A comprehensive soil pollution detection method and system

Abstract

The application relates to the technical field of soil detection, and discloses a soil pollution comprehensive detection method and system.The method comprises the following steps: carrying out denoising filtering on the spectral reflectivity data and ion activity detection data of a target region soil sample, and carrying out normalization processing on the denoised data to obtain a spectral feature vector and an ion activity feature vector of the target region soil sample; a multi-modal feature matrix of the target region soil sample is constructed; the pollution types of the target region soil sample are classified to obtain the pollution types of the soil sample; the pollution types are associated with sensitive feature factors in the multi-modal feature matrix for evaluation to obtain a comprehensive pollution index of the soil sample; the numerical interval of the pollution types and the comprehensive pollution index is cooperatively integrated to obtain soil pollution grade evaluation results of the target region soil sample; and the application can improve the comprehensive evaluation efficiency of the soil pollution comprehensive detection.

Description

Technical Field

[0001] This invention relates to the field of soil testing technology, and in particular to a comprehensive method and system for detecting soil pollution. Background Technology

[0002] Existing soil pollution detection technologies mostly employ a single detection dimension for analysis, relying solely on spectral reflectance data or ion activity detection data for independent calculations. They fail to achieve collaborative integration of multi-source detection data, and the data preprocessing stage lacks a complete noise reduction, filtering, and normalization process. This results in insufficient accuracy in extracting soil pollution features, failing to truly reflect the actual situation of soil pollution.

[0003] Existing detection methods are insufficient for accurately identifying soil pollution types and have a weak ability to distinguish between different pollution patterns. Furthermore, they fail to correlate pollution types with sensitive characteristic factors, lack scientific weight allocation and nonlinear correction logic for pollution classification, and employ a single method for calculating the comprehensive pollution index. Ultimately, this results in insufficient accuracy of detection results, low efficiency of comprehensive assessment, and an inability to achieve rapid, in-situ comprehensive detection of soil pollution. Summary of the Invention

[0004] This invention provides a comprehensive soil pollution detection method and system to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a comprehensive soil pollution detection method, comprising: S1. Denoise filtering is performed on the spectral reflectance data and ion activity detection data of the soil samples in the target area, and the denoised data is normalized to obtain the spectral feature vector and ion activity feature vector of the soil samples in the target area. S2. The spectral feature vector and the ion activity feature vector are fused in the feature layer to construct a multimodal feature matrix of the soil sample in the target area; S3. Based on the multimodal feature matrix, the pollution types of soil samples in the target area are classified to obtain the pollution type of the soil samples; S4. Based on the preset pollution classification threshold, the pollution type is correlated with the sensitive feature factors in the multimodal feature matrix to obtain the comprehensive pollution index of the soil sample. S5. By coordinating and integrating the numerical ranges of pollution types and comprehensive pollution indices, the soil pollution level assessment results of soil samples in the target area are obtained.

[0006] In a preferred embodiment, the step of denoising and filtering the spectral reflectance data and ion activity detection data of the soil samples in the target area, and then normalizing the denoised data to obtain the spectral feature vector and ion activity feature vector of the soil samples in the target area, includes: The reflectance values ​​at each wavelength point in the spectral reflectance data of the soil sample in the target area are statistically averaged by performing adjacent arithmetic mean calculations to obtain the reflectance value at each wavelength point. The ion activity values ​​of the sampling points in the ion activity detection data of the soil samples in the target area are proportionally projected to obtain the ion activity values ​​of the sampling points. Thresholds were set for reflectance and ion activity values, and noise was removed from the defined data to obtain the denoised ion activity and reflectance values ​​of the soil samples in the target area. The denoised ion activity values ​​and denoised reflectance values ​​are linearly mapped to obtain the spectral feature vector and ion activity feature vector of the soil sample in the target area.

[0007] In a preferred embodiment, the step of fusing the spectral feature vector and the ion activity feature vector at a feature layer to construct a multimodal feature matrix of the soil sample in the target area includes: The correlation strength between the spectral feature values ​​at each band position in the spectral feature vector and the activity feature values ​​of each heavy metal ion in the ion activity feature vector is coupled and calculated. Based on the calculation results, the target spectral bands of the soil samples in the target area are selected. The spectral feature values ​​corresponding to the target spectral band are concatenated with all ion activity feature values ​​in the ion activity feature vector to obtain the combined feature sequence of the target spectral band. Based on the sampling location and sampling time of the soil samples in the target area, the combined feature sequences are arranged in a regular manner to obtain the multimodal feature matrix of the soil samples in the target area.

[0008] In a preferred embodiment, the formula for calculating the correlation strength is as follows: ; In the formula, For the first The position of the first band and the first The correlation strength between heavy metal ions The spectral eigenvector is the th Spectral characteristic values ​​at each band position The ion activity eigenvector is the first... Characteristic values ​​of the ion activity of various heavy metal ions. This is the arithmetic mean of the spectral eigenvalues ​​at all band positions in the spectral eigenvector. This is the arithmetic mean of the ion activity eigenvalues ​​of all heavy metal ions in the ion activity eigenvector. It is a very small positive number.

[0009] In a preferred embodiment, the step of systematically arranging the combined feature sequences according to the sampling location and sampling time of the soil samples in the target area to obtain the multimodal feature matrix of the soil samples in the target area includes: The combined feature sequences corresponding to different sampling times at the same sampling location in the soil samples of the target area are arranged sequentially according to the sampling time to form a temporal feature submatrix of the sampling location; The temporal feature submatrix is ​​adaptively spatially fused according to the spatial distribution order of the sampling locations within the target area to obtain the multimodal feature matrix of the soil samples in the target area.

[0010] In a preferred embodiment, the step of classifying the pollution types of soil samples in the target area based on a multimodal feature matrix to obtain the pollution type of the soil samples includes: The similarity assessment between the multimodal feature matrix and the preset polluted soil feature benchmark is performed to obtain the degree of feature difference of the multimodal feature matrix; Based on the degree of difference in characteristics, the pollution type corresponding to the selected polluted soil characteristic benchmarks is taken as the pollution type of the soil sample. The pollution types include single heavy metal pollution, compound heavy metal pollution, or organic-inorganic compound pollution.

[0011] In a preferred embodiment, the step of correlating and evaluating pollution types with sensitive feature factors in a multimodal feature matrix based on a preset pollution classification threshold to obtain a comprehensive pollution index for the soil sample includes: Based on the influence relationship of pollution types, the peak activity of heavy metal ions, the characteristic value of spectral absorption depth, and the change in spectral characteristic curvature in the multimodal characteristic matrix are integrated into the sensitivity characteristic factors of pollution types. The sensitivity feature factors are mapped to preset pollution classification thresholds to generate the pollution level of the sensitivity feature factors; Based on the specific pollutant types identified in the pollution types, weights are assigned to the sensitivity characteristic factors to construct weight adjustment rules for the sensitivity characteristic factors. According to the weighting adjustment rules, the individual pollution contribution of sensitive characteristic factors is superimposed and summarized with the pollution level to obtain the comprehensive pollution index of the soil sample.

[0012] In a preferred embodiment, the comprehensive pollution index is calculated using the following formula: ; In the formula, The comprehensive pollution index, This is a sensitivity characteristic factor index. A value of 1 indicates the peak activity of heavy metal ions, a value of 2 indicates the characteristic value of spectral absorption depth, and a value of 3 indicates the change in spectral characteristic curvature. For the first The weight coefficients of each sensitivity feature factor under the current pollution type. For the first Measured values ​​of each sensitivity characteristic factor For the first Background reference values ​​for each sensitivity feature factor The preset coupling suppression coefficient, It is the natural logarithm function. It is a very small positive number. As a pollution type modulator, It is a natural exponential function. This is a preset nonlinear sensitivity index.

[0013] In a preferred embodiment, the step of synergistically integrating the numerical ranges of pollution type and comprehensive pollution index to obtain the soil pollution level assessment result of the soil sample in the target area includes: Based on the pollution type and comprehensive pollution index, the preset pollution level interval mapping table is searched and filtered to obtain the graded interval set of soil samples in the target area. The target interval of the soil sample in the target area is obtained by sequentially comparing the value of the comprehensive pollution index with the endpoint value of the concentrated interval of the classification interval. The pollution level labels associated with the target interval are used as the soil pollution level assessment results for soil samples in the target area; however, different pollution level labels are allowed for the same comprehensive pollution index value under different pollution types.

[0014] To address the above problems, the present invention also provides a comprehensive soil pollution detection system, the system comprising: The data preprocessing module is used to denoise and filter the spectral reflectance data and ion activity detection data of soil samples in the target area, and to normalize the denoised data to obtain the spectral feature vector and ion activity feature vector of the soil samples in the target area. The multimodal feature fusion module is used to fuse spectral feature vectors and ion activity feature vectors at the feature layer to construct a multimodal feature matrix of soil samples in the target area. The pollution type identification module is used to classify the types of pollution in soil samples in the target area based on a multimodal feature matrix, and to obtain the pollution type of the soil samples. The comprehensive pollution index assessment module is used to correlate pollution types with sensitive feature factors in the multimodal feature matrix based on preset pollution classification thresholds to obtain the comprehensive pollution index of soil samples. The pollution level determination module is used to integrate the numerical ranges of pollution type and comprehensive pollution index to obtain the soil pollution level assessment results of soil samples in the target area.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention systematically denoises and normalizes soil spectral reflectance data and ion activity detection data to accurately generate standardized feature vectors. Through deep coupling and fusion of feature layers, a multimodal feature matrix is ​​constructed to fully extract the core feature information of soil pollution, effectively improving the accuracy of detecting multiple pollutants in soil and significantly enhancing the reliability and authenticity of in-situ detection results.

[0016] 2. This invention achieves accurate identification of pollution types based on a multimodal feature matrix, and completes the scientific calculation of the comprehensive pollution index and the collaborative determination of pollution level by combining pollution classification thresholds. It optimizes the overall calculation process of pollution assessment, greatly improves the work efficiency of comprehensive soil pollution assessment, realizes rapid and comprehensive detection of soil pollution, strengthens the comprehensiveness and operability of detection work, and fully meets the actual application needs of comprehensive soil pollution detection. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart of a comprehensive soil pollution detection method provided in an embodiment of the present invention; Figure 2 This is a functional block diagram of a comprehensive soil pollution detection system provided in an embodiment of the present invention.

[0018] Figure reference numerals: 101, Data preprocessing module; 102, Multimodal feature fusion module; 103, Pollution type identification module; 104, Comprehensive pollution index assessment module; 105, Pollution level determination module. Detailed Implementation

[0019] The realization of the objectives, functional features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings; it should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] This application provides a comprehensive soil pollution detection method. The executing entity of this comprehensive soil pollution detection method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application embodiment: a server, a terminal, etc. In other words, the comprehensive soil pollution detection method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0021] Reference Figure 1 The diagram shown is a flowchart illustrating a comprehensive soil pollution detection method according to an embodiment of the present invention. In this embodiment, the comprehensive soil pollution detection method includes: S1. Denoise filtering is performed on the spectral reflectance data and ion activity detection data of the soil samples in the target area, and the denoised data is normalized to obtain the spectral feature vector and ion activity feature vector of the soil samples in the target area. In this embodiment of the invention, the step of denoising and filtering the spectral reflectance data and ion activity detection data of the soil sample in the target area, and then normalizing the denoised data to obtain the spectral feature vector and ion activity feature vector of the soil sample in the target area, includes: The reflectance values ​​at each wavelength point in the spectral reflectance data of the soil sample in the target area are statistically averaged by performing adjacent arithmetic mean calculations to obtain the reflectance value at each wavelength point. The ion activity values ​​of the sampling points in the ion activity detection data of the soil samples in the target area are proportionally projected to obtain the ion activity values ​​of the sampling points. Thresholds were set for reflectance and ion activity values, and noise was removed from the defined data to obtain the denoised ion activity and reflectance values ​​of the soil samples in the target area. The denoised ion activity values ​​and denoised reflectance values ​​are linearly mapped to obtain the spectral feature vector and ion activity feature vector of the soil sample in the target area.

[0022] For each wavelength point in the spectral reflectance data of the soil sample in the target area, extract the original reflectance values ​​of that wavelength point and the two directly adjacent wavelength points. Add these three values ​​together and then divide the sum by three to obtain the average value. Use this average value as the reflectance value of that wavelength point. Complete the calculation and processing of all wavelength points in this way to obtain the reflectance values ​​of all wavelength points in the soil sample of the target area.

[0023] According to the actual spatial distribution ratio of each sampling point in the soil sample of the target area, the original ion activity values ​​in the ion activity detection data are projected onto the standard positions of each sampling point, so that the ion activity values ​​and the sampling point positions form a unique correspondence. After the projection processing is completed, the ion activity values ​​of all sampling points of the soil sample in the target area are obtained.

[0024] Using preset acceptable thresholds for spectral reflectance and ion activity, the reflectance values ​​at wavelength points are compared with the acceptable thresholds for spectral reflectance, and the ion activity values ​​at sampling points are compared with the acceptable thresholds for ion activity. Values ​​exceeding the corresponding acceptable threshold range are identified as invalid noise points, which are then removed from the data. Values ​​within the acceptable threshold range are retained to obtain the denoised ion activity values ​​and denoised reflectance values ​​of the soil samples in the target area.

[0025] According to the preset standardized numerical range, the denoised reflectance values ​​are converted into standard values ​​within the range one by one. All the converted standard values ​​are arranged and combined according to the order of wavelength points to form the spectral feature vector of the soil sample in the target area. Similarly, according to the same preset standardized numerical range, the denoised ion activity values ​​are converted into standard values ​​within the range one by one. All the converted standard values ​​are arranged and combined according to the distribution order of sampling points to form the ion activity feature vector of the soil sample in the target area.

[0026] The beneficial effects are that by performing adjacent arithmetic mean processing on spectral reflectance data, performing proportional position projection processing on ion activity detection data, and then combining preset thresholds to define and remove noise, and finally standardizing the data form through linear mapping, standard spectral feature vectors and ion activity feature vectors can be stably obtained, ensuring the accuracy and standardization of data processing, and providing a reliable data foundation for subsequent soil pollution detection.

[0027] S2. The spectral feature vector and the ion activity feature vector are fused in the feature layer to construct a multimodal feature matrix of the soil sample in the target area; In this embodiment of the invention, the step of fusing spectral feature vectors and ion activity feature vectors at a feature layer to construct a multimodal feature matrix of soil samples in the target area includes: The correlation strength between the spectral feature values ​​at each band position in the spectral feature vector and the activity feature values ​​of each heavy metal ion in the ion activity feature vector is coupled and calculated. Based on the calculation results, the target spectral bands of the soil samples in the target area are selected. The spectral feature values ​​corresponding to the target spectral band are concatenated with all ion activity feature values ​​in the ion activity feature vector to obtain the combined feature sequence of the target spectral band. Based on the sampling location and sampling time of the soil samples in the target area, the combined feature sequences are arranged in a regular manner to obtain the multimodal feature matrix of the soil samples in the target area.

[0028] The formula for calculating the correlation strength is as follows: ; In the formula, For the first The position of the first band and the first The correlation strength between heavy metal ions The spectral eigenvector is the th Spectral characteristic values ​​at each band position The ion activity eigenvector is the first... Characteristic values ​​of the ion activity of various heavy metal ions. This is the arithmetic mean of the spectral eigenvalues ​​at all band positions in the spectral eigenvector. This is the arithmetic mean of the ion activity eigenvalues ​​of all heavy metal ions in the ion activity eigenvector. It is a very small positive number.

[0029] The method involves systematically arranging the combined feature sequences according to the sampling location and sampling time of soil samples in the target area to obtain a multimodal feature matrix of soil samples in the target area, including: The combined feature sequences corresponding to different sampling times at the same sampling location in the soil samples of the target area are arranged sequentially according to the sampling time to form a temporal feature submatrix of the sampling location; The temporal feature submatrix is ​​adaptively spatially fused according to the spatial distribution order of the sampling locations within the target area to obtain the multimodal feature matrix of the soil samples in the target area.

[0030] The spectral feature values ​​of each band position in the spectral feature vector are extracted one by one. The spectral feature values ​​are matched and correlated with the activity feature values ​​of each heavy metal ion in the ion activity feature vector to obtain the correlation strength value of each set of spectral feature values ​​and ion activity feature values. All correlation strength values ​​are compared with the preset correlation strength judgment threshold. The spectral bands whose correlation strength values ​​reach the correlation strength judgment threshold are retained and these spectral bands are determined as the target spectral bands of the soil samples in the target area.

[0031] All spectral feature values ​​corresponding to the target spectral band are arranged in chronological order of the bands. Then, all ion activity feature values ​​in the ion activity feature vector are arranged in chronological order of heavy metal ion types. The arranged spectral feature values ​​and the arranged ion activity feature values ​​are then sequentially connected in a fixed direction to form a combined feature sequence of the target spectral band.

[0032] The combined feature sequences corresponding to different sampling times at the same sampling location within the soil sample of the target area are arranged sequentially according to the order of sampling time to form the temporal feature sub-matrix corresponding to that sampling location. Then, all the temporal feature sub-matrices corresponding to the sampling locations are integrated and arranged according to the actual spatial distribution order of the sampling locations within the target area to finally form the multimodal feature matrix of the soil sample of the target area.

[0033] Correlation strength is a numerical value used to measure the degree of correlation between a single spectral position and a single heavy metal ion. This value is calculated by coupling spectral characteristics and ion activity characteristics, and is the core criterion for screening target spectral bands.

[0034] The spectral feature values ​​at a single band location are extracted from the spectral feature vector generated after denoising and normalization, and are the basic values ​​characterizing the corresponding band spectral characteristics of soil samples in the target area.

[0035] The ion activity feature values ​​of a single heavy metal ion are extracted from the ion activity feature vector generated after noise reduction and normalization, and are the basic values ​​for characterizing the activity characteristics of corresponding heavy metal ions in soil samples of the target area.

[0036] The arithmetic mean of the spectral characteristic values ​​of all band positions in the spectral characteristic vector is calculated by summing the spectral characteristic values ​​of all bands in the spectral characteristic vector and dividing by the total number of bands. It is used to reflect the average level of the overall spectral characteristics of the soil sample.

[0037] The arithmetic mean of the ion activity characteristic values ​​of all heavy metal ions in the ion activity characteristic vector is calculated by summing the ion activity characteristic values ​​of all heavy metal ions in the ion activity characteristic vector and dividing by the total number of heavy metal ion species. It is used to reflect the average level of the overall ion activity characteristics of the soil sample.

[0038] The minimum positive number is a fixed value determined based on historical practice data and engineering experience in the field of soil pollution detection. It is used to prevent the denominator from being zero during the calculation process and to ensure that the correlation strength calculation process can be completed stably.

[0039] This formula can accurately quantify the correlation between spectral bands and heavy metal ions. At the same time, it can correct the calculation results by combining the overall average level of the two types of features, which can eliminate the influence of single sample feature bias, provide accurate and reliable numerical support for the screening of target spectral bands, and ensure the accuracy of subsequent multimodal feature matrix construction.

[0040] Determine a single sampling location within the target area, collect the combined feature sequences corresponding to that sampling location across all sampling times, and arrange these combined feature sequences in a regular manner according to the natural chronological order of the sampling times, so that each combined feature sequence and the sampling time form a stable one-to-one correspondence. After the arrangement is completed, a temporal feature submatrix for that sampling location is formed.

[0041] Based on the preset spatial distribution coordinate rules of sampling locations within the target area, the spatial arrangement order of all sampling locations is determined. The temporal feature sub-matrices corresponding to each sampling location are then spliced ​​and integrated in sequence according to this spatial arrangement order, so that each temporal feature sub-matrice is matched to its own sampling location spatial node. After integration, the multimodal feature matrix of the soil sample in the target area is obtained.

[0042] The beneficial effects include the ability to accurately extract and match spectral feature values ​​and heavy metal ion activity feature values, screen effective target spectral bands through correlation strength, and then organize them into a combined feature sequence. A multimodal feature matrix is ​​constructed by combining sampling time and spatial distribution rules. The calculation process introduces mean correction to eliminate sample feature bias, and sets a very small positive number to avoid the denominator being zero, ensuring the stability and reliability of correlation strength calculation. This provides accurate support for the screening of target spectral bands and the construction of multimodal feature matrices, and improves the accuracy and stability of soil sample feature analysis.

[0043] S3. Based on the multimodal feature matrix, the pollution types of soil samples in the target area are classified to obtain the pollution type of the soil samples; In this embodiment of the invention, the step of classifying the pollution types of soil samples in the target area based on a multimodal feature matrix to obtain the pollution type of the soil samples includes: The similarity assessment between the multimodal feature matrix and the preset polluted soil feature benchmark is performed to obtain the degree of feature difference of the multimodal feature matrix; Based on the degree of difference in characteristics, the pollution type corresponding to the selected polluted soil characteristic benchmarks is taken as the pollution type of the soil sample. The pollution types include single heavy metal pollution, compound heavy metal pollution, or organic-inorganic compound pollution.

[0044] The pre-defined contaminated soil characteristic benchmarks are constructed in advance for three pollution types: single heavy metal pollution, complex heavy metal pollution, and organic-inorganic complex pollution. Standard soil samples corresponding to each pollution type are selected, and three standard multimodal feature matrices are built according to the method for constructing multimodal feature matrices of soil samples in the target area. Each pollution type corresponds to a specific contaminated soil characteristic benchmark, and all benchmarks have undergone multiple verification and calibrations to ensure accurate characterization of the corresponding pollution type. The multimodal feature matrices of soil samples from the target area are compared one by one with these three pre-defined contaminated soil characteristic benchmarks. During the comparison, each position in the multimodal feature matrix is ​​compared. The feature values ​​are compared one-to-one with the feature values ​​at the same location in the corresponding polluted soil feature benchmark. The matching degree of the feature values ​​at each location is counted. The similarity between the two is measured by the proportion of matching feature values ​​to the total number of feature values ​​in the multimodal feature matrix. The feature difference degree is obtained by subtracting the similarity value from 1. Finally, the feature difference degree of the multimodal feature matrix of the target area soil sample is obtained from the three preset polluted soil feature benchmarks. The feature difference degree ranges from 0 to 1. The smaller the value, the higher the similarity between the multimodal feature matrix and the corresponding polluted soil feature benchmark. The larger the value, the greater the difference between the two.

[0045] A threshold for judging the degree of feature difference is pre-set. This threshold is determined based on historical practice data in the field of soil pollution detection, the characteristic difference patterns of soils with different pollution types, and engineering application experience. Specifically, it is set to 0.2. This threshold is used to judge whether the multimodal feature matrix matches the preset polluted soil characteristic benchmark. The three feature difference degrees obtained in the previous step are compared with this threshold one by one. The preset polluted soil characteristic benchmark with a feature difference degree less than or equal to 0.2 is selected. If only the polluted soil characteristic benchmark corresponding to a single heavy metal pollution is obtained after screening, then the single heavy metal pollution is determined as the pollution type of the soil sample in the target area. If only the polluted soil characteristic benchmark corresponding to a compound heavy metal pollution is obtained after screening, then the polluted soil characteristic benchmark corresponding to a compound heavy metal pollution is determined as the pollution type of the soil sample in the target area. If the soil pollution characteristic benchmark for metal pollution is used, then the pollution type of the soil sample in the target area is determined to be complex heavy metal pollution. If only the soil pollution characteristic benchmark corresponding to organic-inorganic complex pollution is obtained after screening, then the organic-inorganic complex pollution is determined to be the pollution type of the soil sample in the target area. If multiple preset soil pollution characteristic benchmarks with a characteristic difference of less than or equal to 0.2 appear after screening, then the preset soil pollution characteristic benchmark with the smallest characteristic difference value is selected, and the pollution type corresponding to the benchmark is determined to be the pollution type of the soil sample in the target area. This ensures that the pollution type of the soil sample obtained in the end is unique and accurate, and completely corresponds to one of the three preset pollution types.

[0046] The beneficial effects are that the preset contaminated soil characteristic benchmarks are constructed for three types of pollution and have been verified and calibrated multiple times. They can accurately characterize the characteristics of the corresponding pollution types. By comparing the multimodal feature matrix of the soil sample in the target area with the corresponding location feature values ​​of the three preset benchmarks one by one, the degree of feature difference between the two can be accurately determined. A reasonable threshold for judging the degree of feature difference can be determined in advance based on relevant practical experience. By comparing the degree of feature difference with the threshold to select the matching benchmark, the unique and accurate pollution type of the soil sample can be identified. This ensures that the pollution type determination completely corresponds to the three preset types, thereby improving the accuracy and reliability of pollution type identification.

[0047] S4. Based on the preset pollution classification threshold, the pollution type is correlated with the sensitive feature factors in the multimodal feature matrix to obtain the comprehensive pollution index of the soil sample. In this embodiment of the invention, the step of correlating and evaluating pollution types with sensitive feature factors in a multimodal feature matrix based on a preset pollution classification threshold to obtain a comprehensive pollution index for the soil sample includes: Based on the influence relationship of pollution types, the peak activity of heavy metal ions, the characteristic value of spectral absorption depth, and the change in spectral characteristic curvature in the multimodal characteristic matrix are integrated into the sensitivity characteristic factors of pollution types. The sensitivity feature factors are mapped to preset pollution classification thresholds to generate the pollution level of the sensitivity feature factors; Based on the specific pollutant types identified in the pollution types, weights are assigned to the sensitivity characteristic factors to construct weight adjustment rules for the sensitivity characteristic factors. According to the weighting adjustment rules, the individual pollution contribution of sensitive characteristic factors is superimposed and summarized with the pollution level to obtain the comprehensive pollution index of the soil sample.

[0048] The formula for calculating the comprehensive pollution index is as follows: ; In the formula, The comprehensive pollution index, This is a sensitivity characteristic factor index. A value of 1 indicates the peak activity of heavy metal ions, a value of 2 indicates the characteristic value of spectral absorption depth, and a value of 3 indicates the change in spectral characteristic curvature. For the first The weight coefficients of each sensitivity feature factor under the current pollution type. For the first Measured values ​​of each sensitivity characteristic factor For the first Background reference values ​​for each sensitivity feature factor The preset coupling suppression coefficient, It is the natural logarithm function. It is a very small positive number. As a pollution type modulator, It is a natural exponential function. This is a preset nonlinear sensitivity index.

[0049] The comprehensive pollution index is derived through the integrated calculation of various relevant parameters in the formula. Specifically, it comprehensively calculates the weighted pollution-related values ​​of each sensitive characteristic factor, the coupling inhibition effect, and the pollution type moderating effect, ultimately obtaining a quantitative value that reflects the overall pollution status of soil samples in the target area. The sensitive characteristic factor index is based on the types of sensitive characteristic factors identified earlier. As clearly stated earlier, sensitive characteristic factors include three categories: heavy metal ion activity peaks, spectral absorption depth characteristic values, and spectral characteristic curvature changes. Therefore, index values ​​are set to correspond to these three types of characteristic factors to clearly distinguish different sensitive characteristic factors.

[0050] The weighting coefficients are determined based on the weighting adjustment rules established earlier. First, the specific types of pollutants included in the currently identified pollution types are identified. Then, based on the degree of harm of different pollutants to the soil environment, the pollution diffusion rate, and the detection priority, the three types of sensitive characteristic factors are weighted. The specific value corresponding to each sensitive characteristic factor determined after allocation is the weighting coefficient, and the sum of all weighting coefficients is strictly controlled to be 1.

[0051] The measured values ​​are obtained by extracting and screening from the multimodal feature matrix of soil samples in the target area. First, the pollution type of the soil sample is determined, and then three types of feature data are extracted from the multimodal feature matrix: peak activity of heavy metal ions, spectral absorption depth feature value, and spectral feature curvature change. After removing feature data that is not related to the current pollution type, the remaining feature data related to the pollution type are the measured values ​​of the corresponding sensitivity feature factors.

[0052] The background reference values ​​are derived from historical practice data in soil pollution detection, the environmental hazard levels of different pollutants, and engineering application experience. Consistent with the preset pollution classification thresholds, these values ​​serve as benchmarks for assessing the presence and degree of pollution in the measured values ​​of each sensitive characteristic factor, ensuring the rationality of subsequent comparisons and calculations. The coupling inhibition coefficient, based on historical practice data and engineering application experience in soil pollution detection, and considering the interrelationships between sensitive characteristic factors under different pollution types, is a fixed value determined after multiple verifications and calibrations. It is used to adjust the degree of coupling influence between sensitive characteristic factors, avoiding the abnormal impact of a single characteristic factor on the calculation results.

[0053] The natural logarithm function is used by taking the value in parentheses as input and performing a natural logarithmic operation. The calculation process involves adding a very small positive number to the ratio of the measured value to the background reference value and then performing a natural logarithmic transformation on the result. This quantifies the degree of deviation between the measured value and the background reference value, providing basic data for subsequent suppression of extreme effects.

[0054] The extremely small positive number is chosen to avoid the inability to perform natural logarithmic calculations when the ratio of the measured value to the background reference value is 0. Its value is determined based on engineering application experience, selecting a fixed, small positive number to ensure smooth calculation and avoid errors. The pollution type adjustment factor is determined based on the identified pollution type of the soil samples. As mentioned earlier, pollution types are divided into three categories: single heavy metal pollution, compound heavy metal pollution, or organic-inorganic compound pollution. For different pollution types, corresponding adjustment values ​​are set based on the environmental hazard characteristics and diffusion patterns of that type of pollution to correct the impact of different pollution types on the calculation results of the comprehensive pollution index.

[0055] The natural exponential function is used by taking the value in parentheses as input for natural exponential calculation. The calculation process is to perform natural exponential transformation on the product of the negative coupling inhibition coefficient and the average of the absolute values ​​of the natural logarithms of the ratios of the measured values ​​of the three types of sensitive characteristic factors to the background reference values, plus a very small positive number. This suppresses the extreme influence of each characteristic factor deviating from the baseline value, making the calculation results more consistent with the actual pollution situation.

[0056] The nonlinear sensitivity index is derived from historical practical data and engineering application experience in soil pollution detection. It is a fixed value determined after multiple experimental verifications, combining the sensitivity of different sensitivity characteristic factors to pollution levels. It is used to adjust the nonlinear influence of the ratio of measured values ​​of each sensitivity characteristic factor to the background reference value, allowing the calculation results to more accurately reflect the actual pollution situation. The significance of the formula lies in systematically integrating and calculating the measured values, weighting coefficients, and other parameters of the sensitivity characteristic factors determined above, ultimately obtaining a comprehensive pollution index, achieving a quantitative assessment of the pollution level of soil samples. The formula compares the measured value of each sensitivity characteristic factor with the background reference value, quantifies its deviation from the benchmark using the nonlinear sensitivity index, multiplies it by the corresponding weighting coefficient, obtains the weighted influence value of each characteristic factor, and sums them to reflect the different importance of each characteristic factor.

[0057] By using natural logarithms and natural exponents, the extreme influence of deviations from baseline values ​​by individual characteristic factors is suppressed, avoiding the interference of abnormal values ​​of single characteristic factors on the accuracy of the overall results. Pollution type adjustment factors correct the impact of different pollution types on the calculation results, ensuring that the calculation results are consistent with the actual situation of the current pollution type. The final calculated comprehensive pollution index accurately reflects the overall pollution status of soil samples in the target area and forms a complete and coherent logical relationship with the pollution types identified above, the extracted sensitive characteristic factors, the determined pollution levels, and the established weight adjustment rules. The value ranges from 0 to 75, with higher values ​​indicating more severe soil pollution, providing an accurate and reasonable quantitative basis for assessing the degree of soil pollution.

[0058] First, identify the pollution type of the determined soil sample, which may be one of the following: single heavy metal pollution, compound heavy metal pollution, or organic-inorganic compound pollution. Then, extract three types of feature data from the multimodal feature matrix of the soil sample in the target area: peak activity of heavy metal ions, spectral absorption depth feature value, and change in spectral curvature. Combine the influence relationship of pollution type, eliminate feature data that is irrelevant to the current pollution type. For example, for single heavy metal pollution, only the peak activity of the ion corresponding to the single heavy metal is retained, and the peak activity data of other irrelevant heavy metals are excluded. All remaining peak activity of heavy metal ions, spectral absorption depth feature value, and change in spectral curvature related to the current pollution type are summarized and integrated to form the sensitivity feature factor corresponding to the pollution type. Ensure that the sensitivity feature factor only contains the three types of feature data closely related to the current pollution type, and that all data are derived from the multimodal feature matrix.

[0059] The preset pollution classification thresholds are determined based on historical practice data in the field of soil pollution detection, the environmental hazard levels of different pollutants, and engineering application experience. Corresponding classification thresholds are set for the heavy metal ion activity peak, spectral absorption depth characteristic value, and spectral characteristic curvature change among the sensitive characteristic factors. Specifically, the classification thresholds for the heavy metal ion activity peak are divided into four levels: unpolluted, slightly polluted, moderately polluted, and heavily polluted; the classification thresholds for the spectral absorption depth characteristic value are also divided into four levels; and the classification thresholds for the spectral characteristic curvature change are also divided into four levels. Each characteristic value in the integrated sensitive characteristic factors is compared one by one with the corresponding pollution classification threshold. Based on the threshold range of the characteristic value, the pollution level corresponding to each sensitive characteristic factor is determined. Finally, the pollution level corresponding to each of the sensitive characteristic factors is generated, ensuring that each sensitive characteristic factor has a unique and clear pollution level.

[0060] First, clearly define the specific pollutant types included in the currently identified pollution types. For example, single heavy metal pollution should specify the exact heavy metal pollutant, compound heavy metal pollution should specify the two or more heavy metal pollutants, and organic-inorganic compound pollution should specify the specific organic pollutants and heavy metal pollutants. Then, based on the degree of harm of different pollutants to the soil environment, the pollution diffusion rate, and the detection priority, assign weights to the sensitivity characteristic factors. The weight of sensitivity characteristic factors directly related to the main pollutant is set to 0.5, the weight of sensitivity characteristic factors related to the secondary pollutant is set to 0.3, and the weight of sensitivity characteristic factors that help characterize the degree of pollution is set to 0.2. The sum of the weights of all sensitivity characteristic factors is 1. For example, in single heavy metal pollution, the weight of the peak ion activity of the corresponding heavy metal is 0.5, the weight of the spectral absorption depth characteristic value is 0.3, and the weight of the spectral characteristic curvature change is 0.2. Organize this weight allocation method into a standardized weight adjustment rule, clarify the specific weight corresponding to each sensitivity characteristic factor, and ensure that the weight adjustment rule can be directly used for subsequent calculation of the contribution of individual pollution.

[0061] First, based on the pollution level corresponding to each sensitive characteristic factor, determine its corresponding individual pollution contribution. The contribution is 0 for no pollution level, 25 for light pollution level, 50 for moderate pollution level, and 75 for heavy pollution level. Then, according to the specific weight of each sensitive characteristic factor in the weighting adjustment rules, associate the individual pollution contribution of each sensitive characteristic factor with its corresponding weight. Multiply the individual pollution contribution of each sensitive characteristic factor by its corresponding weight to obtain the weighted pollution contribution of each sensitive characteristic factor. Finally, sum the weighted pollution contributions of all sensitive characteristic factors. The sum is the comprehensive pollution index of the soil sample in the target area. The comprehensive pollution index ranges from 0 to 75, with higher values ​​indicating more severe soil pollution. This ensures that the comprehensive pollution index accurately reflects the overall pollution status of the soil sample and forms a complete and coherent logical connection with the previous pollution types, sensitive characteristic factors, pollution levels, and weighting adjustment rules.

[0062] The beneficial effects include the ability to clearly identify the pollution type of soil samples, extract sensitive characteristic factors closely related to the pollution type, determine the pollution level corresponding to each characteristic factor by setting a classification threshold, formulate reasonable weight adjustment rules based on the pollutant-related characteristics, and then calculate a comprehensive pollution index that can accurately reflect the overall pollution status of the soil sample and is logically coherent, thus clearly presenting the degree of soil pollution.

[0063] S5. By coordinating and integrating the numerical ranges of pollution types and comprehensive pollution indices, the soil pollution level assessment results of soil samples in the target area are obtained.

[0064] In this embodiment of the invention, the step of synergistically integrating the numerical ranges of pollution type and comprehensive pollution index to obtain the soil pollution level assessment result of the soil sample in the target area includes: Based on the pollution type and comprehensive pollution index, the preset pollution level interval mapping table is searched and filtered to obtain the graded interval set of soil samples in the target area. The target interval of the soil sample in the target area is obtained by sequentially comparing the value of the comprehensive pollution index with the endpoint value of the concentrated interval of the classification interval. The pollution level labels associated with the target interval are used as the soil pollution level assessment results for soil samples in the target area; however, different pollution level labels are allowed for the same comprehensive pollution index value under different pollution types.

[0065] The pre-defined pollution level interval mapping table is developed in advance based on historical practice data in the field of soil pollution detection, the environmental hazard levels of different pollutants, and engineering application experience. The table clearly distinguishes three pollution types: single heavy metal pollution, compound heavy metal pollution, and organic-inorganic compound pollution. Each pollution type corresponds to multiple continuous and non-overlapping comprehensive pollution index intervals. Each interval is associated with a unique pollution level label, and under different pollution types, the same comprehensive pollution index interval can be associated with different pollution level labels. During the search and screening process, the pollution type of the soil sample in the target area is first identified, and then the specific value of the comprehensive pollution index calculated from the soil sample is determined. Subsequently, the pollution type is used as the primary matching condition to find an entry in the pre-defined pollution level interval mapping table that is completely consistent with the pollution type. Under that entry, all comprehensive pollution index intervals that can contain the current comprehensive pollution index value are then selected. The set formed by integrating all the selected comprehensive pollution index intervals is the graded interval set of the soil sample in the target area, ensuring that each interval in the graded interval set corresponds to the pollution type of the current soil sample, and that the value range of each interval can cover the current comprehensive pollution index value.

[0066] Each comprehensive pollution index interval in the hierarchical interval set has clearly defined endpoint values, namely the minimum and maximum values ​​of the interval. All endpoint values ​​of all intervals are arranged in ascending order, with no overlap or omission between intervals, completely covering the entire range of comprehensive pollution index values ​​from 0 to 75. During sequential comparison, the endpoint values ​​of all intervals in the hierarchical interval set are first extracted and arranged in ascending order. Then, the specific comprehensive pollution index values ​​of the soil samples in the target area are extracted. Starting from the minimum endpoint value after arrangement, the comprehensive pollution index values ​​are compared with the minimum and maximum values ​​of each interval in turn. The comparison follows the judgment condition that "the comprehensive pollution index value is greater than or equal to the minimum value of the interval and less than or equal to the maximum value of the interval". When an interval is found whose minimum value is less than or equal to the comprehensive pollution index value and whose maximum value is greater than or equal to the comprehensive pollution index value, the comparison is stopped immediately. This interval is the target interval for the soil samples in the target area, ensuring that each comprehensive pollution index value can find a unique target interval, and there is no situation where one value corresponds to multiple intervals or there is no corresponding interval.

[0067] In the pre-defined pollution level range mapping table, each comprehensive pollution index range is pre-associated with a specific pollution level label. These labels are categorized into four levels: unpolluted, lightly polluted, moderately polluted, and heavily polluted. This association is pre-determined based on the environmental hazard level, pollution diffusion rate, and detection priority of different pollution types. It clarifies that even with the same comprehensive pollution index value and within the same range, different pollution types may have different associated pollution level labels. During implementation, after determining the target range, the corresponding pollution level label in the pre-defined pollution level range mapping table is directly retrieved. This label is then used directly as the soil pollution level assessment result for the soil sample in the target area. This ensures a one-to-one correspondence between the assessment result and the target range, and a complete and coherent logical connection with the current soil sample's pollution type and comprehensive pollution index. Furthermore, it strictly adheres to the requirement that the same comprehensive pollution index value can correspond to different pollution level labels for different pollution types, guaranteeing the relevance and accuracy of the assessment results.

[0068] The beneficial effect is that a pollution level interval mapping table is developed based on historical practice data in the field of soil pollution detection, the environmental hazard degree of different pollutants, and engineering application experience. By clearly defining the pollution type and comprehensive pollution index of soil samples, a set of graded intervals is obtained through retrieval and screening. Then, a unique target interval is determined through sequential comparison. Finally, the pollution level label corresponding to the target interval is obtained as the soil pollution level assessment result. This can ensure that the assessment results are targeted and accurate, ensure that the assessment process is reproducible, and form a complete and coherent logical relationship with the pollution type and comprehensive pollution index of the soil sample, avoiding assessment deviations or situations where there are no corresponding assessment results.

[0069] like Figure 2The diagram shown is a functional block diagram of a comprehensive soil pollution detection system provided in an embodiment of the present invention.

[0070] The comprehensive soil pollution detection system described in this invention can be installed in an electronic device. Depending on the functions implemented, the comprehensive soil pollution detection system may include a data preprocessing module 101, a multimodal feature fusion module 102, a pollution type identification module 103, a comprehensive pollution index assessment module 104, and a pollution level determination module 105. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, and are stored in the memory of the electronic device.

[0071] In this embodiment, the functions of each module / unit are as follows: The data preprocessing module 101 is used to denoise and filter the spectral reflectance data and ion activity detection data of the soil samples in the target area, and to normalize the denoised data to obtain the spectral feature vector and ion activity feature vector of the soil samples in the target area. The multimodal feature fusion module 102 is used to fuse spectral feature vectors and ion activity feature vectors in a feature layer to construct a multimodal feature matrix of soil samples in the target area. The pollution type identification module 103 is used to classify the pollution types of soil samples in the target area based on the multimodal feature matrix, and obtain the pollution type of the soil samples. The comprehensive pollution index assessment module 104 is used to correlate and assess the pollution type with the sensitive feature factors in the multimodal feature matrix based on the preset pollution classification threshold, so as to obtain the comprehensive pollution index of the soil sample. The pollution level determination module 105 is used to integrate the numerical ranges of pollution type and comprehensive pollution index to obtain the soil pollution level assessment results of soil samples in the target area.

[0072] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0073] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0074] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0075] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0076] This application provides a method for acquiring and processing relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A comprehensive detection method for soil pollution, characterized in that, The method includes: S1. Denoise filtering is performed on the spectral reflectance data and ion activity detection data of the soil samples in the target area, and the denoised data is normalized to obtain the spectral feature vector and ion activity feature vector of the soil samples in the target area. S2. The spectral feature vector and the ion activity feature vector are fused in the feature layer to construct a multimodal feature matrix of the soil sample in the target area; S3. Based on the multimodal feature matrix, the pollution types of soil samples in the target area are classified to obtain the pollution type of the soil samples; S4. Based on the preset pollution classification threshold, the pollution type is correlated with the sensitive feature factors in the multimodal feature matrix to obtain the comprehensive pollution index of the soil sample. S5. By coordinating and integrating the numerical ranges of pollution types and comprehensive pollution indices, the soil pollution level assessment results of soil samples in the target area are obtained.

2. The comprehensive soil pollution detection method as described in claim 1, characterized in that, The process of denoising and filtering the spectral reflectance data and ion activity detection data of soil samples in the target area, and then normalizing the denoised data to obtain the spectral feature vector and ion activity feature vector of the soil samples in the target area, includes: The reflectance values ​​at each wavelength point in the spectral reflectance data of the soil sample in the target area are statistically averaged by performing adjacent arithmetic mean calculations to obtain the reflectance value at each wavelength point. The ion activity values ​​of the sampling points in the ion activity detection data of the soil samples in the target area are proportionally projected to obtain the ion activity values ​​of the sampling points. Thresholds were set for reflectance and ion activity values, and noise was removed from the defined data to obtain the denoised ion activity and reflectance values ​​of the soil samples in the target area. The denoised ion activity values ​​and denoised reflectance values ​​are linearly mapped to obtain the spectral feature vector and ion activity feature vector of the soil sample in the target area.

3. The comprehensive soil pollution detection method as described in claim 1, characterized in that, The process of fusing spectral feature vectors and ion activity feature vectors at a feature layer to construct a multimodal feature matrix for soil samples in the target area includes: The correlation strength between the spectral feature values ​​at each band position in the spectral feature vector and the activity feature values ​​of each heavy metal ion in the ion activity feature vector is coupled and calculated. Based on the calculation results, the target spectral bands of the soil samples in the target area are selected. The spectral feature values ​​corresponding to the target spectral band are concatenated with all ion activity feature values ​​in the ion activity feature vector to obtain the combined feature sequence of the target spectral band. Based on the sampling location and sampling time of the soil samples in the target area, the combined feature sequences are arranged in a regular manner to obtain the multimodal feature matrix of the soil samples in the target area.

4. The comprehensive soil pollution detection method as described in claim 3, characterized in that, The formula for calculating the correlation strength is as follows: ； In the formula, For the first The position of the first band and the first The correlation strength between heavy metal ions The spectral eigenvector is the first... Spectral characteristic values ​​at each band position The ion activity eigenvector is the first... Characteristic values ​​of ion activity for several heavy metal ions. This is the arithmetic mean of the spectral eigenvalues ​​at all band positions in the spectral eigenvector. This is the arithmetic mean of the ion activity eigenvalues ​​of all heavy metal ions in the ion activity eigenvector. It is a very small positive number.

5. The comprehensive soil pollution detection method as described in claim 3, characterized in that, The step involves systematically arranging the combined feature sequences according to the sampling location and time of soil samples in the target area to obtain a multimodal feature matrix of soil samples in the target area, including: The combined feature sequences corresponding to different sampling times at the same sampling location in the soil samples of the target area are arranged sequentially according to the sampling time to form a temporal feature submatrix of the sampling location; The temporal feature submatrix is ​​adaptively spatially fused according to the spatial distribution order of the sampling locations within the target area to obtain the multimodal feature matrix of the soil samples in the target area.

6. The comprehensive soil pollution detection method as described in claim 1, characterized in that, The method of classifying and identifying pollution types in soil samples from the target area based on a multimodal feature matrix yields the pollution types of the soil samples, including: The similarity assessment between the multimodal feature matrix and the preset polluted soil feature benchmark is performed to obtain the degree of feature difference of the multimodal feature matrix; Based on the degree of difference in characteristics, the pollution type corresponding to the selected polluted soil characteristic benchmarks is taken as the pollution type of the soil sample. The pollution types include single heavy metal pollution, compound heavy metal pollution, or organic-inorganic compound pollution.

7. The comprehensive soil pollution detection method as described in claim 1, characterized in that, The process involves correlating pollution types with sensitive feature factors in a multimodal feature matrix based on a preset pollution classification threshold to obtain a comprehensive pollution index for the soil sample, including: Based on the influence relationship of pollution types, the peak activity of heavy metal ions, the characteristic value of spectral absorption depth, and the change in spectral characteristic curvature in the multimodal characteristic matrix are integrated into the sensitivity characteristic factors of pollution types. The sensitivity feature factors are mapped to preset pollution classification thresholds to generate the pollution level of the sensitivity feature factors; Based on the specific pollutant types identified in the pollution types, weights are assigned to the sensitivity characteristic factors to construct weight adjustment rules for the sensitivity characteristic factors. According to the weighting adjustment rules, the individual pollution contribution of sensitive characteristic factors is superimposed and summarized with the pollution level to obtain the comprehensive pollution index of the soil sample.

8. The comprehensive soil pollution detection method as described in claim 7, characterized in that, The formula for calculating the comprehensive pollution index is as follows: ； In the formula, The comprehensive pollution index, This is a sensitivity characteristic factor index. A value of 1 indicates the peak activity of heavy metal ions, a value of 2 indicates the characteristic value of spectral absorption depth, and a value of 3 indicates the change in spectral characteristic curvature. For the first The weight coefficients of each sensitivity feature factor under the current pollution type. For the first Measured values ​​of each sensitivity characteristic factor For the first Background reference values ​​for each sensitivity feature factor The preset coupling suppression coefficient, It is the natural logarithm function. It is a very small positive number. As a pollution type modulator, It is a natural exponential function. This is a preset nonlinear sensitivity index.

9. The comprehensive soil pollution detection method as described in claim 1, characterized in that, The method of synergistically integrating the numerical ranges of pollution type and comprehensive pollution index to obtain the soil pollution level assessment results of soil samples in the target area includes: Based on the pollution type and comprehensive pollution index, the preset pollution level interval mapping table is searched and filtered to obtain the graded interval set of soil samples in the target area. The target interval of the soil sample in the target area is obtained by sequentially comparing the value of the comprehensive pollution index with the endpoint value of the concentrated interval of the classification interval. The pollution level labels associated with the target interval are used as the soil pollution level assessment results for soil samples in the target area; however, different pollution level labels are allowed for the same comprehensive pollution index value under different pollution types.

10. A comprehensive soil pollution detection system, characterized in that, The system for implementing the comprehensive soil pollution detection method according to claim 1 includes: The data preprocessing module is used to denoise and filter the spectral reflectance data and ion activity detection data of soil samples in the target area, and to normalize the denoised data to obtain the spectral feature vector and ion activity feature vector of the soil samples in the target area. The multimodal feature fusion module is used to fuse spectral feature vectors and ion activity feature vectors at the feature layer to construct a multimodal feature matrix of soil samples in the target area. The pollution type identification module is used to classify the types of pollution in soil samples in the target area based on a multimodal feature matrix, and to obtain the pollution type of the soil samples. The comprehensive pollution index assessment module is used to correlate pollution types with sensitive feature factors in the multimodal feature matrix based on preset pollution classification thresholds to obtain the comprehensive pollution index of soil samples. The pollution level determination module is used to integrate the numerical ranges of pollution type and comprehensive pollution index to obtain the soil pollution level assessment results of soil samples in the target area.

Images