second curved plate 4 toward the upper end of the handle 1, so that most of the second curved plate 4 is located at the upper end of the mounting plate 5, move the second sleeve 11 toward the first sleeve 7, rotate the second sleeve 11 counterclockwise to move it along the external thread 10 of the first sleeve 7, and lock the second curved plate 4 and the first sleeve 7 from the handle 1.
[0053] Step 3: If the soil at the sampling point is clay, insert the first curved plate 3 and the second curved plate 4 vertically into the soil, rotate the handle 1 to make the curved plates cut into the soil to a certain depth, and when the sampling cavity is filled with soil, pull up the handle 1 to take out the sample.
[0054] If the soil at the sampling point is sandy, the first curved plate 3 is inserted obliquely into the soil to shovel up the soil.
[0055] Step 4: If the soil at the sampling point is clay, rotate the second sleeve 11 clockwise to move it along the external thread 10 of the first sleeve 7, unlock the second curved plate 4 and the first sleeve 7 from the handle 1, move the first sleeve 7, the second sleeve 11 and the second curved plate 4 toward the upper end of the handle 1, so that most of the second curved plate 4 is located at the upper end of the mounting plate 5, move the second sleeve 11 toward the first sleeve 7, rotate the second sleeve 11 counterclockwise to move it along the external thread 10 of the first sleeve 7, and lock the second curved plate 4 and the first sleeve 7 with the handle 1 to facilitate pouring out the soil.
[0056] Compared with the prior art, the position of the second arc plate 4 is adjusted by adjusting the adjustment component 6, so that the first arc plate 3 and the second arc plate 4 can be enclosed into a circle for clay soil sampling, or only the semicircular first arc plate 3 can be retained for sand sampling.
[0057] In this embodiment: a push rod 14 is slidably connected to the handle 1, a first through hole 15 adapted to the push rod 14 is opened on the handle 1, a second through hole 16 adapted to the push rod 14 is opened on the mounting plate 5, the push rod 14 passes through the first through hole 15 and the second through hole 16, the push rod 14 is fixedly connected to a piston block 17 at one end close to the mounting plate 5, the piston block 17 is adapted to the first curved plate 3 and the second curved plate 4, and the push rod 14 is fixedly connected to a handle 18 at one end away from the mounting plate 5.
[0058] In this solution: a push rod 14 is slidably connected to the handle 1, a first through hole 15 adapted to the push rod 14 is provided on the handle 1, a second through hole 16 adapted to the push rod 14 is provided on the mounting plate 5, the push rod 14 passes through the first through hole 15 and the second through hole 16, the end of the push rod 14 close to the mounting plate 5 is fixedly connected to a piston block 17, the piston block 17 is adapted to the first curved plate 3 and the second curved plate 4, and the end of the push rod 14 away from the mounting plate 5 is fixedly connected to a handle 18.
[0059] After sampling the clay soil, push the handle 18 to move the push rod 14 downward, and the piston block 17 pushes the soil sample out from between the first curved plate 3 and the second curved plate 4 to avoid sample residue. Since soil sampling requires soil samples at multiple points, after taking out the soil sample at the current point, sampling at different positions can be performed to improve sampling efficiency.
[0060] In this embodiment: a first bolt 19 is provided at the end of the handle 1 away from the mounting plate 5, a third through hole 20 adapted to the first bolt 19 is provided on the handle 1, a fourth through hole 21 adapted to the first bolt 19 is provided on the push rod 14, and the first bolt 19 passes through the third through hole 20 and the fourth through hole 21.
[0061] In this solution, a first bolt 19 is provided at the end of the handle 1 away from the mounting plate 5, a third through hole 20 adapted to the first bolt 19 is provided on the handle 1, and a fourth through hole 21 adapted to the first bolt 19 is provided on the push rod 14, and the first bolt 19 passes through the third through hole 20 and the fourth through hole 21.
[0062] After the first bolt 19 is passed through the third through hole 20 and the fourth through hole 21, the position of the push rod 14 can be locked to prevent accidental pushing during transportation. The push rod 14 can serve as an extension part to facilitate the user to apply force.
[0063] In this embodiment, a plurality of protruding teeth 22 are fixedly connected to the side of the first arc plate 3 and the second arc plate 4 away from the mounting plate 5 .
[0064] In this solution, a plurality of protruding teeth 22 are fixedly connected to the first arc-shaped plate 3 and the second arc-shaped plate 4 on one side away from the mounting plate 5 .
[0065] During the sampling process of clay soil, the first curved plate 3 and the second curved plate 4 rotate and insert into the soil, driving the convex teeth 22 to rotate. The convex teeth 22 increase local stress, break the hard soil, and speed up the sampling speed.
[0066] Example 2:
[0067] See also Figure 6-9 The difference between this embodiment and the first embodiment is that: an annular protrusion 23 is fixedly connected to the second curved plate 4, a limiting groove 24 adapted to the protrusion 23 is opened on the mounting plate 5, the limiting groove 24 is annular, and the protrusion 23 is rotatably connected in the limiting groove 24, the adjusting assembly 6 includes a second bolt 25, a fifth through hole 26 adapted to the second bolt 25 is opened on the side of the first curved plate 3 close to the mounting plate 5, and a sixth through hole 27 adapted to the second bolt 25 is opened on the second curved plate 4, and the second bolt 25 passes through the fifth through hole 26 and the sixth through hole 27.
[0068] In this solution, an annular protrusion 23 is fixedly connected to the second arc-shaped plate 4 , and a groove 8 adapted to the protrusion 23 is provided on the mounting plate 5 to limit the sliding track of the protrusion 23 and the second arc-shaped plate 4 .
[0069] The second bolt 25 passes through the fifth through hole 26 and the sixth through hole 27 to mechanically lock the relative positions of the first curved plate 3 and the second curved plate 4, so that the first curved plate 3 and the second curved plate 4 can switch between an enclosed circular state and a semicircular state.
[0070] Working principle and usage:
[0071] Step 1: The user determines whether the soil at the sampling point is clay or sand by visual observation.
[0072] Step 2: If the soil at the sampling point is clay, pull the second bolt 25 out of the fifth through hole 26 and the sixth through hole 27, rotate the second curved plate 4 so that the first curved plate 3 and the second curved plate 4 form a circle, and then insert the second bolt 25 into the fifth through hole 26 and the sixth through hole 27 to fix the relative position of the first curved plate 3 and the second curved plate 4.
[0073] If the soil at the sampling point is sandy, pull the second bolt 25 out of the fifth through hole 26 and the sixth through hole 27, rotate the second curved plate 4 so that the first curved plate 3 and the second curved plate 4 fit together in a semicircle, and then insert the second bolt 25 into the fifth through hole 26 and the sixth through hole 27 to fix the relative position of the first curved plate 3 and the second curved plate 4.
[0074] Step 3: If the soil at the sampling point is clay, insert the first curved plate 3 and the second curved plate 4 vertically into the soil, rotate the handle 1 to make the curved plates cut into the soil to a certain depth, and when the sampling cavity is filled with soil, pull up the handle 1 to take out the sample.
[0075] If the soil at the sampling point is sandy, the first curved plate 3 and the second curved plate 4 are inserted obliquely into the soil to shovel up the soil.
[0076] Step 4: If the soil at the sampling point is clay, pull the second bolt 25 out of the fifth through hole 26 and the sixth through hole 27, rotate the second curved plate 4 so that the first curved plate 3 and the second curved plate 4 fit together in a semicircular shape, and then insert the second bolt 25 into the fifth through hole 26 and the sixth through hole 27 to fix the relative position of the first curved plate 3 and the second curved plate 4 to facilitate pouring out the soil.
[0077] Compared with the prior art, the position of the second arc plate 4 is adjusted by adjusting the adjustment component 6, so that the first arc plate 3 and the second arc plate 4 can be enclosed into a circle for sampling clay soil, and can also be fitted into a semicircle for sampling sand soil.
[0078] Example 3:
[0079] See also Figure 10-11 , a soil heavy metal content prediction method, comprising a soil sampler,
[0080] Step 1: Collect soil using a soil sampler based on the existing sampling point location information in the target area;
[0081] Step 2: Collect soil heavy metal data, and at the same time obtain data on normalized vegetation index, saturated water vapor pressure difference, temperature, precipitation, relative humidity, light absorption scaling factor, carbon dioxide concentration, soil sand content, soil clay content, vapor pressure, maximum root depth, drought index, ten-meter high wind speed and soil moisture through satellite remote sensing spectral inversion, and perform preprocessing; the preprocessing includes filling missing values and removing outliers for the collected normalized vegetation index, saturated water vapor pressure difference, temperature, precipitation, relative humidity, light absorption scaling factor, carbon dioxide concentration, soil sand content, soil clay content, vapor pressure, maximum root depth, drought index, ten-meter high wind speed, soil moisture and soil heavy metal content.
[0082] Step 3: Build a fully connected neural network model, which includes a four-layer network structure. The first layer is a fully connected layer with 100 neurons and a sigmoid activation function. The second layer is a fully connected layer with 100 neurons and a sigmoid activation function. The third layer is a fully connected layer with 100 neurons and a sigmoid activation function. The fourth layer is a fully connected layer with 1 neuron and no activation function. The data collected in step 2 is standardized using the mean-standard deviation standardization method, and the normalized data is used as an input variable. The normalized data is input into the fully connected neural network, and the model is trained using the training set data. The loss function is calculated based on the output value and the true value, and the model parameters are updated using the back-propagation algorithm until the model performance converges. During the training process, the model performance is regularly evaluated using the validation set, and the coefficient of determination R of the training set and the validation set is 1. 2 When the set value is reached and it is confirmed that the model has no significant overfitting or underfitting phenomenon, the heavy metal content prediction model is obtained, and the independent variables are input into the fully connected neural network model for calculation. The output result is the predicted value of the heavy metal content in the soil to be tested.
[0083] The coefficient of determination R 2 The calculation formula is
[0084]
[0085] R 2 represents the coefficient of determination, y i represents the true value of sample i, represents the predicted value of sample i, represents the average value of sample i.
[0086] In this scheme: the method is divided into two stages: model training and prediction: first, execute steps A to B to build a prediction model for the heavy metal zinc content in the soil of the target area; then use steps I to III to obtain the predicted value of the heavy metal zinc content in the soil to be tested in the target area.
[0087] Step A. Data acquisition and preprocessing
[0088] Based on the existing sampling point location information in the target area, soil heavy metal data such as Fe, Mn, As, Ba, Cr, Cu, Hg, Ni, Pb, V, and Co were collected. At the same time, the normalized difference vegetation index (NDVI), saturated vapor pressure difference, temperature, precipitation, relative humidity, light absorption scale factor, carbon dioxide concentration, soil sand content, soil clay content, vapor pressure, maximum root depth, drought index, wind speed at 10 meters high, and monthly or annual time series data of soil moisture were obtained through satellite remote sensing spectral inversion.
[0089] The collected data was preprocessed. For missing values, we used the mean imputation method, which imputed missing values with the mean of the variable across all samples. For outliers, we used the 3-standard-deviation rule to eliminate them. If a data point deviated from the mean by more than 3 standard deviations, it was considered an outlier and eliminated. After preprocessing, we obtained a training set (630 samples) and a validation set (270 samples), and then proceeded to Step B.
[0090] Step B. Model construction and training
[0091] Use Python's TensorFlow framework to build a fully connected neural network model, use the preprocessed data in step A as input variables, extract features and perform model training, and output concept correlation features representing multi-source data.
[0092] In practical applications, the above step B specifically includes the following steps B1 to B4:
[0093] Step B1. Build a fully connected neural network model
[0094] A fully connected neural network model was constructed with four layers: the first layer was a fully connected (dense) layer with 100 neurons and a sigmoid activation function; the second layer was a fully connected (dense) layer with 100 neurons and a sigmoid activation function; the third layer was a fully connected (dense) layer with 100 neurons and a sigmoid activation function; and the fourth layer was a fully connected (dense) layer with 1 neuron, no activation function, and directly outputting the predicted value. The model parameters were set as follows: 100 training epochs, Adam optimizer, learning rate 0.01, L2 regularization type, regularization rate 0.001, no dropout, and cyclic dropout.
[0095] Step B2. Data normalization
[0096] The data collected in step A were input standardized using the mean-standard deviation standardization method (avgstd). This standardization normalized the numerical range of all input variables to a distribution with a mean of 0 and a standard deviation of 1, and used as the input variables of the model.
[0097] Step B3. Model training
[0098] The model was trained using a training set (630 samples). During training, the measured soil heavy metal zinc content was used as the target value. The gradient of the loss function with respect to the model parameters was calculated using the backpropagation algorithm, and the model parameters were updated using the Adam optimizer. The forward computation, loss calculation, backpropagation, and parameter update process were repeated until the preset number of training rounds was 100. L2 regularization (regularization rate 0.001) was applied during training to mitigate overfitting.
[0099] Step B4. Model Validation
[0100] During the training process, the validation set (270 samples) is used regularly to evaluate the model performance to check whether there is overfitting or underfitting problem. The performance evaluation indicators include the coefficient of determination R 2 , the calculation formula is as follows:
[0101]
[0102] In the formula: represents the true value of sample i, represents the predicted value of sample i, and represents the average value of sample i.
[0103] The final training results show that the R 2 The R of the validation set is 0.984. 2 It is 0.565, indicating that the model has high prediction accuracy and stability, and there is no significant overfitting or underfitting phenomenon.
[0104] Step I. Collect soil data to be tested
[0105] Collect soil heavy metal data of Fe, Mn, As, Ba, Cr, Cu, Hg, Ni, Pb, V, and Co in the target area, and obtain the normalized vegetation index, saturated water vapor pressure difference, temperature, precipitation, relative humidity, light absorption scale factor, carbon dioxide concentration, soil sand content, soil clay content, vapor pressure, maximum root depth, drought index, 10-meter high wind speed, and soil moisture data corresponding to the soil through satellite remote sensing spectral inversion, and then proceed to step II.
[0106] Step II. Extraction of soil characteristics
[0107] Following the method in step B2, perform mean-standard deviation normalization on the data collected in step I to extract the characteristics corresponding to the soil to be tested. During normalization, use the mean and standard deviation calculated in the training set to ensure that the test data and the training data fall within the same distribution range. After normalization, obtain the characteristic vector of the soil to be tested, and proceed to step III.
[0108] Step III. Prediction of soil heavy metal zinc content
[0109] Taking the features extracted in step II as input, the fully connected neural network model trained in step B is used for calculation, and the output result is the predicted value of the heavy metal zinc content in the soil to be tested. Figure 2 As shown in the figure, the scatter plot of the measured values and the predicted values shows that the two show a good linear relationship in both the training set and the validation set, indicating that the model prediction results are highly consistent with the actual values.
[0110] The present invention integrates multi-source dynamic data such as normalized vegetation index, saturated water vapor pressure deficit, drought index, and related variables such as soil heavy metal Fe and Mn to fully capture the environmental and soil factors that affect zinc content, thereby improving the comprehensiveness and accuracy of the prediction. The fully connected neural network is combined with input standardization and L2 regularization to effectively alleviate the overfitting problem and improve the generalization ability of the model on small sample data sets. The model training and verification results show that the training set R 2 The validation set R 2 The prediction accuracy is 0.565, which is relatively high and suitable for environmental monitoring and soil pollution assessment.
[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A soil sampler comprising a handle (1) and a drill (2), characterized in that: The drilling tool (2) comprises a first curved plate (3) and a second curved plate (4); one end of the handle (1) is fixedly connected to a mounting plate (5); the first curved plate (3) is fixedly connected to the mounting plate (5); an adjustment component (6) is installed between the second curved plate (4) and the mounting plate (5); the second curved plate (4) can be adjusted by the adjustment component (6) to enclose a circle with the first curved plate (3) to form a circle.
2. The soil sampler according to claim 1, characterized in that: The adjusting assembly (6) comprises a first sleeve (7) sleeved on the handle (1), the first sleeve (7) being slidably connected to the handle (1), the mounting plate (5) being provided with a groove (8) adapted to the second arc-shaped plate (4), one end of the second arc-shaped plate (4) passing through the groove (8) and fixedly connected to the first sleeve (7), the first sleeve (7) being provided with a plurality of notches (9), the plurality of notches (9) extending along the length direction of the first sleeve (7), the first notches (9) being provided with a plurality of notches (9 ... An external thread (10) is cut on the outer wall of a sleeve (7), a second sleeve (11) is sleeved on the first sleeve (7), an internal thread (12) adapted to the external thread (10) is cut on the inner wall of the second sleeve (11), the second sleeve (11) is threadedly connected to the first sleeve (7), an anti-slip strip (13) is fixedly connected to the outer wall of the second sleeve (11), and a plurality of the anti-slip strips (13) are uniformly distributed circumferentially around the central axis of the second sleeve (11).
3. The soil sampler according to claim 1, characterized in that: The handle (1) is slidably connected to a push rod (14), the handle (1) is provided with a first through hole (15) adapted to the push rod (14), the mounting plate (5) is provided with a second through hole (16) adapted to the push rod (14), the push rod (14) passes through the first through hole (15) and the second through hole (16), the push rod (14) is fixedly connected to an end of the push rod (14) close to the mounting plate (5) with a piston block (17), the piston block (17) is adapted to the first arc plate (3) and the second arc plate (4), and the push rod (14) is fixedly connected to an end of the push rod (14) away from the mounting plate (5) with a handle (18).
4. The soil sampler according to claim 3, characterized in that: A first bolt (19) is provided at one end of the handle (1) away from the mounting plate (5), a third through hole (20) adapted to the first bolt (19) is provided on the handle (1), a fourth through hole (21) adapted to the first bolt (19) is provided on the push rod (14), and the first bolt (19) passes through the third through hole (20) and the fourth through hole (21).
5. The soil sampler according to claim 1, characterized in that: A plurality of protruding teeth (22) are fixedly connected to the first arc-shaped plate (3) and the second arc-shaped plate (4) on one side away from the mounting plate (5).
6. The soil sampler according to claim 1, characterized in that: The second arc-shaped plate (4) is fixedly connected with an annular protrusion (23), the mounting plate (5) is provided with a limiting groove (24) adapted to the protrusion (23), the limiting groove (24) is annular, and the protrusion (23) is rotatably connected in the limiting groove (24), the adjusting assembly (6) includes a second bolt (25), a fifth through hole (26) adapted to the second bolt (25) is provided on a side of the first arc-shaped plate (3) close to the mounting plate (5), the second arc-shaped plate (4) is provided with a sixth through hole (27) adapted to the second bolt (25), and the second bolt (25) passes through the fifth through hole (26) and the sixth through hole (27).
7. A method for predicting heavy metal content in soil, comprising the soil sampler according to claim 1, characterized in that: Step 1: Collect soil using a soil sampler based on the existing sampling point location information in the target area; Step 2: Collect soil heavy metal data and, through satellite remote sensing spectral inversion, obtain data on the normalized vegetation index, saturated water vapor pressure deficit, temperature, precipitation, relative humidity, light absorption scale factor, carbon dioxide concentration, soil sand content, soil clay content, vapor pressure, maximum root depth, drought index, wind speed at 10 meters, and soil moisture, and perform preprocessing. Step 3: Build a fully connected neural network model, and use the mean-standard deviation standardization method to standardize the data collected in step 2. The normalized data is used as the input variable; the normalized data is input into the fully connected neural network, and the model is trained using the training set data. The loss function is calculated based on the output value and the true value, and the model parameters are updated using the back propagation algorithm until the model performance converges; during the training process, the model performance is regularly evaluated using the validation set, and the determination coefficient R of the training set and validation set is 0. 2 When the set value is reached and it is confirmed that the model has no significant overfitting or underfitting phenomenon, the heavy metal content prediction model is obtained, and the independent variables are input into the fully connected neural network model for calculation. The output result is the predicted value of the heavy metal content in the soil to be tested.
8. The method for predicting heavy metal content in soil according to claim 7, wherein: The fully connected neural network model includes a four-layer network structure, the first layer of the network structure is a fully connected layer, the number of neurons is 100, and the activation function is sigmoid; the second layer of the network structure is a fully connected layer, the number of neurons is 100, and the activation function is sigmoid; the third layer of the network structure is a fully connected layer, the number of neurons is 100, and the activation function is sigmoid; the fourth layer of the network structure is a fully connected layer, the number of neurons is 1, and there is no activation function.
9. The method for predicting heavy metal content in soil according to claim 7, wherein: The preprocessing includes filling missing values and removing outliers for the collected normalized vegetation index, saturated water vapor pressure difference, temperature, precipitation, relative humidity, light absorption scaling factor, carbon dioxide concentration, soil sand content, soil clay content, vapor pressure, maximum root depth, drought index, ten-meter high wind speed, soil moisture and soil heavy metal content.
10. The method for predicting heavy metal content in soil according to claim 7, wherein: Coefficient of determination R 2 The calculation formula is R 2 represents the coefficient of determination, y i represents the true value of sample i, represents the predicted value of sample i, represents the average value of sample i.