Soil detection system and detection method

Abstract

The invention relates to a soil detection system and a detection method. Soil detection equipment comprises a sampling module, a positioning module for generating original point sampling coordinates, a sample treatment module for chemical, electrochemical and biological treatment means, a detection module, a detection database, a detection control module, a generating and sending module and a sending and receiving module. The detection method comprises the steps of sampling and processing, binding coordinates, comparing abnormal data through preliminary detection, generating a pollution map, carrying out secondary detection on abnormal data, calculating a spatial uncertainty index, dynamically generating a near-origin secondary sampling coordinate, circularly sampling and detecting until the index reaches the standard or the density upper limit, and finally generating the pollution map and a prediction model. And the detection database is updated and optimized through machine learning.

Description

Technical Field

[0001] The present invention relates to the field of soil detection, and in particular to a soil detection system and method. Background Art

[0002] Existing soil pollution detection methods have systemic defects: the fixed grid sampling strategy cannot dynamically respond to the diffusion characteristics of pollutants such as differences in migration rates, resulting in redundant sampling in low-pollution areas and missed sampling at high-gradient pollution boundaries, and the misjudgment rate of pollution range remains high; on-site rapid detection equipment is limited to basic physical and chemical parameter analysis such as pH value and organic matter, and has insufficient ability to identify complex pollution, and relies on laboratory re-inspection, resulting in a lengthy cycle; the existing graded detection scheme uses a fixed threshold to trigger secondary detection, which is difficult to adapt to the synergistic effect of multiple pollutants, and the static setting of the secondary sampling radius ignores the dynamic diffusion law of pollutants, resulting in the loss of key data; more seriously, the sampling coordinates, detection data and pollution maps are stored separately, lacking a real-time binding mechanism and a laboratory verification closed loop, resulting in low accuracy in pollution traceability and diffusion prediction, which urgently needs to be solved through adaptive sampling and intelligent data fusion technology. Summary of the Invention The purpose of this application is to provide a soil detection system and detection method, which have the advantages of dynamically responding to the diffusion characteristics of pollutants, improving the accuracy of complex pollution identification, optimizing sampling strategies and realizing closed-loop verification of detection.

[0003] In order to solve the above technical problems, the present invention is solved by the following technical solutions: A soil detection system, comprising: A sampling module for sampling soil; Positioning module, used to locate the sampling location and generate the origin sampling coordinates; A sample processing module, used to separately store the soil samples obtained by the sampling module and bind the storage location to the origin sampling coordinates; A detection module, comprising a preliminary detection unit and a secondary detection unit, wherein the preliminary detection unit is used to perform preliminary detection on the treated soil sample to obtain preliminary detection data; A detection database storing a plurality of abnormal soil data packets, wherein the abnormal soil data packets record data on abnormal soil including pre-inspection data and re-inspection data; The detection control module is used to match the preliminary detection data with the pre-inspection data of the abnormal soil data packet. When the preliminary detection data is within the pre-inspection data range, the secondary detection unit is triggered to perform a secondary detection on the soil sample at the corresponding storage location, and the advanced detection data obtained by the secondary detection is matched with the re-inspection data to confirm the abnormal concentration value of the soil sample; A sending module is generated to generate a pollution location map based on the origin sampling coordinates and the matching results. The near-origin secondary sampling coordinates are dynamically generated based on the pollution location map. The near-origin secondary sampling coordinates are coordinate points within a close range from the origin sampling coordinates. The close range of the origin sampling coordinates is dynamically adjusted according to the pollution diffusion characteristics, and the sampling module is controlled to sample the secondary coordinate points until the sampling density reaches a preset threshold.

[0004] By adopting the above technical solution, the soil detection equipment solves the problem of inaccurate secondary detection triggering by clarifying the matching threshold between the preliminary detection data and the pre-inspection data, ensuring the accurate start of the secondary detection; independently stores soil samples to ensure the accurate binding of the samples and the origin coordinates; combines the generation of pollution location maps to improve the accuracy of pollution positioning; dynamically adjusts the near-origin secondary sampling range according to the type of pollutant, soil permeability, etc., and optimizes the sampling strategy in combination with the pollution concentration gradient, achieving more targeted sampling, effectively improving the accuracy and efficiency of soil pollution detection, and can determine the pollution location and range more quickly and accurately.

[0005] The present invention is further configured as follows: the preliminary detection unit is used to detect at least one of the hardness, moisture, pH value, organic matter and spectral characteristics of the soil; and the secondary detection unit is used to perform secondary detection on the soil sample to obtain advanced detection data.

[0006] By adopting the above technical solution, the equipment clearly uses the secondary detection unit to obtain advanced detection data, and cooperates with the primary detection unit to form a gradient detection system. The primary detection quickly screens abnormal samples, and the secondary detection deeply analyzes the pollution details. The synergistic effect of the two enhances the layering and depth of soil detection, provides more comprehensive data support for subsequent pollution analysis, and helps to more accurately judge the soil pollution status.

[0007] The present invention is further configured as follows: the sampling module is equipped with a segmented sampling rotating rod group, and the sampling drill rod group is integrated with a soil texture sensor, and the soil texture sensor is used to detect changes in soil resistivity in real time; when the resistivity mutation value exceeds a threshold, it is determined to be a geological stratification interface and segmented sampling is performed.

[0008] By adopting the above technical solution, the sampling module of the equipment integrates soil texture sensors and acoustic wave sensors, and uses multi-sensor fusion technology to determine the geological stratification interface, thereby solving the problem of insufficient reliability of single resistivity detection and improving the accuracy of geological stratification judgment; after segmented sampling, each segmented sample is bound to the origin coordinates to ensure the correspondence of information of soil samples in different strata, improve the scientific nature of sampling and the representativeness of samples, and provide more reliable basic samples for subsequent detection.

[0009] The present invention is further configured such that: the sample processing module is equipped with a processing means adapted to the secondary detection unit, including: Chemical treatment method, which is equipped with a number of chemical reagents, uses the chemical reagents to perform titration tests on soil samples and generate chemical treatment result data; The electrochemical treatment method is configured with an aqueous solution and a test electrode, wherein the soil sample is formed into a solution or suspension by the aqueous solution, and the control electrode generates electrochemical treatment result data; The biological detection method is equipped with several closed chambers for culturing soil samples in a sealed environment to detect carbon dioxide produced or oxygen consumed by microbial respiration and obtain biological activity information.

[0010] By adopting the above technical solution, the three processing methods of the sample processing module are respectively targeted at different detection needs. Chemical treatment is suitable for the quantification of heavy metal ions, electrochemical treatment is suitable for the detection of characteristics such as soluble ions, and biological detection is suitable for microbial activity and organic pollution assessment, covering the detection needs of various pollution types; the data conversion interface is configured to unify the data formats of different methods, solving the data adaptation problem; the temperature control module is set in the closed chamber to ensure a constant temperature and humidity environment for biological detection, ensure the consistency and reliability of biological detection results, and overall improve the effectiveness and accuracy of secondary detection data.

[0011] The present invention is further configured such that: the detection module is further configured to generate pollution degree data, a pollution concentration map and a pollution heat map based on the detection data of the secondary detection unit.

[0012] By adopting the above technical solution, the detection module can generate pollution degree data, pollution concentration maps and pollution heat maps based on the secondary detection data, converting abstract detection data into intuitive visual charts, making the degree, distribution and aggregation of soil pollution clear at a glance, and providing clear visual support for staff to quickly understand the pollution status and analyze pollution trends, which helps to improve the efficiency of pollution analysis and decision-making.

[0013] The present invention is further configured to include: The sending and receiving module establishes an encrypted communication connection with the terminal laboratory and is used to: Sending the advanced detection data and pollution analysis map to the terminal laboratory; Receive precision verification data returned by the terminal laboratory; The detection database is configured with a machine learning update engine for: Compare the feature vector differences between the field test data of the sample to be verified and the precision verification data; When the difference exceeds the preset tolerance threshold, the abnormal soil data package is updated, and the matching threshold range of the pre-inspection data or re-inspection data is adjusted to generate a new abnormal soil data package; the external environmental data and expert knowledge base are integrated to dynamically optimize the feature weight distribution model.

[0014] By adopting the above technical solution, the sending and receiving modules of the device simplify the communication with the terminal laboratory, focus on the core function of data transmission, and avoid unnecessary restrictions on the protection scope of encrypted communication; the machine learning update engine of the detection database makes the update conditions of abnormal soil data packets clearer by clarifying the feature vector construction method, and optimizes the feature weight distribution model by combining external environmental data and expert knowledge base, continuously improving the accuracy and adaptability of the database, and enhancing the equipment's detection capabilities for different soil pollution conditions.

[0015] A soil testing method comprises the following steps: S1: Control the sampling module to collect soil samples at a specified depth at the target location; S2: The soil samples are crushed, homogenized and stored separately through the sample processing module; S3: Determine the origin sampling coordinates of the sampling location through the positioning module and bind them to the sample storage location; S4: Performing preliminary testing on the soil sample through the preliminary testing unit of the testing module to obtain preliminary testing data; S5: Compare the preliminary test data with the abnormal soil data package in the test database to determine whether it is consistent with the pre-test data; S6: Generate a contamination location map based on the comparison results and the origin sampling coordinates; if the preliminary detection data is consistent with the pre-inspection data, start the secondary detection unit; the contamination location map is generated by marking the sampling coordinates of the matching abnormal data as contamination points; S7: Generate a suspected pollution diffusion map based on the secondary detection results and calculate the spatial uncertainty index of the current pollution area. The spatial uncertainty index is calculated based on the spatial distribution density of the sampling points, the variance of the detection data, and the gradient change rate of the pollution concentration. The gradient weight is dynamically adjusted according to the pollution diffusion direction. S8: When the spatial uncertainty index is greater than the preset threshold, the following loop steps are executed: S801: Based on the pollution concentration gradient map and the spatial uncertainty index, generate near-origin secondary sampling coordinates, where the near-origin secondary sampling coordinates are coordinate points within a close range from the origin sampling coordinates, and the close range is dynamically adjusted according to pollution diffusion characteristics; S802: Control the sampling module to sample the secondary sampling coordinates near the origin; S803: Perform preliminary testing, secondary testing, update the pollution concentration gradient map, and recalculate the spatial uncertainty index on the new sample in sequence; S9: When the spatial uncertainty index is less than or equal to the preset threshold or the sampling density reaches the preset upper limit, exit the loop; S10: Generate the final pollution location map, pollution heat map and pollution diffusion prediction model.

[0016] By adopting the above technical solutions, the soil detection method makes the uncertainty assessment of the contaminated area more accurate by clarifying the spatial uncertainty index; clarifies the adjustment rules of the close range and the priority of the exit conditions in the cycle step, avoids the risk of dead loops, and ensures an orderly and efficient detection process; and can gradually reduce spatial uncertainty by dynamically adjusting the sampling strategy and cyclic detection. The resulting pollution location map, heat map and prediction model are more comprehensive and accurate, improving the scientific nature and reliability of soil pollution detection.

[0017] The present invention is further configured such that when the preliminary detection data is consistent with the pre-detection data, the secondary detection unit adopts at least one of the following methods: Chemical treatment method: soil samples are titrated using chemical reagents to generate chemical treatment result data; Electrochemical treatment method: The soil sample is converted into a solution or suspension through an aqueous solution, and the control electrode generates the electrochemical treatment result data; Biological detection method: culturing soil samples in the closed chamber, detecting carbon dioxide produced or oxygen consumed by microbial respiration, and obtaining biological activity information; The application scenarios of the chemical treatment method, electrochemical treatment method and biological detection method include: Chemical treatment methods are suitable for detecting the quantitative composition of heavy metal ions in soil through chemical reactions; The electrochemical treatment method is suitable for detecting the concentration of soluble ions and electrochemical characteristics of redox potential in soil; Biological detection methods are suitable for evaluating soil microbial activity and the degree of organic pollution.

[0018] By adopting the above technical solution, the method clarifies the chemical, electrochemical and biological detection methods used by the secondary detection unit, and corresponds to different applicable scenarios, so that the selection of detection methods is more targeted, and the most appropriate detection method can be selected according to the type of pollutants in the soil; at the same time, it is clarified that the biological detection uses the closed chamber described in claim 4, which ensures the stability of the biological detection environment, improves the accuracy and effectiveness of the secondary detection results, and provides a reliable basis for accurately judging the type and degree of pollution.

[0019] The present invention is further configured to: generate the suspected pollution diffusion map in step S7 by obtaining the soil permeability, porosity and moisture content of the sampling points through the preliminary detection unit; Based on the fluid mechanics model and in combination with the soil physical parameters, the lateral / vertical migration rate of pollutants is calculated; Access to real-time meteorological data to dynamically correct migration rate and diffusion direction; Generate a pollution migration time series map, marking the spatial and temporal boundaries of pollutant diffusion within a preset time period; The pollution migration time series map is connected to the secondary sampling. When the time series map shows that the pollutants are migrating in a certain direction at an accelerated rate, the secondary sampling coordinates near the origin are added to the migration path; the time series map is iteratively updated based on the new sampling point data until the prediction converges.

[0020] By adopting the above technical solution, when generating a suspected pollution diffusion map, this method calculates the migration rate through Darcy's law combined with the convection-diffusion equation, and accesses real-time meteorological data for correction, thereby improving the accuracy of the calculation of pollutant migration rate and diffusion direction; generating a pollution migration time series map and combining it with secondary sampling dynamic update, it can timely track the migration trend of pollutants and add new sampling points on the migration path, making the pollution diffusion analysis more in line with the actual situation, and improving the scientific nature and timeliness of pollution diffusion prediction.

[0021] The present invention is further configured to include the following steps: S11: Submit the sample set to be verified to the terminal laboratory, including samples whose spatial uncertainty index is higher than the preset risk threshold; and samples whose detection data matches the historical abnormal data packets in the fuzzy range; S12: Based on the precise verification data, if significant deviations are confirmed in the on-site testing, the abnormal soil data package is expanded or corrected; If the laboratory denies the existence of contamination, a counterexample training set is generated to optimize the machine learning model; Associate external environmental data to build a regional pollution risk profile.

[0022] By adopting the above technical solution, this method ensures the representativeness of the samples to be verified by submitting a specific set of samples to be verified to the terminal laboratory, which helps to correct abnormal soil data packets through precise verification data and improve the credibility of the detection data; if the laboratory denies the existence of pollution, a counterexample training set is generated, which can continuously optimize the machine learning model and improve the model's discrimination ability; external environmental data is associated to construct a regional pollution risk portrait, which provides comprehensive support for the overall assessment and prevention and control of regional soil pollution and enhances the practicality and scalability of the detection method.

[0023] Due to the adoption of the above technical solution, the present invention has significant technical effects: the soil detection system and method provided by the present application effectively solves the problems of sampling redundancy, missed sampling and low detection accuracy in the existing technology by dynamically generating secondary sampling coordinates near the origin and adjusting the sampling range according to the pollution diffusion characteristics, combining multi-level detection with a machine learning update mechanism, and has the advantages of dynamically responding to the diffusion characteristics of pollutants, improving the accuracy of complex pollution identification, optimizing sampling strategies and realizing closed-loop verification of detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a structural framework diagram of a soil detection system. DETAILED DESCRIPTION

[0025] The present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0026] Example: In the existing technology, soil pollution detection methods have systematic defects. The fixed grid sampling strategy cannot dynamically respond to the diffusion characteristics of pollutants, resulting in redundant sampling in low-pollution areas and missed sampling at high-gradient pollution boundaries. The misjudgment rate of pollution range remains high. On-site rapid detection equipment is limited to basic physical and chemical parameter analysis, and its ability to identify complex pollution is insufficient. Reliance on laboratory re-inspection leads to lengthy cycles. The existing graded detection scheme uses a fixed threshold to trigger secondary detection, which is difficult to adapt to the synergistic effect of multiple pollutants. The static setting of the secondary sampling radius ignores the dynamic diffusion law of pollutants, resulting in the loss of key data. Sampling coordinates, detection data and pollution maps are stored separately, and there is a lack of real-time binding mechanism and laboratory verification closed loop, resulting in low accuracy in pollution tracing and diffusion prediction.

[0027] The inventors discovered that the core problem of the existing technology is that the sampling strategy does not match the dynamic diffusion characteristics of pollutants, and the detection process lacks an intelligent triggering mechanism. By analyzing the relationship between the migration rate of pollutants and their spatial distribution, they proposed an idea of ​​dynamically adjusting the secondary sampling density. To address the problem of detection data fragmentation, they designed a coordinate and storage location binding mechanism. To solve the laboratory verification delay, they constructed a closed-loop process in which abnormal data packets are matched to trigger secondary detection. They combined on-site detection with database intelligent analysis to achieve adaptive sampling and data fusion.

[0028] This application proposes a soil detection device including a sampling module, a positioning module, a sample processing module, a detection module, a detection database, a detection control module and a generation and sending module. The sampling module is used to collect soil samples, the positioning module generates the origin sampling coordinates, the sample processing module binds the storage location with the coordinates, the detection module includes a preliminary detection unit and a secondary detection unit, the detection database stores abnormal soil data packets, the detection control module triggers secondary detection by matching preliminary detection data with pre-inspection data, and the generation and sending module dynamically generates near-origin secondary sampling coordinates based on the pollution location map and controls the sampling density.

[0029] The sampling module refers to a component that obtains soil samples through mechanical devices. Specifically, it can be implemented by an electric sampler with a drill bit or a spiral drill rod, and is used to collect soil samples at different depths. The positioning module refers to a device for obtaining geographic coordinates. Specifically, it can be implemented by a GPS positioning chip or a Beidou navigation system, and is used to record the latitude and longitude information of the sampling point. The sample processing module refers to a unit for storing and managing samples. Specifically, it can be implemented by a sample storage box with an independent code in combination with a QR code scanning device to ensure that each sample corresponds to the coordinate information one by one. The detection module includes a preliminary detection unit and a secondary detection unit. The preliminary detection unit can use a multi-parameter sensor integrated probe to realize basic physical and chemical index detection. The secondary detection unit can be configured with a spectrometer or a chromatograph for in-depth analysis. The detection database refers to a storage system for storing abnormal data features. Specifically, it can be implemented by a relational database or a time series database, and is used to record the matching rules between pre-inspection data and re-inspection data. The generation and sending module refers to a logical unit that generates pollution maps and controls sampling. Specifically, it can be implemented by using geographic information system software combined with a dynamic path planning algorithm, and adjusts the secondary sampling range according to the pollution diffusion model.

[0030] After the sampling module collects soil samples in the target area, the positioning module synchronously records the geographic coordinates of the sampling points. The sample processing module stores the samples in an independent storage unit and associates the storage location with the coordinates through data binding technology. The preliminary detection unit quickly detects the samples. The detection control module compares the results with the pre-inspection data in the database. When the data matches, the secondary detection unit is triggered to perform in-depth analysis and generate advanced detection data. The generation and sending module dynamically calculates the secondary sampling range according to the pollution diffusion characteristics, generates new sampling point coordinates near the origin coordinates, and gradually increases the sampling density through cyclic sampling and detection until the preset pollution boundary identification accuracy is reached.

[0031] The traditional method uses a fixed grid distribution, which results in a mismatch between the distribution of sampling points and the actual diffusion direction of pollutants. However, this solution dynamically generates secondary sampling coordinates near the origin, which can adjust the sampling density in real time according to changes in the pollution concentration gradient. The existing technology relies on fixed thresholds to trigger re-inspections and cannot adapt to the synergistic effects of multiple pollutants. This solution uses a dual matching mechanism of pre-inspection data and re-inspection data, combined with machine learning models to optimize matching rules, significantly improving the ability to identify complex pollution. In traditional equipment, detection data and sampling coordinates are stored separately. This solution binds the storage location and geographic coordinates to achieve integrated management of detection data and spatial location, providing an accurate data foundation for pollution diffusion modeling.

[0032] This application effectively solves the problem of pollution boundary identification error caused by fixed sampling strategies. By dynamically adjusting the secondary sampling range, the detection coverage of high-gradient pollution areas is significantly improved. The dual detection mechanism shortens the laboratory re-inspection cycle. The intelligent matching of abnormal data packets reduces the number of invalid detections. The real-time binding of coordinates and data provides accurate spatial reference for pollution tracing. The dynamic sampling density control strategy reduces the overall sampling cost while ensuring detection accuracy.

[0033] The primary detection unit is used to detect at least one of the hardness, moisture, pH value, organic matter and spectral characteristics of the soil; the secondary detection unit is used to perform secondary detection on the soil sample to obtain advanced detection data.

[0034] Hardness refers to the bonding strength between soil particles, which can be achieved by using an indentation hardness tester or a shear force sensor. The hardness data is obtained by measuring the resistance of the probe to penetrate the soil. Humidity refers to the moisture content of the soil, which can be achieved by using a capacitive moisture sensor or a near-infrared spectrometer. The moisture content is calculated by detecting changes in the dielectric constant or the absorbance at a specific wavelength. pH value refers to the acidity and alkalinity of the soil, which can be achieved by using a glass electrode pH meter or an ion-sensitive field-effect transistor. The pH data is generated by measuring the hydrogen ion concentration. Organic matter refers to the total amount of carbon-containing organic matter in the soil. It can be achieved by using the loss on ignition method or an ultraviolet-visible spectrophotometer. The mass loss or color intensity is detected after high-temperature decomposition or chemical oxidation reaction. Spectral characteristics refer to the optical response of the soil in a specific band. It can be achieved by using a multispectral imager or a Raman spectrometer. The characteristic peaks of pollutants are identified by analyzing the reflectivity or scattering spectrum.

[0035] The preliminary detection unit uses a multi-parameter combination detection mode to simultaneously collect physical, chemical and optical property data of the soil. When an abnormal increase in hardness is detected, it can be associated with humidity data to determine whether there is compaction; when the pH value deviates from the normal range, the cause of the acid-base imbalance is analyzed in combination with the organic matter content; spectral feature detection can capture the characteristic absorption peaks of heavy metal ions or organic pollutants. The secondary detection unit selects targeted detection methods based on the abnormal indicators found in the preliminary detection. For example, when the preliminary detection shows abnormal organic matter, gas chromatography-mass spectrometry is used to analyze the organic composition; when the spectral characteristics indicate heavy metal pollution, atomic absorption spectrometry is used to determine the metal ion concentration.

[0036] Existing soil testing equipment usually only detects a single physical and chemical parameter, such as only measuring pH or humidity, and is unable to build multi-dimensional data associations. This solution integrates hardness, humidity, pH, organic matter and spectral feature detection functions to form a parameter complementary verification mechanism. For example, when the spectral characteristics show the presence of a pollutant but the pH value does not exceed the standard, the secondary detection unit can be triggered to perform pollutant morphological analysis to avoid misjudgment of a single parameter. The fixed combination of detection items in existing technologies cannot adapt to complex pollution scenarios. This solution dynamically selects a combination of detection items, for example, prioritizing spectral feature detection in industrial pollution areas and focusing on organic matter and pH value detection in agricultural pollution areas, to achieve adaptive adjustment of detection strategies.

[0037] The present application can effectively identify cross-interference phenomena in complex pollution scenarios. For example, when heavy metal pollution and organic pollution coexist, soil structure damage can be detected through hardness testing, and the characteristic peak of chromium ions can be identified by combining spectral characteristics. The hexavalent chromium concentration is then confirmed through secondary testing to avoid missed detections caused by a single test item. At the same time, multi-parameter preliminary testing can reduce the number of unnecessary secondary tests. For example, when the pH value and organic matter are both within the normal range, the possibility of acidic organic pollution can be directly ruled out. Compared with the fixed trigger secondary detection mode in the existing technology, the detection efficiency is significantly improved.

[0038] The sampling module is equipped with a segmented sampling rod group, and the sampling drill rod group is integrated with a soil texture sensor, which is used to detect changes in soil resistivity in real time. When the resistivity mutation value exceeds the threshold, it is determined to be a geological stratification interface and segmented sampling is performed.

[0039] The segmented sampling rod assembly is a sampling device consisting of multiple detachable drill rod segments, which can be implemented using a threaded connection or a snap-on structure. Each drill rod segment is equipped with an independent sampling chamber to store soil samples at the corresponding depth. The soil texture sensor is a detection device embedded in the front end of the drill rod, which can be implemented using a four-electrode resistivity measurement module. The resistivity value is calculated by applying current to the soil and measuring the voltage difference. The resistivity mutation value refers to the magnitude of the resistivity change between adjacent sampling points. Specifically, a threshold value, such as 200Ω·m, can be set as the judgment basis. When the change exceeds this threshold, a control signal is triggered. The geological stratification interface is the boundary between soil layers with different physical properties, such as the junction of a clay layer and a gravel layer. The resistivity mutation can be used to identify the structural differences between the layers. Segmented sampling automatically separates the current drill rod segment when a stratification interface is detected, retaining the soil sample in that layer and replacing it with a new drill rod segment to continue sampling the next layer.

[0040] During the drilling process, the soil texture sensor continuously collects resistivity data. When the drill bit passes through soil layers of different textures, the resistivity value changes suddenly, for example, from a low-resistivity clay layer to a high-resistivity gravel layer. The control system compares the resistivity difference between adjacent sampling points in real time. If it exceeds the preset threshold, it is determined that the geological stratification interface has been reached. At this time, the corresponding detachable section in the drill rod group is automatically detached, and the soil sample of this layer is sealed and stored in an independent sampling chamber. At the same time, the subsequent drill rod sections continue to sample downward. By separating and sampling layer by layer, the mixing of samples from different soil layers is avoided, ensuring the correspondence between the detection data and the soil stratification.

[0041] Traditional sampling equipment uses fixed-length drill rods for continuous sampling, which cannot identify the soil layer structure. This leads to the mixing of samples from different soil layers, and the test data cannot accurately reflect the pollution status of each layer. This solution realizes automatic identification of layered interfaces through resistivity mutation detection. Combined with a segmented drill rod group, it realizes independent sampling by soil layer, solving the problem of data distortion caused by mixed sampling.

[0042] This application can accurately distinguish different geological layers and obtain independent layered samples, avoiding the obscuration of the diffusion paths of pollutants in different soil layers, thereby improving the accuracy of pollution detection. The layered sampling data can be bound to the positioning coordinates to provide reliable basic data support for the establishment of a three-dimensional pollution diffusion model.

[0043] The sample processing module is equipped with processing means compatible with the secondary detection unit, including a chemical treatment method, which is equipped with a number of chemical reagents, and the soil sample is titrated by chemical reagents to generate chemical treatment result data; an electrochemical treatment method, which is equipped with an aqueous solution and a test electrode, and the soil sample is formed into a solution or suspension by the aqueous solution, and the electrode is controlled to generate electrochemical treatment result data; a biological detection method, which is equipped with a number of closed chambers, and is used to culture soil samples in a sealed environment to detect carbon dioxide produced or oxygen consumed by microbial respiration and obtain biological activity information.

[0044] The chemical treatment method refers to the use of chemical reagents of preset concentrations to produce a color development or precipitation reaction with soil samples. Specifically, an automatic titrator combined with a spectral analysis module can be used to achieve quantitative detection, which is used to quickly identify the concentration of heavy metal ions. The electrochemical treatment method refers to dissolving the soil sample and measuring the redox potential or ion mobility through electrodes. Specifically, a three-electrode system can be used in conjunction with a constant potentiostat to analyze the electrochemical properties of soluble pollutants. The biological detection method refers to the use of the correlation between microbial metabolic activity and pollutant concentration. Specifically, a closed culture chamber with an integrated gas sensor can be used to monitor the oxygen consumption rate to evaluate the biodegradation activity of organic pollutants.

[0045] When the secondary detection unit is triggered, the sample processing module selects an appropriate treatment method according to the preset program. For example, for samples suspected of heavy metal contamination, the chemical treatment method injects a chelating agent through an automatic titration device and determines the ion concentration based on the reaction endpoint; for soluble salt pollution, the electrochemical treatment method introduces the soil suspension into the electrolytic cell and determines the characteristic peak potential through cyclic voltammetry; if organic pollution needs to be assessed, the biological detection method places the sample in a constant temperature incubation chamber and continuously monitors the carbon dioxide release through an infrared sensor. The data of the three treatment methods are transmitted to the detection control module through a unified interface to achieve multi-dimensional pollution feature cross-validation.

[0046] Traditional soil testing equipment is usually equipped with only a single chemical detection method, which cannot cover the synergistic analysis of heavy metals, soluble salts and organic pollutants. This solution integrates chemical, electrochemical and biological detection methods to build a composite pollution identification system. It can simultaneously complete ion concentration measurement, redox characteristics analysis and microbial activity detection on site, avoiding the problem of missed detection due to a single detection method.

[0047] This application solves the problem of insufficient identification ability of complex pollutants in existing technologies, and realizes the simultaneous rapid detection of heavy metal ions, soluble salts and organic pollutants. The chemical treatment method ensures the quantitative accuracy of heavy metal ions, the electrochemical treatment method improves the sensitivity of soluble salt detection, and the biological detection method effectively captures the biological activity characteristics of organic pollutants. The coordinated work of the three significantly shortens the laboratory re-inspection cycle and provides multi-dimensional data support for the precise positioning of pollution sources.

[0048] The detection module is also used to generate pollution degree data, pollution concentration map and pollution heat map based on the detection data of the secondary detection unit.

[0049] Pollution degree data refers to the results of quantitative analysis of the concentration, distribution range and diffusion trend of pollutants through the advanced detection data of the secondary detection unit. Specifically, it can be achieved by using a numerical grading algorithm combined with a pollutant toxicity equivalent calculation model, and is used to distinguish the pollution hazard levels in different regions. The pollution concentration map refers to a spatial distribution visualization map generated based on sampling coordinates and secondary detection data. Specifically, it can be achieved by using a geographic information system interpolation algorithm combined with a pollution concentration gradient calculation model, and is used to mark the differences in pollutant concentrations in different geographical locations. The pollution heat map refers to a graphical expression that reflects the spatial changes in pollution intensity through color gradient mapping technology. Specifically, it can be achieved by using a kernel density estimation algorithm combined with dynamic rendering technology, and is used to intuitively display the pollution core area and diffusion boundary.

[0050] After the secondary detection unit completes the advanced detection, the detection module inputs the detection data into the pollution degree calculation model and generates pollution level classification data based on the preset pollutant threshold range. At the same time, the concentration data of discrete sampling points is converted into a continuous distribution surface through the spatial interpolation algorithm to generate a pollution concentration map. Furthermore, the kernel density estimation method is used to calculate the pollution intensity distribution within a unit area, and the map is rendered into a heat map based on the geographic coordinates. The red area in the pollution heat map represents the pollution core area, and the yellow to green gradient area represents the diffusion transition area, thereby realizing the visual identification of the pollution boundary.

[0051] Traditional methods can only generate a single numerical report or a static concentration distribution map, and cannot dynamically integrate secondary detection data to generate multidimensional analysis results. This solution can simultaneously reflect pollution intensity, spatial heterogeneity and diffusion trends by superimposing pollution degree data, concentration maps and heat maps. For example, the fixed sampling density in existing technologies leads to data redundancy in low-pollution areas, while this solution uses the gradient changes of heat maps to identify high-diffusion risk areas and guide the dynamic adjustment of secondary sampling coordinates near the origin.

[0052] This application solves the problem of low prediction accuracy caused by the separate storage of pollution maps and detection data in the existing technology. The generation of pollution heat maps can be directly linked to the density adjustment of secondary sampling coordinates to avoid missing high-gradient pollution boundaries; the combination of pollution concentration maps and pollution degree data can accurately identify areas of synergistic effects of complex pollution and reduce the frequency of laboratory re-inspections; dynamic rendering technology is used to achieve real-time updating of pollution diffusion prediction models, thereby improving the efficiency of pollution tracing.

[0053] The sending and receiving module establishes an encrypted communication connection with the terminal laboratory, sends advanced detection data and pollution analysis maps to the terminal laboratory, and receives precise verification data returned by the terminal laboratory; the detection database is configured with a machine learning update engine to compare the feature vector differences between the on-site detection data of the sample to be verified and the precise verification data. When the difference exceeds the preset tolerance threshold, the abnormal soil data package is updated, the matching threshold range of the pre-inspection data or re-inspection data is adjusted, and a new abnormal soil data package is generated. The external environmental data and expert knowledge base are integrated to dynamically optimize the feature weight distribution model.

[0054] Encrypted communication connection refers to the use of asymmetric encryption algorithms to establish a data transmission channel. Specifically, the TLS protocol can be used to achieve end-to-end encryption to prevent the detection data from being tampered with or leaked during transmission. The machine learning update engine refers to an adaptive model based on the gradient descent algorithm. Specifically, it can extract multidimensional features in the detection data by building a convolutional neural network and automatically identify pattern offsets in abnormal data packets. The feature vector difference refers to the Euclidean distance after mapping the on-site detection data and the laboratory data to a high-dimensional space. Specifically, the similarity can be calculated after dimensionality reduction using the principal component analysis method to quantify the degree of data consistency. The preset tolerance threshold refers to the maximum deviation range allowed between the on-site detection and laboratory verification results. Specifically, a dynamic threshold can be set based on the historical detection error distribution. For example, For example, when the calibration cycle of the detection equipment changes, the tolerance value is automatically adjusted. The newly added abnormal soil data package refers to the extended database unit formed by integrating the verification difference data. Specifically, a multidimensional data matrix can be constructed by associating the pollution source type, soil physical and chemical properties and environmental parameters. External environmental data refers to meteorological conditions, hydrogeology and land use information. Specifically, real-time environmental parameters can be obtained by accessing satellite remote sensing data and GIS systems. The expert knowledge base refers to a rule library that stores pollution migration patterns and remediation experience. Specifically, knowledge graph technology can be used to construct a pollutant-medium-effect association network. The feature weight allocation model refers to the dynamic adjustment of the contribution of detection indicators in data matching. Specifically, the attention mechanism can be used to calculate the impact weights of different pollution factors in a specific environment.

[0055] The sending and receiving modules transmit the advanced detection data and pollution analysis maps obtained from on-site detection to the terminal laboratory through an encrypted channel. The laboratory uses precision instruments to verify the test of the same sample and return the results. The machine learning update engine extracts multi-dimensional indicators such as spectral characteristics, ion concentration, and biological activity from the on-site detection data, and compares them with the feature vectors corresponding to the laboratory verification data. When the difference between the two exceeds the dynamic threshold set based on historical error statistics, it is determined that there is a deviation in the matching rules in the current abnormal soil data package, and a new data package containing new verification data and environmental parameters is automatically generated. External environmental data obtains information such as rainfall and groundwater flow rate in real time, and combines it with the pollutant migration model stored in the expert knowledge base to dynamically adjust the weight coefficients of different detection indicators in data matching, such as increasing the weight of heavy metal ion solubility under heavy rainfall conditions.

[0056] Traditional detection systems lack a closed-loop feedback mechanism for laboratory verification data. Updates to abnormal data packets rely on manual experience and lag behind environmental changes. The fixed threshold matching method in existing technologies cannot adapt to the synergistic effects of multiple pollutants and dynamic diffusion scenarios, resulting in an increased misjudgment rate. This solution uses encrypted communication to achieve real-time interactive verification of detection data and laboratory results, uses machine learning models to automatically identify data deviations and update matching rules, and combines environmental parameters with expert knowledge to dynamically optimize feature weights, effectively solving the problem of misjudgment caused by changes in pollutant migration characteristics.

[0057] This application realizes the real-time calibration of on-site detection data and laboratory precision verification, eliminates the mismatch problem caused by equipment error or environmental interference, and improves the adaptability of the detection system to complex pollution and migration and diffusion scenarios by dynamically updating abnormal data packets and adjusting feature weights, ensuring the accuracy and timeliness of pollution identification. It uses the fusion analysis of external environmental data and expert knowledge base to enhance the system's ability to predict the dynamic behavior of pollutants and provide data support for the precise formulation of secondary sampling strategies.

[0058] The soil detection method includes the following steps: controlling the sampling module to collect soil samples of a specified depth at a target location; crushing and homogenizing the soil samples through the sample processing module and storing them separately; determining the origin sampling coordinates of the sampling location through the positioning module and binding them to the sample storage location; performing preliminary detection on the soil samples through the preliminary detection unit of the detection module to obtain preliminary detection data; comparing the preliminary detection data with the abnormal soil data packets in the detection database to determine whether they are consistent with the pre-inspection data; generating a pollution location map based on the comparison results and the origin sampling coordinates; if the preliminary detection data are consistent with the pre-inspection data, starting the secondary detection unit; generating the pollution location map by marking the sampling coordinates matching the abnormal data as pollution points; generating a suspected pollution diffusion map based on the secondary detection results, and calculating the spatial uncertainty index of the current pollution area ... The spatial uncertainty index is calculated based on the spatial distribution density of sampling points, the variance of detection data and the rate of change of pollution concentration gradient, and the gradient weight is dynamically adjusted according to the direction of pollution diffusion. When the spatial uncertainty index is greater than the preset threshold, the loop steps are executed: based on the pollution concentration gradient map and the spatial uncertainty index, the near-origin secondary sampling coordinates are generated. The near-origin secondary sampling coordinates are the coordinate points in the close range from the origin sampling coordinates, and the close range is dynamically adjusted according to the pollution diffusion characteristics; the sampling module is controlled to sample the near-origin secondary sampling coordinates; preliminary detection, secondary detection, updating of the pollution concentration gradient map, and recalculation of the spatial uncertainty index are performed on the new samples in sequence; when the spatial uncertainty index is less than or equal to the preset threshold or the sampling density reaches the preset upper limit, the loop is exited; and the final pollution location map, pollution heat map and pollution diffusion prediction model are generated.

[0059] The near-origin secondary sampling coordinates refer to the supplementary sampling points dynamically generated in the adjacent area in the direction of pollutant diffusion based on the origin sampling coordinates. Specifically, they can be achieved by superimposing the pollution concentration gradient change rate and the diffusion model prediction path to capture the dynamic changes of the pollution boundary. The spatial uncertainty index refers to an indicator that quantifies the credibility of the pollution diffusion area. Specifically, it can be achieved through weighted calculation of the sampling point density and the pollution concentration variance, and is used to trigger adaptive supplementary sampling. Dynamic adjustment of pollution diffusion characteristics refers to real-time correction of the sampling range according to the pollutant migration rate, soil permeability and meteorological parameters. Specifically, it can be achieved through the fusion calculation of fluid mechanics models and real-time meteorological data to adapt to the nonlinear diffusion law of pollutants.

[0060] During the pollution location map generation stage, when the preliminary detection data matches the pre-inspection data, the pollution point is marked and a secondary detection is triggered. The secondary detection results are used to construct a pollution concentration gradient map. The spatial uncertainty index is calculated based on the sampling point distribution density and the volatility of the detection data. When the index exceeds the preset threshold, the near-origin secondary sampling coordinates are generated based on the pollution diffusion direction and the concentration gradient change rate. The sampling distance range is dynamically adjusted to achieve accurate capture of high-gradient pollution boundaries. The supplementary sampling data is iteratively detected and the pollution map is updated until the spatial uncertainty index drops below the threshold or reaches the maximum sampling density, and finally a converged pollution diffusion prediction model is output.

[0061] The existing fixed grid sampling strategy uses preset spacing to distribute sampling points, and is unable to adjust the sampling density according to the migration rate of pollutants, resulting in missed detection of high-gradient pollution boundaries. However, this method dynamically triggers supplementary sampling through the spatial uncertainty index, and combines the pollution diffusion model to correct the sampling coordinates in real time, so that the sampling points are focused on high-uncertainty areas. In addition, the existing technology relies on laboratory re-inspection, which prolongs the detection cycle. This method uses a closed-loop verification mechanism of preliminary and secondary detection to significantly shorten the on-site detection iteration cycle while retaining the laboratory data calibration function.

[0062] This application solves the problem of delayed response of fixed sampling strategies to the dynamic diffusion of pollutants. Through the synergistic effect of spatial uncertainty index and dynamic sampling coordinates, adaptive tracking of pollution boundaries is achieved. At the same time, the triggering mechanism of preliminary detection and secondary detection reduces redundant detection operations, reducing detection time while ensuring the accuracy of complex pollution identification. The pollution diffusion prediction model improves the accuracy of pollutant migration path prediction through an iterative update mechanism, providing reliable data support for pollution control decisions.

[0063] When the preliminary test data is consistent with the pre-test data, the secondary test unit adopts at least one of the following methods: the chemical treatment method uses chemical reagents to perform titration tests on soil samples to generate chemical treatment result data; the electrochemical treatment method uses aqueous solution to form a solution or suspension of soil samples to control electrodes to generate electrochemical treatment result data; the biological detection method cultivates soil samples in a closed room to detect carbon dioxide produced or oxygen consumed by microbial respiration to obtain biological activity information; the chemical treatment method is suitable for detecting the quantitative composition of heavy metal ions in the soil through chemical reactions; the electrochemical treatment method is suitable for detecting the concentration of soluble ions in the soil, the electrochemical characteristics of redox potential; the biological detection method is suitable for evaluating the activity of soil microorganisms and the degree of organic pollution.

[0064] Chemical treatment method refers to the detection means of using specific reagents to react quantitatively with soil components. Specifically, silver nitrate titration can be used to detect chloride ion concentration, and the reaction end point can be determined by color change. This method establishes a standardized detection process for heavy metal ions to solve the problem of quantitative analysis of specific components in complex pollution. Electrochemical treatment method refers to the electrochemical parameter measurement technology based on solution system. Specifically, a three-electrode system can be used to measure the cyclic voltammetry curve of soil suspension, and soluble pollutants can be identified by redox peaks. This method improves the detection efficiency of soluble pollutants by establishing a mapping relationship between electrical signals and ion concentrations. Biological detection method refers to biosensor technology based on microbial metabolic activities. Specifically, a closed culture chamber can be connected to a gas sensor to monitor the rate of change of carbon dioxide concentration in real time. This method indirectly reflects the bioavailability of organic pollutants by quantifying the intensity of microbial respiration.

[0065] When the preliminary detection data matches an abnormal soil data package, the detection mode is automatically selected according to the type of pollutant. For example, when an abnormal pH value is detected in the soil of a lead-zinc mining area, the chemical treatment method is preferentially activated for heavy metal ion titration analysis. When salinized soil with abnormally increased conductivity is detected, the electrochemical treatment method is automatically switched to determine the sodium ion concentration. For areas with organic pollution risks, biological detection methods are enabled to evaluate microbial activity. The selection logic of the detection method is embedded in the pollution diffusion model. When the pollution migration direction shows a trend of organic matter diffusion, the frequency of calling the biological detection method is automatically increased. The detection data is linked to the sampling coordinates in real time to form a space-time matrix, providing multi-dimensional data support for pollution boundary determination.

[0066] Traditional methods use a single detection method, resulting in incomplete identification of complex pollution, and fixed detection processes cannot adapt to dynamic pollution characteristics. This solution builds a multi-method collaborative detection system to establish exclusive detection paths for different pollutant types such as heavy metals, soluble salts, and organic matter. The detection strategy is dynamically optimized according to the direction of pollution diffusion. The separation of chemical detection and biological detection in existing technologies leads to data fragmentation. This solution realizes the fusion analysis of multimodal data in a unified spatiotemporal coordinate system.

[0067] This application effectively solves the technical defect of incomplete identification of complex pollutants, improves the collaborative detection capability of heavy metal ions and organic pollutants, avoids the waste of detection resources caused by traditional fixed processes through a dynamic matching mechanism between detection methods and pollution characteristics, and the spatiotemporal binding of multi-dimensional detection data provides a data basis for the precise delineation of pollution boundaries, reducing misjudgment of pollution scope due to a single detection method.

[0068] When generating a suspected pollution diffusion map, the soil permeability, porosity and moisture content of the sampling point are obtained through the preliminary detection unit; based on the fluid mechanics model, the lateral or vertical migration rate of pollutants is calculated in combination with the soil physical parameters; real-time meteorological data is connected to dynamically correct the migration rate and diffusion direction; a pollution migration time series map is generated to mark the spatiotemporal boundaries of the pollutant diffusion within a preset time period; the pollution migration time series map is connected to the secondary sampling, and when the time series map shows that the pollutant is migrating rapidly in a certain direction, the secondary sampling coordinates near the origin are added to the migration path; the time series map is iteratively updated based on the new sampling point data until the prediction converges.

[0069] Soil permeability refers to the ability of soil to allow fluid to pass through it. This can be achieved using a permeability coefficient meter or on-site permeability test and is used to assess the vertical migration capacity of pollutants in soil. Porosity refers to the ratio of pore volume to total volume in soil. It can be calculated using soil bulk density and particle density and reflects the storage space and lateral diffusion potential of pollutants in soil. Moisture content refers to the percentage of water content in soil to dry soil mass. This can be achieved using drying methods or real-time monitoring using sensors and is used to determine the level of pollutant dissolution and migration activity. Fluid dynamics models are pollutant migration simulation algorithms based on Darcy's law or the Navier-Stokes equations. They can be implemented using finite element analysis or numerical simulation software and are used to quantify the diffusion rate of pollutants under specific soil conditions. Real-time meteorological data, including rainfall, wind speed, and temperature, can be obtained from meteorological stations or satellite remote sensing and is used to correct for shifts in pollutant migration paths caused by changing meteorological conditions. Pollution migration time series maps are pollutant diffusion prediction maps that overlay time and spatial coordinates. They can be generated using dynamic rendering of geographic information system layer overlays to visualize pollutant diffusion trends and boundary ranges.

[0070] When generating a pollution diffusion prediction map, the soil permeability, porosity and moisture content parameters are first collected through the preliminary detection unit. These parameters are input into the fluid mechanics model, and the lateral and vertical migration rates of pollutants are obtained by calculation. Subsequently, the system accesses real-time meteorological data. For example, when an increase in rainfall is detected, the model automatically adjusts the migration rate to reflect the promoting effect of water penetration on the solubility of pollutants. Based on the corrected data, a pollution migration time series map is generated. The map shows the spatiotemporal boundaries of possible pollutant diffusion in the next 24 hours. If the map shows that the migration rate of pollutants in the northeast direction has increased significantly, new near-origin secondary sampling coordinates are automatically generated on the diffusion path in the northeast direction. After the sampling module collects samples according to the new coordinates, the detection data is fed back to the model for iterative calculation to update the diffusion boundary in the time series map until the difference between the boundaries of two consecutive calculations is less than the convergence threshold.

[0071] In some specific embodiments, the fluid mechanics model can use the Darcy-Fock diffusion equation, with permeability coefficient and porosity as input variables, to calculate the migration rate of pollutants in unsaturated soil. Meteorological data correction can dynamically adjust the hydraulic conductivity coefficient in the model by establishing a linear relationship function between rainfall intensity and permeability rate. The generation of time series maps can be combined with a geographic information system to superimpose the sampling point coordinates with the diffusion rate vector to generate a dynamic thermal map with a time scale.

[0072] Existing methods rely on fixed sampling grids and static diffusion models, and are unable to adjust prediction results based on differences in soil physical properties and meteorological changes. This solution integrates real-time detected soil parameters and meteorological data to enable the diffusion model to dynamically respond to changes in environmental conditions. At the same time, existing technologies use secondary sampling at fixed time intervals, while this solution actively triggers sampling based on the migration trend of the time series map, prioritizing increasing the sampling density in areas where pollutant diffusion is accelerated, thereby reducing the risk of missed detection.

[0073] This application can dynamically correct the pollutant diffusion prediction results according to the actual physical properties of the soil and changes in the external environment. By iteratively updating the time series map and secondary sampling coordinates, it can effectively capture the mutation area of ​​the pollutant migration path and avoid boundary prediction errors caused by model deviation. This method solves the problem of misjudgment of pollution range caused by static models and fixed sampling strategies in the existing technology, and significantly improves the spatiotemporal accuracy of diffusion prediction.

[0074] Laboratory verification and model optimization steps are added to the soil detection method, including: submitting a set of samples to be verified to the terminal laboratory, samples with a spatial uncertainty index higher than the preset risk threshold; samples whose detection data matches the historical abnormal data package in a fuzzy interval; based on precise verification data, if it is confirmed that there is a significant deviation in the on-site detection, the abnormal soil data package is expanded or corrected; if the laboratory denies the existence of pollution, a counterexample training set is generated to optimize the machine learning model; and external environmental data are associated to construct a regional pollution risk portrait.

[0075] The sample set to be verified refers to a collection of soil samples that require precise laboratory verification. Specifically, a screening algorithm can be used to extract samples whose spatial uncertainty index is higher than the preset risk threshold from the on-site test data, as well as samples whose matching degree between the test data and the historical abnormal data packet is in the fuzzy interval. For example, samples whose matching confidence is lower than the set value calculated using a probability model. This feature is used to identify samples with insufficient reliability in on-site testing to ensure that key data is verified in the laboratory.

[0076] Precision verification data refers to the verification results generated by the laboratory through sample testing using high-precision instruments. Specifically, it can be achieved by mass spectrometry analysis, chromatography analysis or isotope tracing methods. This data is used to verify the accuracy of on-site test results and provide a basis for updating abnormal soil data packages.

[0077] The counterexample training set refers to a set of sample data that has been confirmed to be uncontaminated by the laboratory. It can be generated by comparing the laboratory verification results with the on-site detection data. This training set is used to optimize the classification boundaries of the machine learning model and reduce the probability of misjudgment.

[0078] A regional pollution risk portrait refers to a pollution distribution characteristic model constructed in combination with external environmental data. It can be generated by integrating meteorological data, hydrogeological data and a historical pollution event database. This portrait is used to dynamically predict pollution diffusion trends and improve the accuracy of regional risk assessment.

[0079] After completing the on-site detection, the system automatically screens out samples with high spatial distribution uncertainty or test results in the fuzzy matching range, and packages their coordinates, test data and related environmental parameters and sends them to the terminal laboratory. The laboratory uses precision testing equipment to re-inspect the samples. If a significant deviation is found between the on-site test data and the laboratory data, the difference feature vector is input into the machine learning update engine of the detection database, triggering the expansion or correction of the abnormal soil data packet. If the laboratory confirms that the sample is not contaminated, the test data of the sample is combined with the environmental parameters to generate a counterexample training set for optimizing the feature weight distribution of the machine learning model. At the same time, the system links the laboratory verification data with the external environmental database, generates a regional pollution risk portrait through the pollution source resolution algorithm, and dynamically updates the pollution diffusion prediction model.

[0080] Existing soil testing methods lack a closed-loop feedback mechanism for laboratory verification and on-site testing, resulting in delayed updates of abnormal data packets and reliance on manual experience for model optimization. This solution, by automatically screening samples to be verified and establishing a linkage mechanism between laboratory data and machine learning models, enables dynamic correction of abnormal data packets and self-optimization of models. At the same time, it combines external environmental data to construct a multidimensional pollution risk profile, significantly improving the accuracy of pollution source tracing and diffusion prediction.

[0081] This application can effectively solve the misjudgment problem caused by the separation of on-site testing and laboratory data, reduce the missed detection rate of pollution boundaries through a closed-loop verification mechanism, optimize the model classification ability using counterexample training sets, and dynamically generate pollution risk portraits based on environmental data to provide accurate decision-making support for pollution control.

Claims

1. A soil detection system, characterized in that: include: A sampling module for sampling soil; Positioning module, used to locate the sampling location and generate the origin sampling coordinates; A sample processing module, used to separately store the soil samples obtained by the sampling module and bind the storage location to the origin sampling coordinates; A detection module, comprising a preliminary detection unit and a secondary detection unit, wherein the preliminary detection unit is used to perform preliminary detection on the treated soil sample to obtain preliminary detection data; A detection database storing a plurality of abnormal soil data packets, wherein the abnormal soil data packets record data on abnormal soil including pre-inspection data and re-inspection data; The detection control module is used to match the preliminary detection data with the pre-inspection data of the abnormal soil data packet. When the preliminary detection data is within the pre-inspection data range, the secondary detection unit is triggered to perform a secondary detection on the soil sample at the corresponding storage location, and the advanced detection data obtained by the secondary detection is matched with the re-inspection data to confirm the abnormal concentration value of the soil sample; A sending module is generated to generate a pollution location map based on the origin sampling coordinates and the matching results. The near-origin secondary sampling coordinates are dynamically generated based on the pollution location map. The near-origin secondary sampling coordinates are coordinate points within a close range from the origin sampling coordinates. The close range of the origin sampling coordinates is dynamically adjusted according to the pollution diffusion characteristics, and the sampling module is controlled to sample the secondary coordinate points until the sampling density reaches a preset threshold.

2. A soil detection system according to claim 1, characterized in that: The primary detection unit is used to detect at least one of the hardness, moisture, pH value, organic matter and spectral characteristics of the soil; the secondary detection unit is used to perform secondary detection on the soil sample to obtain advanced detection data.

3. A soil detection system according to claim 2, characterized in that: The sampling module is equipped with a segmented sampling rod group, which is equipped with several sampling rods and integrated with a soil texture sensor. The soil texture sensor is used to detect changes in soil resistivity in real time. When the resistivity mutation value exceeds a threshold, it is determined to be a geological stratification interface, and the sampling rod is replaced for segmented sampling.

4. A soil detection system according to claim 1, characterized in that: The sample processing module is equipped with processing means adapted to the secondary detection unit, including: Chemical treatment method, which is equipped with a number of chemical reagents, uses the chemical reagents to perform titration tests on soil samples and generate chemical treatment result data; The electrochemical treatment method is configured with an aqueous solution and a test electrode, wherein the soil sample is formed into a solution or suspension by the aqueous solution, and the control electrode generates electrochemical treatment result data; The biological detection method is equipped with several closed chambers for culturing soil samples in a sealed environment to detect carbon dioxide produced or oxygen consumed by microbial respiration and obtain biological activity information.

5. A soil detection system according to claim 1, characterized in that: The detection module is further configured to generate pollution degree data, a pollution concentration map, and a pollution heat map based on the detection data of the secondary detection unit.

6. A soil detection system according to claim 1, characterized in that: Also includes: The sending and receiving module establishes an encrypted communication connection with the terminal laboratory and is used to: Sending the advanced detection data and pollution analysis map to the terminal laboratory; Receive precision verification data returned by the terminal laboratory; The detection database is configured with a machine learning update engine for: Compare the feature vector differences between the field test data of the sample to be verified and the precision verification data; When the difference exceeds the preset tolerance threshold, the abnormal soil data package is updated, and the matching threshold range of the pre-inspection data or re-inspection data is adjusted to generate a new abnormal soil data package; the external environmental data and expert knowledge base are integrated to dynamically optimize the feature weight distribution model.

7. A soil detection method, applied to a soil detection system according to any one of claims 1 to 6, characterized in that: The following steps are involved: S1: Control the sampling module to collect soil samples at a specified depth at the target location; S2: The soil samples are crushed, homogenized and stored separately through the sample processing module; S3: Determine the origin sampling coordinates of the sampling location through the positioning module and bind them to the sample storage location; S4: Performing preliminary testing on the soil sample through the preliminary testing unit of the testing module to obtain preliminary testing data; S5: Compare the preliminary test data with the abnormal soil data package in the test database to determine whether it is consistent with the pre-test data; S6: Generate a contamination location map based on the comparison results and the origin sampling coordinates; if the preliminary detection data is consistent with the pre-inspection data, start the secondary detection unit; the contamination location map is generated by marking the sampling coordinates of the matching abnormal data as contamination points; S7: Generate a suspected pollution diffusion map based on the secondary detection results and calculate the spatial uncertainty index of the current pollution area. The spatial uncertainty index is calculated based on the spatial distribution density of the sampling points, the variance of the detection data, and the gradient change rate of the pollution concentration. The gradient weight is dynamically adjusted according to the pollution diffusion direction. S8: When the spatial uncertainty index is greater than the preset threshold, the following loop steps are executed: S801: Based on the pollution concentration gradient map and the spatial uncertainty index, generate near-origin secondary sampling coordinates, where the near-origin secondary sampling coordinates are coordinate points within a close range from the origin sampling coordinates, and the close range is dynamically adjusted according to pollution diffusion characteristics; S802: Control the sampling module to sample the secondary sampling coordinates near the origin; S803: Perform preliminary testing, secondary testing, update the pollution concentration gradient map, and recalculate the spatial uncertainty index on the new sample in sequence; S9: When the spatial uncertainty index is less than or equal to the preset threshold or the sampling density reaches the preset upper limit, exit the loop; S10: Generate the final pollution location map, pollution heat map and pollution diffusion prediction model.

8. The method according to claim 7, characterized in that In step S6, when the preliminary detection data is consistent with the pre-detection data, the secondary detection unit adopts at least one of the following methods: Chemical treatment method: soil samples are titrated using chemical reagents to generate chemical treatment result data; Electrochemical treatment method: The soil sample is converted into a solution or suspension through an aqueous solution, and the control electrode generates the electrochemical treatment result data; Biological detection method: culturing soil samples in the closed chamber, detecting carbon dioxide produced or oxygen consumed by microbial respiration, and obtaining biological activity information; The application scenarios of the chemical treatment method, electrochemical treatment method and biological detection method include: Chemical treatment methods are suitable for detecting the quantitative composition of heavy metal ions in soil through chemical reactions; The electrochemical treatment method is suitable for detecting the concentration of soluble ions and electrochemical characteristics of redox potential in soil; Biological detection methods are suitable for evaluating soil microbial activity and the degree of organic pollution.

9. The method according to claim 7, characterized in that In step S7, the suspected pollution diffusion map is generated by obtaining the soil permeability, porosity and moisture content of the sampling points through the preliminary detection unit; Based on the fluid mechanics model and in combination with the soil physical parameters, the lateral / vertical migration rate of pollutants is calculated; Access to real-time meteorological data to dynamically correct migration rate and diffusion direction; Generate a pollution migration time series map, marking the spatial and temporal boundaries of pollutant diffusion within a preset time period; Connecting the pollution migration time series map with the secondary sampling, and when the time series map shows that the pollutants are migrating in a certain direction at an accelerated rate, adding the secondary sampling coordinates near the origin on the migration path; The time series graph is iteratively updated based on the new sampling point data until the prediction converges.

10. The method according to claim 7, characterized in that The following steps are also included: S11: Submit the sample set to be verified to the terminal laboratory, including samples whose spatial uncertainty index is higher than the preset risk threshold; and samples whose detection data matches the historical abnormal data packets in the fuzzy range; S12: Based on the precise verification data, if significant deviations are confirmed in the on-site testing, the abnormal soil data package is expanded or corrected; If the laboratory denies the existence of contamination, a counterexample training set is generated to optimize the machine learning model; Associate external environmental data to build a regional pollution risk profile.

Images