Soil heavy metal monitoring device based on spectrum deep learning and evaluation method

Abstract

The invention discloses a soil heavy metal monitoring device based on spectrum deep learning and an evaluation method, the soil heavy metal monitoring device comprises a plurality of column leaching containers, the plurality of column leaching containers are sequentially connected in series through hoses to form a percolation passage, the front end of the percolation passage is connected with a water outlet bucket through a peristaltic pump, and the rear end of the percolation passage is connected with a water inlet bucket; a branch pipe is led out of the hose between every two adjacent column leaching containers through a three-way pipe, the multiple branch pipes are connected with the multiple collection bottles respectively, and electrochemical detection assemblies are arranged on the collection bottles; the plurality of cuvette brackets are respectively arranged on the plurality of hoses, each cuvette bracket is connected with the optical fiber spectrometers through optical fibers, and the plurality of optical fiber spectrometers are electrically connected with the computer; the scheme aims at providing an effective evaluation method aiming at the soil pollution problem of a heavy metal high background area, the concentration of harmful heavy metal elements in the soil is remarkably reduced, the soil quality is improved, and important technical support and theoretical basis are provided for soil remediation.

Description

Technical Field

[0001] The present invention relates to the field of geological infiltration technology, and in particular to a soil heavy metal monitoring device and an evaluation method based on spectral deep learning. Background Art

[0002] Due to geological history, when excavation projects are carried out in strata that naturally contain toxic metals and metalloids, the excavated soil and rock piles will be exposed to water and oxygen and will undergo varying degrees of leaching, causing varying degrees of pollution to the surrounding soil, posing a significant risk to the environment and human health.

[0003] Therefore, further planning for soil remediation or disposal is necessary. Various types of batch and column tests have been developed to evaluate leaching behavior in contaminated soils. Column percolation tests have been more widely used in recent years because they can more closely simulate the contact between in situ water and granular materials. Currently, column percolation tests have been used to shorten ISO / TS tests on four different soils with four equilibrium periods and two flow rates. 21268-3, the final column percolation test time can be shortened from 20 to 30 days to 7 to 9 days; the initial equilibrium period is shortened from 48h to 12 to 16h; the flow rate is increased from 12mL / h to 36mL / h; by using two columns of different sizes (50mm×300mm and 150mm×700mm) to carry out column percolation tests on marine sediment samples under saturated / unsaturated flow conditions, the leaching behavior of arsenic (As) and other inorganic components is studied; by using the ore of a sandstone uranium deposit to be developed, acid and alkali column leaching tests were carried out, and oxidation leaching was compared; the results show that the acid leaching effect is better than the alkaline leaching effect, and oxidation leaching is better than non-oxidation leaching.

[0004] Traditional column filtration tests are mostly carried out in simple single-column containers, and the leaching solution can only be collected for testing after the test. The sampling method is relatively simple, the efficiency is relatively slow, and the test effect is poor. Therefore, it is necessary to research and develop a new multi-container, continuous column filtration test system, and the reaction temperature, leaching solution pH, Eh, conductivity and other parameters can be intuitively obtained over time during the reaction. Because water chemical reactions will absorb and release heat, the direction and extent of the chemical reaction can be judged by the change in temperature in the column leaching container. Therefore, there is an urgent need for a test device for heavy metal pollution remediation that can effectively improve the efficiency and accuracy of column filtration tests, simplify the test process, and greatly reduce the workload of the test. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a soil heavy metal monitoring device and evaluation method based on spectral deep learning, which solves the problems of cumbersome operation, low efficiency and poor test effect of the existing column infiltration test.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is:

[0007] In a first aspect, a soil heavy metal monitoring device based on spectral deep learning is provided, which includes: a plurality of column immersion containers, which are connected in series in sequence through hoses to form an infiltration path, the front end of the infiltration path is connected to a water outlet bucket filled with deionized water through a peristaltic pump, and the rear end of the infiltration path is connected to a water inlet bucket, and the plurality of column immersion containers are provided with temperature sensors; a plurality of collecting bottles, a hose between two adjacent column immersion containers leads to a branch pipe through a tee pipe, and the plurality of branch pipes are respectively connected to the plurality of collecting bottles, and the collecting bottles are provided with an electrochemical detection component for detecting the leachate; a plurality of cuvette holders, which are respectively arranged on the plurality of hoses, each cuvette holder is connected to a fiber optic spectrometer through an optical fiber, and the plurality of fiber optic spectrometers are electrically connected to a computer; a support frame, which is a multi-layer frame structure for integrated installation of various test components.

[0008] Furthermore, the column immersion container includes an organic glass column tube, both ends of which are provided with flanges, the flanges are connected to the orifice plate and the sealing cover in sequence through butterfly bolts, and rubber pads are provided between the flange and the orifice plate, and between the orifice plate and the sealing cover, the two ends of the hose are respectively arranged on the sealing covers of two adjacent column immersion containers, an insulation layer is provided on the organic glass column tube, and the temperature sensor is located between the organic glass column tube and the insulation layer.

[0009] Furthermore, both ends of the organic glass column tube are filled with quartz sand layers, and the organic glass column tube between the quartz sand layers is filled with mineral pile or adsorption material.

[0010] Furthermore, the electrochemical detection component includes a dissolved oxygen meter, a pH meter, an Eh meter and a conductivity meter electrically connected to the computer.

[0011] In a second aspect, a method for evaluating a soil heavy metal monitoring device based on spectral deep learning is provided, which comprises the following steps:

[0012] S1: Select soil from the high background area of ​​heavy metals as mineral pile material, and sample and measure the concentration of common elements in the mineral pile material;

[0013] S2: sequentially assembling two adjacent column leaching containers from a plurality of column leaching containers into a test unit, and respectively filling the mineral pile into the column leaching container at the front end of each test unit, and respectively filling different types of adsorption materials into the column leaching container at the rear end of each test unit;

[0014] S3: driving deionized water to penetrate several column immersion containers in sequence along the infiltration path through a peristaltic pump, and during the infiltration test, monitoring the temperature of several column immersion containers in real time through a temperature sensor, and monitoring several water chemical indicators of the leachate in several hoses in real time through an electrochemical detection component and an optical fiber spectrometer, wherein the several water chemical indicators include the electrochemical parameters of the leachate in each collection bottle and the concentration of various harmful heavy metal ions in the leachate in each section of the hose; and after the infiltration test, sampling and measuring the concentration of common elements in the mineral stockpile again;

[0015] S4: Analyze the changing trends of several water chemical indicators of the leachate in the two sections of the hose in each test unit to obtain the effects of different types of adsorption materials on reducing the concentration of various harmful heavy metal ions;

[0016] S5: Calculate the percolation adsorption score of each adsorbent material, and select the adsorbent material with the highest percolation adsorption score as the best adsorption inhibition material for the mineral pile.

[0017] Furthermore, the sampling and determination methods for the concentration of common elements in the mineral stockpile include:

[0018] S11: Weigh 0.01 g of dry mineral bulk powder sample into a conical flask and add 30 mL of 6 mg·L -1 of hydrochloric acid solution and shake to dissolve completely;

[0019] S12: The solution was transferred into a 50 mL colorimetric tube, fixed to volume with ultrapure water solution and shaken well, and the concentration of common elements in the mineral pile was determined using a fiber optic spectrometer.

[0020] Furthermore, the analysis method of a single test unit in step S4 includes:

[0021] S41: Select any harmful heavy metal ion concentration in the water chemical index as an analysis index;

[0022] S42: selecting a water chemistry index with the same change trend as the analysis index as a positive correlation index, and selecting a water chemistry index with the opposite change trend as the analysis index as a negative correlation index;

[0023] S43: respectively calculating the positive covariance cov1 between the analysis index of the leachate in the front section hose and the positive correlation index, the negative covariance cov2 between the analysis index of the leachate in the front section hose and the negative correlation index, the positive covariance cov1 between the analysis index of the leachate in the rear section hose and the positive correlation index, and the negative covariance cov2 between the analysis index of the leachate in the rear section hose and the negative correlation index;

[0024] S44: If the positive covariance cov1 is less than the positive covariance cov1 and the negative covariance cov2 is greater than the negative covariance cov2, the adsorption material in the test unit has an adsorption inhibition effect on the percolation of the analysis index, and step S43 is executed; otherwise, the adsorption material in the test unit has no adsorption inhibition effect on the percolation of the analysis index, and the analysis is terminated;

[0025] S45: Construct a mathematical calculation model to reflect the degree of change in the concentration of harmful heavy metal ions:

[0026]

[0027] In the formula, is the degree of reduction of the concentration of harmful heavy metal ions; i represents the seepage section; k is the adjustment value of the exponential function; cov1 is the positive covariance; cov2 is the negative covariance; x is the preset constant;

[0028] S46: reselect the analysis index and return to step S42 until the degree to which the adsorption material in the test unit reduces the concentration of various harmful heavy metal ions is obtained.

[0029] Furthermore, the calculation method of the percolation adsorption score of each adsorbent material is: multiply the degree of reduction in the concentration of various harmful heavy metal ions in the adsorbent material by the corresponding weight value of the harmful heavy metal ions and then sum them up to obtain the percolation adsorption score of the adsorbent material; wherein the weight value is set according to the importance of the harmful heavy metal ions.

[0030] Furthermore, the calculation formula for the concentration of harmful heavy metal ions obtained by the fiber optic spectrometer is:

[0031]

[0032] A=εcl

[0033] Among them, A is the absorbance, I0 is the intensity of the incident light, I is the intensity of the transmitted light, ε is the molar absorption coefficient of the substance, c is the concentration of harmful heavy metal ions, and l is the optical path length of the sample cell.

[0034] The beneficial effects of the present invention are:

[0035] 1. This scheme enables a single mineral or multiple minerals to undergo continuous or different orders of water-rock chemical reactions at the same time, and predicts the changes in parameters such as the leaching rate, pH, Eh, Pe, and ionic strength value of harmful metal ions in mineral piles at different stages over time, so as to facilitate the subsequent reduction of harmful elements in minerals; at the same time, the monitoring device of this scheme is easy to disassemble and assemble, low in cost, and low in later maintenance cost. The evaluation method of this scheme is simple to operate and has high experimental efficiency; at the same time, this scheme provides an effective evaluation method for soil pollution problems in high background areas of heavy metals; it can significantly reduce the concentration of harmful heavy metal elements in the soil; this process aims to control the content of various harmful heavy metal elements in the soil within the national standard value, thereby improving soil quality and protecting the ecological environment and human health; this evaluation method is not only simple to operate, safe and efficient, but also can provide important technical support and theoretical basis for soil remediation in related fields.

[0036] 2. This solution wraps an insulation layer on the column leaching container and installs a temperature sensor to monitor the temperature change during the reaction in real time, thereby monitoring the process of water chemical thermodynamics; a pH meter, Eh meter, dissolved oxygen meter and conductivity meter are added to the collection bottle to monitor the water chemical reaction process of the leaching solution in real time and predict the changes in parameters such as pH, Eh, Pe, and ion strength values ​​in the mineral pile over time; and a fiber optic spectrometer is set at each seepage section. The light beam released by the solution in each seepage section is absorbed by the optical fiber spectrometer through the emission light source, and the optical fiber spectrometer converts the optical signal into an electrical signal, so that the ion concentration and properties in the collected solution can be directly measured. There is no need to send the solution in the collection bottle for inspection to obtain the target ion concentration, which greatly reduces the workload and greatly improves the work efficiency; finally, through the evaluation system of the evaluation method, the selection of the adsorption material that best reduces the concentration of harmful heavy metal ions is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 Schematic diagram of the structure of the experimental device for heavy metal pollution remediation.

[0038] Figure 2 It is a schematic diagram of the structure of the column immersion container.

[0039] Figure 3 Schematic diagram of the structure of the collecting bottle.

[0040] Figure 4 Schematic diagram of the structure of the cuvette holder.

[0041] Figure 5 Schematic diagram of the internal structure of the fiber optic spectrometer.

[0042] Figure 6 Flowchart for deep learning detection of harmful heavy metal elements in soil using fiber optic spectrometer.

[0043] Among them, 1. column immersion container, 2. hose, 3. peristaltic pump, 4. water outlet bucket, 5. water inlet bucket, 6. temperature sensor, 7. collecting bottle, 8. tee pipe, 9. branch pipe, 10. cuvette holder, 11. optical fiber, 12. optical fiber spectrometer, 13. computer, 14. support frame, 15. plexiglass column tube, 16. flange, 17. butterfly bolt, 18. orifice plate, 19. sealing cover, 20. rubber pad, 20. insulation layer, 21. quartz sand layer, 22. dissolved oxygen meter, 23. pH meter, 24. Eh meter, 25. conductivity meter, 26. mineral pile, 27. light inlet, 28. camera, 29. diffraction grating, 30. cover plate. DETAILED DESCRIPTION

[0044] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0045] like Figures 1 to 4 As shown, the soil heavy metal monitoring device based on spectral deep learning in this scheme includes: a plurality of column immersion containers 1, a plurality of collection bottles 7, a plurality of cuvette holders 10 and a support frame 14, and the support frame 14 is a multi-layer frame structure for integrating and installing various test components.

[0046] A plurality of column immersion containers 1 are connected in series in sequence through a hose 2 to form a percolation passage, the front end of the percolation passage is connected to a water outlet bucket 4 filled with deionized water through a peristaltic pump 3, and the rear end of the percolation passage is connected to a water inlet bucket 5; specifically, the column immersion container 1 includes an organic glass column tube 15, both ends of the organic glass column tube 15 are provided with flanges 16, the flange 16 is connected to a orifice plate 18 and a sealing cover 19 in sequence through a butterfly bolt 17, and rubber pads 20 are provided between the flange 16 and the orifice plate 18, and between the orifice plate 18 and the sealing cover 19, the two ends of the hose 2 are respectively arranged on the sealing covers 19 of two adjacent column immersion containers 1, the organic glass column tube 15 is provided with an insulation layer 21, and the temperature sensor 6 is located between the organic glass column tube 15 and the insulation layer 21; both ends of the organic glass column tube 15 are filled with quartz sand layers 22, and the organic glass column tube 15 located between the quartz sand layers 22 is filled with mineral pile 27 or adsorption material.

[0047] A branch pipe 9 is led out from the hose 2 between two adjacent column leaching containers 1 through a tee pipe 8, and several branch pipes 9 are respectively connected to several collecting bottles 7, on which an electrochemical detection component for detecting the leaching solution is provided; the electrochemical detection component may include a dissolved oxygen meter 23, a pH meter 24, an Eh meter 25 and a conductivity meter 26 electrically connected to the computer 13.

[0048] A plurality of cuvette holders 10 are respectively arranged on a plurality of hoses 2, each cuvette holder 10 is connected to a fiber optic spectrometer 12 via an optical fiber 11, and a plurality of fiber optic spectrometers 12 are electrically connected to a computer 13, and the computer 13 displays the pH, Eh, Pe, ion strength values ​​and changes in ion concentration in the collected solution in real time.

[0049] This scheme also provides an evaluation method for a soil heavy metal monitoring device based on spectral deep learning, which includes the following steps:

[0050] S1: Select soil from a high background area of ​​heavy metals as the mineral pile 27, and sample and measure the concentrations of common elements in the mineral pile 27; specifically, including:

[0051] S11: Weigh 0.01 g of dry mineral pile 27 powder sample into a conical flask, and add 30 mL of 6 mg·L -1 of hydrochloric acid solution and shake to dissolve completely;

[0052] S12: The solution is transferred into a 50 mL colorimetric tube, fixed to volume with ultrapure aqueous solution and shaken well, and the concentration of common elements in the mineral pile 27 is measured using a fiber optic spectrometer 12.

[0053] S2: sequentially assembling two adjacent column leaching containers 1 from among the plurality of column leaching containers 1 into a test unit, and respectively filling the mineral pile 27 into the column leaching container 1 at the front end of each test unit, and respectively filling different types of adsorption materials into the column leaching container 1 at the rear end of each test unit;

[0054] When the mineral pile 27 is loaded into the column, in order to prevent the orifice plate 18 of the column immersion container 1 from being blocked, quartz sand of 50 mm in length is respectively loaded into the liquid inlet and liquid outlet of the column immersion container 1. The quartz sand will not react with the leaching liquid, which is conducive to controlling variables. In specific implementation, firstly, 50 mm of quartz is loaded into the liquid inlet, and then the prepared mineral pile 27 samples are filled one by one. Each time a portion is added, a rammer is used to compact each layer three times. The purpose is to make the mineral pile 27 in the column immersion container 1 as uniform as possible, and finally 50 mm of quartz is loaded into the liquid outlet. In order to ensure the uniformity of the porosity of the mineral pile 27 during the column loading process, the column is loaded by a segmented filling method. Specifically, the required ore sample weight is calculated according to the product of the column volume and the bulk density of the ore layer, and the column is divided into 10 sections for loading. Then, the flange 16, rubber pad 20, orifice plate 18, rubber pad 20, and sealing cover 19 are connected in sequence with butterfly bolts to ensure that the entire device is watertight.

[0055] S3: driving the deionized water to penetrate the plurality of column immersion containers 1 in sequence along the infiltration path through the peristaltic pump 3, and during the infiltration test, monitoring the temperature of the plurality of column immersion containers 1 in real time through the temperature sensor 6, and monitoring the plurality of water chemical indices of the leachate in the plurality of hoses 2 in real time through the electrochemical detection component and the optical fiber spectrometer 12, wherein the plurality of water chemical indices include the electrochemical parameters of the leachate in each collection bottle 7 and the concentration of various harmful heavy metal ions in the leachate in each section of the hose 2; and after the infiltration test is completed, sampling and measuring the concentration of common elements in the mineral stockpile 27 again;

[0056] In specific implementation, the reaction time of the solution and the mineral pile 27 can be controlled by changing the rotation speed of the peristaltic pump 3, and a sufficient contact reaction can be generated therewith. According to the test requirements, the number of column leaching containers 1 and the type of filler in the column leaching container 1 can be arbitrarily changed; during the test, since an oxidation reaction occurs in the column leaching container 1, the temperature change of the reaction can be observed by the temperature sensor 6; an electrochemical detection component is placed inside the collecting bottle 7, which can detect changes in parameters such as pH, Eh, and dissolved oxygen concentration of the leaching solution to monitor the water chemical reaction process of the leaching solution in real time; finally, the optical fiber spectrometer 12 can be used to analyze the light absorption or transmission of different harmful heavy metal ions at specific wavelengths to obtain information such as the chemical composition and concentration of the harmful heavy metal ions;

[0057] like Figure 5As shown, the fiber optic spectrometer 12 includes a light inlet 27, a camera 28, a diffraction grating 29 and a cover plate 30; the working principle of the fiber optic spectrometer 12 is: the cuvette holder 10 is set at the hose 2 of each seepage section, and in order to improve the resolution, the optical aperture of the fiber optic spectrometer 12 needs to be reduced; for this purpose, the optical fiber 11 can be installed. Its opening is smaller and the light beam is more focused; when the solution flows through the cuvette holder 10, the light source is used to pass through the solution and then transmitted to the fiber optic spectrometer 12 through the optical fiber 11. The fiber optic spectrometer 12 is composed of a diffraction grating and a camera; the grating is composed of many fine lines, which disperses light by diffraction, and its function is to disperse the light transmitted from the optical fiber 11 into spectra of different wavelengths; the kit also includes a transparent foil with a microscopic periodic structure, which acts like a prism; the camera then detects the dispersed spectrum and converts it into an electrical signal and transmits it to the computer 13; the computer 13 then processes, displays and analyzes the captured signal; thereby, a spectrum diagram of each heavy metal ion can be obtained, showing the relationship between the intensity of light and the wavelength; thereby, according to the dynamic changes in the wavelengths of different heavy metal elements in the fiber optic spectrometer 12, the change process of the concentration of these heavy metal elements can be obtained; wherein, the calculation formula for the concentration of harmful heavy metal ions obtained by the fiber optic spectrometer 12 is:

[0058]

[0059] A=εcl

[0060] Among them, A is the absorbance, I0 is the intensity of the incident light, I is the intensity of the transmitted light, ε is the molar absorption coefficient of the substance, c is the concentration of harmful heavy metal ions, and l is the optical path length of the sample cell;

[0061] After the test, the peristaltic pump 3 is stopped, the leaching liquid in the column leaching container 1 is drained and collected, the column leaching container 1 is removed and the mineral pile 27 is collected; samples of the removed mineral pile 27 are sampled from top to bottom at a spacing of 20 mm, and solid-liquid separation is performed, and the liquid and solid content are analyzed respectively; finally, the mineral pile 27 is dried and sent for inspection.

[0062] S4: Analyze the change trends of several water chemical indicators of the leachate in the two sections of the hose 2 in each test unit to obtain the reduction effect of different types of adsorption materials on the concentration of various harmful heavy metal ions; wherein, the analysis method of a single test unit includes:

[0063] S41: Select any harmful heavy metal ion concentration in the water chemical index as an analysis index;

[0064] S42: selecting a water chemistry index with the same change trend as the analysis index as a positive correlation index, and selecting a water chemistry index with the opposite change trend as the analysis index as a negative correlation index;

[0065] Specifically, according to the difference between the analysis index monitored by each fiber optic spectrometer 12 and the change trend of other water chemical indexes, it is judged whether there is the possibility of precipitation or reduction of other harmful heavy metal ions in the column immersion container 1 at each stage; in the test time period, the water inlet area and the water outlet area of ​​the first column immersion container 1 constitute the first columnar seepage section, and the water inlet area and the water outlet area of ​​the second column immersion container 1 constitute the second columnar seepage section, and so on. There is a certain correlation between various water chemical indexes in the aqueous solution. The trend change of one water chemical index can roughly reflect the characteristics of other water chemical indexes; the changeable concentration of harmful heavy metal elements can break the correlation between various water chemical indexes in the water body, thereby causing various The variation trend of each hydrochemical index is deviated, therefore, the solution of each columnar seepage section collected by the test device is used to obtain the possibility of the water body in the column immersion container 1 in the first columnar seepage section being affected, that is, the possibility of the concentration change of harmful heavy metal ions according to the difference in the variation trend of the data of each hydrochemical index; therefore, any hydrochemical index is selected as the analysis index, any hydrochemical index with the same correlation as the analysis index is used as a positive correlation index, and any hydrochemical index with the opposite correlation to the analysis index is used as a negative correlation index; then in each seepage section, the covariance between the analysis index of its stage and the positive correlation index is a positive covariance; the covariance between the analysis index of its stage and the negative correlation index is a negative covariance;

[0066] S43: respectively calculating the positive covariance cov1 between the analysis index of the leachate in the front section hose 2 and the positive correlation index, the negative covariance cov2 between the analysis index of the leachate in the front section hose 2 and the negative correlation index, the positive covariance cov1 between the analysis index of the leachate in the rear section hose 2 and the positive correlation index, and the negative covariance cov2 between the analysis index of the leachate in the rear section hose 2 and the negative correlation index;

[0067] S44: If the positive covariance cov1 is less than the positive covariance cov1 and the negative covariance cov2 is greater than the negative covariance cov2, the adsorption material in the test unit has an adsorption inhibition effect on the percolation of the analysis index, and step S43 is executed; otherwise, the adsorption material in the test unit has no adsorption inhibition effect on the percolation of the analysis index, and the analysis is terminated;

[0068] S45: Construct a mathematical calculation model to reflect the degree of change in the concentration of harmful heavy metal ions:

[0069]

[0070] In the formula, is the degree of reduction of the concentration of harmful heavy metal ions; i represents the seepage section; k is the adjustment value of the exponential function, which takes an empirical value of 0.1 to prevent the exponential function from approaching 0 too early; cov1 is the positive covariance; cov2 is the negative covariance; x is a preset constant, which takes a correction value of 1.86;

[0071] S46: reselecting the analysis index and returning to step S42 until the degree to which the adsorption material in the test unit reduces the concentration of various harmful heavy metal ions is obtained;

[0072] The following is an example of steps S41-S46:

[0073] The column immersion container 1 at the front end of the test unit is filled with soil from a heavy metal geological high background area, and the solution flowing out of the first section is monitored by a fiber optic spectrometer 12 to obtain the wavelength of each heavy metal element and form a spectrum. According to the spectrum, the analysis index Cd can be obtained. 2+ 、As 3+ 、Zn 2+ , Hg + The plasma concentration has the same trend of change, which is increasing, that is, they are all positively correlated indicators, but the pH value is decreasing, which is a negatively correlated indicator; therefore, Cd is selected 2+ As the analysis indicator, 3+ is a positively correlated water chemistry index, and pH is a negatively correlated water chemistry index; at this time, Cd 2+ Positively correlated with water chemical index As 3+ The covariance is cov1, Cd 2+ The covariance of the concentration and the negatively correlated water chemical index pH is cov2; in the first columnar seepage section, the positive covariance cov1 is positive, and the negative covariance cov2 is negative; montmorillonite is loaded into the column immersion container 1 at the rear end of the test unit, and the optical fiber spectrometer 12 is used to monitor the solution flowing out of the second seepage section to obtain the wavelength of each heavy metal element, form a spectrum, and the concentration of the sample can be obtained, and the positive covariance cov1 of the analysis index of the leachate in the rear hose 2 and the positive correlation index and the negative covariance cov2 of the analysis index of the leachate in the rear hose 2 and the negative correlation index are obtained; at this time, combined with the spectrum of the first seepage section and the second seepage section, the positive covariance decreases, the closer it is to 0, and it may be a negative number; the negative covariance increases, the closer it is to 0, and it may be a positive number; then montmorillonite is for the harmful heavy metal element Cd 2+ In addition, when cov1-cov2≥0, and the closer cov1-cov2 is to 0, the worse the correlation between the water chemical indicators is; Cd 2+ The greater the concentration change; when cov1-cov2<0, it means that the correlation between the water chemical indicators is opposite. Combined with the spectrum derived from the fiber optic spectrometer 12, Cd 2+ If the concentration tends to decrease, then the inhibitory effect on harmful heavy metal ions is greater than the inhibitory effect when cov1-cov2≥0, and the degree of reduction in the concentration of harmful heavy metal ions is calculated. Similarly, through multiple groups of serial column filtration tests, the degree to which different adsorption materials reduce the concentration of various harmful heavy metal ions can be obtained.

[0074] S5: Calculate the percolation adsorption score of each adsorbent material, and select the adsorbent material with the highest percolation adsorption score as the best adsorption inhibition material for the mineral pile 27; wherein the percolation adsorption score of each adsorbent material is calculated as follows: multiply the reduction degree of the concentration of various harmful heavy metal ions in the adsorbent material by the corresponding weight value of the harmful heavy metal ions and then sum them up to obtain the percolation adsorption score of the adsorbent material; wherein the weight value is set according to the importance of the harmful heavy metal ions.

[0075] like Figure 6 As shown, this scheme also gives the specific process of fiber optic spectrometer 12 deep learning to detect harmful heavy metal elements in soil:

[0076] A1: Soil sample collection: Collect representative samples from the soil in the geological high background area; when sampling, factors such as geographical location, soil type, and pollution source should be considered to ensure that the collected samples can truly reflect the heavy metal pollution status of the soil in the area; when operating, use professional soil sampling tools to collect according to the predetermined sampling depth and quantity. Record the latitude and longitude, altitude, soil type and other information of the sampling point for subsequent analysis.

[0077] A2: Sample pretreatment: Grind, sieve and digest the collected soil samples to extract heavy metal elements in the soil; specific operations: Grinding: Put the soil sample into a grinder and grind it to a uniform particle size (less than 200 mesh); Sieving: Use a sieve to screen the ground soil to remove larger particles and impurities; Digestion: Put the screened soil sample into a digestion tank, add an appropriate amount of digester (such as nitric acid, hydrochloric acid, etc.), and perform digestion under high temperature and high pressure to extract heavy metal elements in the soil.

[0078] A3: Measurement by fiber optic spectrometer 12: Use fiber optic spectrometer 12 to measure the pretreated soil sample and obtain its spectral data; in specific operation, set the parameters of fiber optic spectrometer 12: according to the characteristics of the soil sample and the measurement requirements, set the wavelength range, scanning speed, integration time and other parameters of fiber optic spectrometer 12; measurement: put the pretreated soil sample into the measurement area of ​​fiber optic spectrometer 12, and start fiber optic spectrometer 12 for measurement. Record the spectral data, including the light absorption or emission intensity at different wavelengths.

[0079] A4: Spectral data preprocessing: cleaning, smoothing and standardizing the acquired spectral data to improve the accuracy and comparability of the data; in specific operations, cleaning: removing noise and outliers in the spectral data, such as spikes, baseline drift, etc.; smoothing: using a smoothing algorithm to process the spectral data to reduce data fluctuations and noise interference; standardization: standardizing the spectral data as needed so that the data of different samples have the same scale.

[0080] A5: Build a deep learning model: Select a suitable deep learning framework (such as TensorFlow or PyTorch), design a neural network structure suitable for the characteristics of spectral data, and determine the hyperparameters of the model; When performing specific operations, select a deep learning framework: In this example, we choose TensorFlow as the deep learning framework because it provides rich APIs and tools to facilitate the construction and training of complex neural network models;

[0081] Design the neural network structure: Input layer: The input layer receives the preprocessed spectral data. Assuming that the wavelength range of the spectral data is 400-800nm, and a data point is recorded every 1nm, the number of neurons in the input layer is 401 (800-400+1); Hidden layer: Design two hidden layers, each containing 128 neurons. The hidden layer uses ReLU (Rectified Linear Unit) as the activation function to increase the nonlinear expression ability of the model; Output layer: The output layer contains multiple neurons, each neuron corresponds to the concentration prediction of a heavy metal element. Assuming that the concentrations of four heavy metal elements, lead (Pb), cadmium (Cd), chromium (Cr), and mercury (Hg), need to be predicted, the number of neurons in the output layer is 4. The output layer uses a linear activation function because concentration prediction is a regression problem;

[0082] Determine the hyperparameters: Learning rate: Select 0.001 as the initial learning rate, and dynamically adjust it according to the performance of the validation set during training; Batch size: Set the batch size to 32, that is, 32 samples are processed simultaneously during each training; Number of iterations: Set the maximum number of iterations to 100, but stop training early when the performance of the validation set no longer improves; Optimizer: Select the Adam optimizer because it can achieve better performance in most cases.

[0083] A6: Model training and validation: Use soil samples with known heavy metal element concentrations and their corresponding spectral data to train the model, and use the validation data set to validate the model to evaluate the model's prediction accuracy and generalization ability. When doing this, prepare the data set: Training set: 80% of the soil samples with known heavy metal element concentrations and their corresponding spectral data are divided into the training set; Validation set: The remaining 20% ​​are divided into the validation set to evaluate the performance of the model during the training process;

[0084] Training model: Input the training set into the deep learning model to start the training process; after each training batch, calculate the loss value (such as mean square error MSE) and accuracy (for regression problems, you can calculate the correlation coefficient or R between the predicted value and the true value). 2 score as a measure of accuracy); adjust model parameters (such as weights and biases) based on the loss value and accuracy to optimize model performance; during the training process, regularly use the validation set to validate the model to evaluate the model's predictive accuracy and generalization ability. If the performance of the validation set no longer improves, stop training early to avoid overfitting;

[0085] Model validation and adjustment: Evaluate the prediction accuracy and generalization ability of the model based on the performance of the validation set. If the model performance is poor, you can consider adjusting the network structure, hyperparameters, or data preprocessing methods; if the model shows signs of overfitting on the validation set (i.e., the training set performance is good, but the validation set performance is poor), you can consider adding regularization terms, using dropout technology, or increasing the amount of training data to alleviate overfitting;

[0086] Save model: Save the trained model to disk for subsequent prediction and analysis.

[0087] A7: Prediction of heavy metal element concentration: Use the trained deep learning model to predict the spectral data of new soil samples and output the predicted heavy metal element concentration value; in specific operation, input the spectral data of new soil samples (solution in column infiltration) into the trained deep learning model for prediction. Output the predicted heavy metal element concentration value, including harmful heavy metal elements such as lead, cadmium, chromium, and mercury.

[0088] A8: Result analysis: Analyze and compare the predicted heavy metal element concentration values ​​to determine whether there are harmful heavy metal elements exceeding the standard in the soil; in specific operations, analyze the results: compare the predicted heavy metal element concentration values ​​with the soil heavy metal pollution standards to determine whether there are harmful heavy metal elements exceeding the standard in the soil.

[0089] In summary, the test device of this scheme and the evaluation system of the test method are intended to screen and evaluate the best new adsorption materials for reducing harmful heavy metal elements in soils with high heavy metal background areas. The test device has been carefully designed to simulate the soil remediation process under different environmental conditions, thereby ensuring the effectiveness and applicability of the selected materials; by using the test device of the present invention, combined with the new adsorption material, the time required to restore the soil in the high heavy metal background area to normal soil can be accurately determined. This process can not only achieve effective remediation of contaminated soil, but also provide a quantitative basis for evaluating the recovery capacity of soil in different regions. Specifically, by monitoring the change curve of the concentration of harmful heavy metal elements in the soil, the remediation speed and effect of soils in different regions under the same conditions can be analyzed.

Claims

1. A soil heavy metal monitoring device based on spectral deep learning, characterized in that: include: A plurality of column immersion containers, wherein the plurality of column immersion containers are connected in series in sequence through a hose to form a percolation path, wherein the front end of the percolation path is connected to a water outlet bucket filled with deionized water through a peristaltic pump, and the rear end of the percolation path is connected to a water inlet bucket, and temperature sensors are provided on the plurality of column immersion containers; A plurality of collecting bottles, wherein the hose between two adjacent column leaching containers leads to a branch pipe through a tee pipe, and the branch pipes are respectively connected to a plurality of collecting bottles, and the collecting bottles are provided with an electrochemical detection component for detecting the leaching solution; A plurality of cuvette holders, which are respectively arranged on a plurality of hoses, each of the cuvette holders is connected to a fiber optic spectrometer via an optical fiber, and a plurality of the fiber optic spectrometers are electrically connected to a computer; The support frame is a multi-layer frame structure used for integrated installation of various test components.

2. The soil heavy metal monitoring device based on spectral deep learning according to claim 1 is characterized in that: The column immersion container includes an organic glass column tube, both ends of the organic glass column tube are provided with flanges, the flanges are connected to the orifice plate and the sealing cover in sequence through butterfly bolts, and rubber pads are provided between the flange and the orifice plate, and between the orifice plate and the sealing cover, the two ends of the hose are respectively arranged on the sealing covers of two adjacent column immersion containers, the organic glass column tube is provided with an insulation layer, and the temperature sensor is located between the organic glass column tube and the insulation layer.

3. The soil heavy metal monitoring device based on spectral deep learning according to claim 1 is characterized in that: Both ends of the organic glass column tube are filled with quartz sand layers, and the organic glass column tube between the quartz sand layers is filled with mineral pile or adsorption material.

4. The soil heavy metal monitoring device based on spectral deep learning according to claim 1 is characterized in that: The electrochemical detection component includes a dissolved oxygen meter, a pH meter, an Eh meter and a conductivity meter electrically connected to a computer.

5. An evaluation method for a soil heavy metal monitoring device based on spectral deep learning according to any one of claims 1 to 4, characterized in that: The following steps are involved: S1: Select soil from the high background area of ​​heavy metals as mineral pile material, and sample and measure the concentration of common elements in the mineral pile material; S2: sequentially assembling two adjacent column leaching containers from a plurality of column leaching containers into a test unit, and respectively filling the mineral pile into the column leaching container at the front end of each test unit, and respectively filling different types of adsorption materials into the column leaching container at the rear end of each test unit; S3: driving deionized water to penetrate several column immersion containers in sequence along the infiltration path through a peristaltic pump, and during the infiltration test, monitoring the temperature of several column immersion containers in real time through a temperature sensor, and monitoring several water chemical indicators of the leachate in several hoses in real time through an electrochemical detection component and an optical fiber spectrometer, wherein the several water chemical indicators include the electrochemical parameters of the leachate in each collection bottle and the concentration of various harmful heavy metal ions in the leachate in each section of the hose; and after the infiltration test, sampling and measuring the concentration of common elements in the mineral stockpile again; S4: Analyze the changing trends of several water chemical indicators of the leachate in the two sections of the hose in each test unit to obtain the effects of different types of adsorption materials on reducing the concentration of various harmful heavy metal ions; S5: Calculate the percolation adsorption score of each adsorbent material, and select the adsorbent material with the highest percolation adsorption score as the best adsorption inhibition material for the mineral pile.

6. The evaluation method of soil heavy metal monitoring device based on spectral deep learning according to claim 5 is characterized in that: Sampling methods for determining the concentration of common elements in mineral stockpiles include: S11: Weigh 0.01 g of dry mineral bulk powder sample into a conical flask and add 30 mL of 6 mg·L -1 of hydrochloric acid solution and shake to dissolve completely; S12: The solution was transferred into a 50 mL colorimetric tube, fixed to volume with ultrapure water solution and shaken well, and the concentration of common elements in the mineral pile was determined using a fiber optic spectrometer.

7. The evaluation method of soil heavy metal monitoring device based on spectral deep learning according to claim 5 is characterized in that: The analysis method of a single test unit in step S4 includes: S41: Select any harmful heavy metal ion concentration in the water chemical index as an analysis index; S42: selecting a water chemistry index with the same change trend as the analysis index as a positive correlation index, and selecting a water chemistry index with the opposite change trend as the analysis index as a negative correlation index; S43: respectively calculating the positive covariance cov1 between the analysis index of the leachate in the front section hose and the positive correlation index, the negative covariance cov2 between the analysis index of the leachate in the front section hose and the negative correlation index, the positive covariance cov1 between the analysis index of the leachate in the rear section hose and the positive correlation index, and the negative covariance cov2 between the analysis index of the leachate in the rear section hose and the negative correlation index; S44: If the positive covariance cov1 is less than the positive covariance cov1 and the negative covariance cov2 is greater than the negative covariance cov2, the adsorption material in the test unit has an adsorption inhibition effect on the percolation of the analysis index, and step S43 is executed; otherwise, the adsorption material in the test unit has no adsorption inhibition effect on the percolation of the analysis index, and the analysis is terminated; S45: Construct a mathematical calculation model to reflect the degree of change in the concentration of harmful heavy metal ions: In the formula, is the degree of reduction of the concentration of harmful heavy metal ions; i represents the seepage section; k is the adjustment value of the exponential function; cov1 is the positive covariance; cov2 is the negative covariance; x is the preset constant; S46: reselect the analysis index and return to step S42 until the degree to which the adsorption material in the test unit reduces the concentration of various harmful heavy metal ions is obtained.

8. The evaluation method of soil heavy metal monitoring device based on spectral deep learning according to claim 7 is characterized in that: The calculation method for the percolation adsorption score of each adsorbent material is as follows: the degree of reduction in the concentration of various harmful heavy metal ions in the adsorbent material is multiplied by the corresponding weight value of the harmful heavy metal ions and then the sum is calculated to obtain the percolation adsorption score of the adsorbent material; wherein the weight value is set according to the importance of the harmful heavy metal ions.

9. The evaluation method of soil heavy metal monitoring device based on spectral deep learning according to claim 7 is characterized in that: The calculation formula of harmful heavy metal ion concentration obtained by fiber optic spectrometer is: A=εcl Among them, A is the absorbance, I0 is the intensity of the incident light, I is the intensity of the transmitted light, ε is the molar absorption coefficient of the substance, c is the concentration of harmful heavy metal ions, and l is the optical path length of the sample cell.

Images