Groundwater pollution tracing method and device based on suspended particle single-particle mass spectrometry coupled with machine learning, electronic equipment and storage medium

Abstract

The present application relates to the technical field of water quality monitoring, and discloses a suspended matter single-particle mass spectrometry coupled machine learning groundwater pollution tracing method and device, electronic equipment and storage medium, wherein the groundwater pollution tracing method comprises the following steps: collecting source samples of potential pollution sources and receptor samples of contaminated areas, obtaining particle mass spectrum information by using single-particle mass spectrometry detection technology, constructing a pollution source mass spectrum fingerprint library containing a source attribution subset, each subset being determined by a general mass spectrum category and a potential pollution source, and storing average mass spectrum characteristics; determining the composition of the pollution source according to the matching of the mass spectrum information of the receptor sample particles and the fingerprint library, and verifying and correcting in combination with hydrogeological conditions. The method can accurately identify the source differences of particulate pollutants by characterizing the chemical characteristics of suspended particulate matter through high-resolution mass spectrometry, effectively solves the limitations that traditional methods cannot distinguish when multiple pollution sources coexist, and improves the accuracy of groundwater pollution tracing.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring technology, specifically to a method, device, electronic equipment, and storage medium for tracing groundwater pollution sources using single-particle mass spectrometry coupled with machine learning. Background Technology

[0002] Groundwater, as a vital freshwater resource, is crucial for human health and the maintenance of ecosystems. However, groundwater is facing increasingly severe pollution problems, seriously threatening residents' health and ecological safety. Because aquifers are located beneath the surface, the complex migration pathways of pollutants make tracing the sources of groundwater pollution an extremely challenging task.

[0003] Traditional methods for tracing groundwater pollution sources mainly include hydrochemical methods, isotope tracing methods, microbial tracing methods, and tracer methods. Hydrochemical analysis understands the overall hydrochemical characteristics by measuring the concentration of major ions (such as calcium, magnesium, sodium, potassium, chloride, sulfate, and bicarbonate) and identifies potential sources based on characteristic ion ratios. However, multiple sources may have similar chemical characteristics, interfering with source differentiation. Isotope analysis utilizes stable isotopes of water (such as δ¹⁰) 8 O and δ²H are used to trace the source and movement of groundwater and identify mixing between different sources. Stable isotope analysis of contaminants helps distinguish between natural and anthropogenic sources and identify specific sources of pollution. However, isotopic ranges may overlap between different sources. Microbial source tracing (MST) identifies the source of fecal contamination by detecting host-specific microbial markers (such as those from humans, livestock, and wildlife). Commonly used techniques include PCR and qPCR. MST is particularly suitable for tracing contamination from sewage, animal feces, and other fecal sources. However, MST may not be applicable to all types of contamination and may be affected by the survival and migration of microorganisms in groundwater. Tracer techniques trace water flow paths and velocities and identify links between potential sources and points of contamination by introducing artificial tracers (such as fluorescent dyes, salts, radioactive compounds, and microspheres) into groundwater systems. Artificial tracers can provide direct evidence of contaminant transport pathways. However, tracer techniques are costly and time-consuming to apply and may be limited by ethical and regulatory considerations, making them unsuitable for all sites and types of contamination.

[0004] Traditional methods provide information about the overall properties of groundwater and the presence of contaminants. However, many contaminants, especially metals, persistent organic pollutants (POPs), and emerging pollutants such as microplastics and engineered nanomaterials, are often bound to or exist in other forms with particulate matter or colloids in groundwater. Therefore, traditional methods may overlook the unique source information carried by particulate pollutants or colloids as important migration carriers. Particularly when multiple pollution sources simultaneously cause groundwater pollution, their individual particulate characteristics mix together, making it difficult for traditional methods to accurately distinguish between different sources. Summary of the Invention

[0005] This invention provides a method, device, electronic device, and storage medium for tracing groundwater pollution sources using single-particle mass spectrometry coupled with machine learning, to solve the problem that traditional methods are unable to accurately distinguish the sources of particulate or colloidal pollutants when multiple pollution sources coexist.

[0006] In a first aspect, the present invention provides a groundwater pollution source tracing method based on suspended particulate mass spectrometry coupled with machine learning. The method includes the following steps: acquiring multiple source samples, wherein the source samples are collected from potential pollution sources; performing single-particle mass spectrometry detection on each source sample to obtain mass spectrometry information of multiple particles in each source sample; constructing a pollution source mass spectrometry fingerprint database based on the mass spectrometry information of multiple source samples, wherein the pollution source mass spectrometry fingerprint database includes multiple fingerprint database entries, each fingerprint database entry corresponds to a source attribution subset, the source attribution subset is jointly determined by a general mass spectrometry category and a potential pollution source, and includes the average mass spectrometry features of all particles in the source attribution subset; acquiring a receptor sample, wherein the receptor sample is a suspended particulate matter sample in groundwater of a polluted area; performing single-particle mass spectrometry detection on the receptor sample to obtain mass spectrometry information of each type of particle in the receptor sample; and determining the pollution source composition of the receptor sample based on the mass spectrometry information of each type of particle in the receptor sample and the pollution source mass spectrometry fingerprint database.

[0007] The groundwater pollution source tracing method coupled with single-particle mass spectrometry and machine learning provided by this invention uses single-particle mass spectrometry to perform high-resolution characterization of suspended particulate matter in potential pollution sources and polluted groundwater, and constructs a pollution source mass spectrometry fingerprint library containing a subset of source attribution. This method can accurately identify the source characteristics of particulate matter and effectively overcome the limitations of traditional methods in distinguishing the source when multiple pollution sources coexist or when pollutants exist in particulate or colloidal states.

[0008] In some optional implementations, constructing a pollution source mass spectrometry fingerprint database based on mass spectrometry information from multiple source samples includes: selecting from multiple particles contained in source samples belonging to the same potential pollution source to obtain effective source spectrum particles corresponding to each potential pollution source; merging all effective source spectrum particles into a particle set and performing global clustering on the particle set to obtain multiple general mass spectrometry categories; wherein each general mass spectrometry category contains particles from one or more potential pollution sources; for each general mass spectrometry category, according to the potential pollution source to which each particle in the general mass spectrometry category belongs, dividing the general mass spectrometry category into multiple source attribution subsets, each source attribution subset corresponding to a combination of a general mass spectrometry category and a potential pollution source; associating and storing each source attribution subset with its corresponding potential pollution source and the average mass spectrometry features of all particles in that subset to form a fingerprint database entry; traversing all general mass spectrometry categories and their divided source attribution subsets to obtain the pollution source mass spectrometry fingerprint database.

[0009] This implementation method first effectively screens particles from the same pollution source, then performs global clustering to form a universal mass spectrometry category, and further subdivides the source-attribution subsets according to the pollution source. This not only preserves the subtle differences between different pollution sources under the same particle chemical type, but also effectively avoids the problem of category redundancy or cross-confusion caused by directly classifying by source. The pollution source mass spectrometry fingerprint library constructed in this way has both chemical commonality and source specificity, which significantly improves the ability to distinguish multi-source particulate matter in complex mixed pollution scenarios and the accuracy of source tracing.

[0010] In some optional implementations, selecting from multiple particles contained in a source sample belonging to the same potential pollution source to obtain effective source spectrum particles corresponding to each potential pollution source includes: performing a first-level clustering on the mass spectrometry information of multiple particles contained in the source sample belonging to the same potential pollution source to obtain multiple sub-particle categories corresponding to each potential pollution source; filtering the multiple sub-particle categories obtained after the first-level clustering for each potential pollution source, and retaining only the particle categories whose particle number ratio exceeds a preset threshold and have significant chemical characteristics to obtain the effective source spectrum particles for each potential pollution source.

[0011] This implementation method effectively eliminates noise particles and non-representative components by performing first-level clustering on particle mass spectrometry data of the same potential pollution source and screening based on the proportion of quantity and the significance of chemical characteristics, while retaining effective source spectrum particles with statistical significance and source-specific chemical identifiers. This not only improves the stability and representativeness of pollution source fingerprint features, but also enhances the reliability and discriminativeness of subsequent fingerprint database construction.

[0012] In some optional implementations, determining the contamination source composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the contamination source mass spectrometry fingerprint database includes: for particle j in the receptor sample, calculating the similarity between the mass spectrometry information of particle j and the average mass spectrometry feature corresponding to each fingerprint database entry in the contamination source mass spectrometry fingerprint database to obtain multiple source matching similarity scores; comparing each source matching similarity score with a preset similarity threshold; if at least one source matching similarity score is greater than the similarity threshold, then particle j is assigned to the potential contamination source with the highest similarity score; if all source matching similarity scores are less than or equal to the similarity threshold, then particle j is classified as an unknown source or background source; based on the assignment results of all particles in the receptor sample, calculating the proportion of particles corresponding to each potential contamination source to determine the contamination source composition of the receptor sample.

[0013] This implementation method quantitatively compares the average mass spectrometry characteristics of receptor particles with those of each source subset in the fingerprint database, and combines this with a preset threshold to achieve accurate source attribution of particles. This not only effectively identifies the main pollution sources, but also reasonably distinguishes unknown or background particles, avoiding misjudgments. Furthermore, by statistically analyzing the particle proportions corresponding to each pollution source, the contribution of different pollution sources to the receptor sample is objectively quantified, significantly improving the accuracy, interpretability, and practicality of groundwater pollution tracing results.

[0014] In some optional implementations, determining the contaminant composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the contaminant source mass spectrometry fingerprint database includes: training a preset model based on the mass spectrometry information of each particle in the contaminant source mass spectrometry fingerprint database and its corresponding potential contaminant source label to obtain a contaminant source classification model; converting the mass spectrometry information of each particle in the receptor sample into a feature vector of the same dimension as the training data and inputting it into the contaminant source classification model to obtain the potential contaminant source category to which each particle in the receptor sample belongs; and based on the classification results of all particles in the receptor sample, calculating the proportion of particles corresponding to each potential contaminant source to determine the contaminant composition of the receptor sample.

[0015] This implementation method trains a dedicated pollution source classification model using labeled data from a pollution source mass spectrometry fingerprint database, automatically mapping the mass spectrometry information of receptor particles to the corresponding pollution source category, thus achieving a high-throughput and intelligent source apportionment process.

[0016] In some optional implementations, after determining the pollution source composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the pollution source mass spectrometry fingerprint database, the method further includes: verifying the rationality of the pollution source composition by combining hydrogeological conditions and environmental background information, and correcting the pollution source composition based on the verification results.

[0017] After initially determining the composition of pollution sources based on a mass spectrometry fingerprint database, this implementation method further integrates hydrogeological conditions and environmental background information to conduct multi-dimensional verification and necessary corrections to the source tracing results. This effectively eliminates unreasonable attributions caused by complex particle migration paths, overlapping source characteristics, or model misjudgments, significantly improving the scientific validity, credibility, and practical applicability of the pollution source analysis results.

[0018] Secondly, the present invention also provides a groundwater pollution source tracing device based on single-particle mass spectrometry coupled with machine learning for suspended solids. The device includes a first acquisition module, a mass spectrometry information acquisition module, a fingerprint database construction module, a second acquisition module, and a source tracing module. The first acquisition module is used to acquire multiple source samples, wherein the source samples are collected from potential pollution sources. The mass spectrometry information acquisition module is used to perform single-particle mass spectrometry detection on each source sample to obtain the mass spectrometry information of multiple particles in each source sample. The fingerprint database construction module is used to construct a pollution source mass spectrometry fingerprint database based on the mass spectrometry information of multiple source samples, wherein the pollution source mass spectrometry... The fingerprint database includes multiple fingerprint entries, each corresponding to a source attribution subset. The source attribution subset is determined by a general mass spectrometry category and a potential pollution source, and includes the average mass spectrometry characteristics of all particles in the source attribution subset. The second acquisition module is used to acquire receptor samples, which are suspended particulate matter samples from groundwater in the contaminated area. The mass spectrometry information acquisition module is also used to perform single-particle mass spectrometry detection on the receptor samples to obtain the mass spectrometry information of each type of particle in the receptor samples. The source tracing module is used to determine the pollution source composition of the receptor samples based on the mass spectrometry information of each type of particle in the receptor samples and the pollution source mass spectrometry fingerprint database.

[0019] Thirdly, the present invention also provides an electronic device, including a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the groundwater pollution source tracing method of suspended solids single particle mass spectrometry coupled with machine learning described in the first aspect or any of its corresponding embodiments.

[0020] Fourthly, the present invention also provides a computer-readable storage medium storing computer instructions, which are used to cause a computer to execute the groundwater pollution source tracing method of suspended solids single-particle mass spectrometry coupled with machine learning according to the first aspect or any corresponding embodiment described above.

[0021] Fifthly, the present invention also provides a computer program product, including computer instructions for causing a computer to execute the groundwater pollution source tracing method using suspended solids single-particle mass spectrometry coupled with machine learning as described in the first aspect or any corresponding embodiment. Attached Figure Description

[0022] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of an application scenario according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the first process of the groundwater pollution source tracing method based on suspended single-particle mass spectrometry coupled with machine learning according to an embodiment of the present invention. Figure 3 This is a schematic diagram of the second process of the groundwater pollution source tracing method based on suspended single-particle mass spectrometry coupled with machine learning according to an embodiment of the present invention. Figure 4 This is a schematic diagram of suspended particulate matter detection in water according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the single-particle mass spectrometry principle according to an embodiment of the present invention; Figure 6 This is a flowchart illustrating the groundwater pollution source tracing method based on single-particle mass spectrometry coupled with machine learning according to the present invention. Figure 7 This is a schematic diagram of the positive and negative spectral characteristics of rainwater pool sample particles according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the positive and negative spectral characteristics of wastewater treatment plant sample particles according to an embodiment of the present invention; Figure 9 This is a schematic diagram of the positive and negative spectral characteristics of groundwater sample 2A01 particles according to an embodiment of the present invention; Figure 10 This is a schematic diagram of the positive and negative spectral characteristics of groundwater sample GW2 particles according to an embodiment of the present invention; Figure 11 This is a schematic diagram of the spectral features of six types of particles obtained by ART-2a clustering according to an embodiment of the present invention; Figure 12 This is a schematic diagram showing the proportion of particle types in each sample according to an embodiment of the present invention; Figure 13 This is a schematic diagram comparing the source tracing results of the wastewater treatment plant with the actual values ​​according to an embodiment of the present invention; Figure 14 This is a structural block diagram of a groundwater pollution tracing device based on suspended single-particle mass spectrometry coupled with machine learning according to an embodiment of the present invention. Figure 15 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention; The components include: 1. Aerosol generator; 2. Drying tube; 3. Aerodynamic lens; 4. First photomultiplier tube; 5. Second photomultiplier tube; 6. First diameter measuring laser; 7. Second diameter measuring laser; 8. Reflection zone; 9. Acceleration zone; 10. Resolution / ionization laser; 11. Microchannel plate; 12. Ellipsoidal mirror; 13. Sample inlet orifice. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0026] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0027] As an optional application scenario of this invention, such as Figure 1 As shown, a groundwater pollution tracing system based on single-particle mass spectrometry coupled with machine learning may include at least one terminal device and at least one server. Figure 1 The system is illustrated in the example, which includes a computer 101, a mobile terminal 102, and a server 103, and the terminal devices such as the computer 101 and the mobile terminal 102 are connected to the server 103 through a network 110.

[0028] Specifically, the terminal device can be a smartphone, tablet, laptop, PDA, desktop computer, game console, smart TV, smart wearable device, in-vehicle terminal, VR (Virtual Reality) device, AR (Augmented Reality) device, etc. Server 103 can be a standalone physical server, a server cluster, a distributed system, or a cloud server providing cloud services. Network 110 can be a wired or wireless network, examples of which include, but are not limited to, the Internet, corporate intranet, local area network, wide area network, mobile communication network, and combinations thereof.

[0029] Suspended particulate matter (SPM) refers to solid particles ranging in size from colloidal to approximately 100 micrometers suspended in water. These particles are composed of inorganic and organic matter and can originate from weathering and erosion of soil and rocks, as well as from waste generated by human activities. The large specific surface area of ​​SPMs makes them effective carriers of various pollutants. Heavy metals, organic pollutants, nutrients, and even pathogenic microorganisms are easily adsorbed onto the surface of SPMs and diffuse with the migration of particles. This adsorption allows pollutants to migrate over greater distances and at higher concentrations than in their dissolved state. Because SPMs can interact with and transport various pollutants, they can serve as potential pollution indicators. Analyzing the composition of SPMs helps us understand potential pollution sources.

[0030] Single-particle aerosol mass spectrometry (SPMS) refers to a series of mass spectrometry techniques capable of real-time or near-real-time chemical analysis of individual particles in air or liquids. The basic principle of SPMS involves introducing individual particles from an aerosol or liquid sample (such as groundwater) into a mass spectrometer, then ionizing the individual particles using techniques such as laser ablation or inductively coupled plasma (ICP). The generated ions are then analyzed by a mass analyzer to determine their mass-to-charge ratio, providing information about the elemental and / or molecular composition of the individual particles. Finally, the ions are detected to obtain the mass spectrum of each particle, thus acquiring real-time, online information on the particle size and chemical composition, giving each particle a unique "chemical fingerprint."

[0031] SPMS analysis of suspended particulate matter (SPM) in groundwater is an effective way to trace pollution sources. SPMS can reveal the elemental composition of inorganic particles, helping to link them to specific pollution sources. For example, the detection of a specific heavy metal in a single particle may point to industrial emissions. For organic particles, SPMS can provide information about their molecular composition, thus distinguishing between natural organics and synthetic pollutants such as microplastics or petroleum products. Different pollution sources may release particles with different elemental ratios or unique organic compounds, which can be identified by SPMS. SPMS's high resolution and specificity in characterizing the chemical composition of individual particles provide a powerful tool for fingerprinting groundwater pollution sources. By identifying unique markers within these particles, we can link pollution to sources with greater certainty than with holistic analysis. Comparing the characteristics of SPM in groundwater samples from different locations or at different times helps to trace the movement of pollution plumes and may identify point sources of pollutants. Especially when multiple pollution sources simultaneously cause groundwater pollution, their individual particulate characteristics may be mixed together. Conventional analysis provides a complex profile of a multi-particle mixture, making it difficult to distinguish the contributions of each pollution source. Single-particle mass spectrometry (SPS) offers the possibility of identifying and quantifying different particle groups by analyzing each particle individually, even in mixed samples, where each particle group may be associated with a specific source of contamination. This represents a significant advancement compared to existing methods discussed in the literature that struggle to resolve the sources of multiple chemical pollutants.

[0032] However, SPMS generates a large amount of high-dimensional spectral data containing complex patterns and noise, making manual interpretation extremely difficult and inefficient. Machine learning (ML), as a powerful data analysis tool, excels at identifying patterns, classifying, and clustering from high-dimensional and complex data. Applying ML algorithms to SPMS data analysis can automate particle classification, pollution source clustering and identification, and contribution rate calculation for massive particle mass spectrometry data, significantly improving the accuracy and efficiency of source tracing.

[0033] Based on this, according to an embodiment of the present invention, a groundwater pollution source tracing method based on suspended single-particle mass spectrometry coupled with machine learning is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0034] This embodiment provides a groundwater pollution source tracing method using suspended solid single-particle mass spectrometry coupled with machine learning, which can be used with computer equipment. Figure 2 This is a flowchart of the first method for tracing groundwater pollution sources using single-particle mass spectrometry coupled with machine learning according to an embodiment of the present invention, as follows: Figure 2As shown, the process includes the following steps: Step S201: Obtain multiple source samples, wherein the source samples are collected from potential pollution sources.

[0035] Among them, source samples refer to particulate matter samples collected from various sources that may cause groundwater pollution; potential pollution sources refer to sources in the survey area that may release pollutants into groundwater.

[0036] Step S202: Perform single-particle mass spectrometry detection on each source sample to obtain the mass spectrometry information of multiple particles in each source sample.

[0037] Specifically, a single-particle aerosol mass spectrometer can be used to perform single-particle mass spectrometry detection on the particulate matter in each source sample to obtain mass spectrometry information of multiple particles.

[0038] Step S203: Construct a pollution source mass spectrometry fingerprint database based on the mass spectrometry information of multiple source samples. The pollution source mass spectrometry fingerprint database includes multiple fingerprint database entries. Each fingerprint database entry corresponds to a source attribution subset. The source attribution subset is jointly determined by a general mass spectrometry category and a potential pollution source, and includes the average mass spectrometry features of all particles in the source attribution subset.

[0039] Step S204: Obtain a receptor sample, wherein the receptor sample is a suspended particulate matter sample from the groundwater in the contaminated area.

[0040] Among them, the receptor sample refers to a water sample containing suspended particulate matter collected from a groundwater contaminated area.

[0041] Step S205: Perform single-particle mass spectrometry detection on the receptor sample to obtain the mass spectrometry information of each particle in the receptor sample.

[0042] Specifically, a single-particle aerosol mass spectrometer can be used to perform single-particle mass spectrometry detection on particulate matter in the receptor sample to obtain mass spectrometry information for each type of particle in the receptor sample.

[0043] Step S206: Determine the contamination source composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the contamination source mass spectrometry fingerprint database.

[0044] The groundwater pollution source tracing method coupled with single-particle mass spectrometry and machine learning provided in this embodiment uses single-particle mass spectrometry to perform high-resolution characterization of suspended particulate matter in potential pollution sources and polluted groundwater, and constructs a pollution source mass spectrometry fingerprint library containing a subset of source attribution. This method can accurately identify the source characteristics of particulate matter and effectively overcome the limitations of traditional methods in distinguishing the source when multiple pollution sources coexist or when pollutants exist in particulate or colloidal states.

[0045] This embodiment provides a groundwater pollution source tracing method using suspended solid single-particle mass spectrometry coupled with machine learning, which can be used with computer equipment. Figure 3 This is a second flowchart of the groundwater pollution source tracing method based on single-particle mass spectrometry coupled with machine learning according to an embodiment of the present invention, as shown below. Figure 3 As shown, the process includes the following steps: Step S301: Obtain multiple source samples, wherein the source samples are collected from potential pollution sources.

[0046] Specifically, groundwater sampling and testing includes the following steps: 1. Analysis of the pollution status of contaminated groundwater Analyze the survey results of the polluted area, summarize the types and concentrations of pollutants exceeding the standards, the spatial distribution of the locations exceeding the standards, and determine the main pollution indicators and sampling scope based on the historical groundwater monitoring data of the target points.

[0047] 2. Analysis of potential pollution sources in the surrounding area Before conducting sampling and source tracing analysis, the hydrogeological conditions of the contaminated area and its surroundings, as well as the distribution of potential pollution sources, must be identified. The hydrogeological conditions of the site (such as aquifer structure, groundwater flow direction and velocity), the location of potential pollution sources, and their upstream and downstream spatial relationship with the contaminated site in terms of groundwater flow direction should be comprehensively considered. Based on this, a list of source tracing comparison targets should be compiled. Since sites requiring pollution source tracing generally have already undergone preliminary groundwater environmental surveys, information such as aquifer structure and groundwater flow direction is usually relatively complete and can be directly obtained through data collection. By analyzing the groundwater flow direction, the upstream and downstream locations of the contaminated area's groundwater should be determined, along with the upstream and downstream relationships between potential pollution sources such as surrounding industrial and mining enterprises and the contaminated area. For upstream and adjacent potential pollution sources, data on raw materials, production and discharge information, etc., should be collected and analyzed to determine the types of characteristic pollutants discharged or potentially leaked. Those overlapping with pollutants exceeding standards in the contaminated area should be included in the potential pollution source tracing comparison scope.

[0048] 3. Pollution source and groundwater sampling (1) Groundwater sampling: Several types of monitoring wells, such as background wells, pollution source wells, and migration path wells, are deployed along the direction of groundwater flow to monitor the background conditions of groundwater in the polluted area, the potential source areas of pollutants, and the migration paths of pollutants. Among them, background wells are deployed upstream of the groundwater flow in the polluted area to obtain uncontaminated groundwater samples and determine the characteristics of background particulate matter; pollution source wells are deployed directly in the known or suspected pollution source area or immediately downstream to capture and characterize the original pollutant particles released from the source; migration path wells (plume wells) are deployed along the predicted pollutant migration path (plume) to track the changes in the concentration, composition, and characteristics of particulate matter during migration.

[0049] When collecting groundwater environmental samples, the target stratum and the contaminated stratum should be the same. Groundwater environmental monitoring wells, groundwater sample collection, transportation, and preservation should be carried out in accordance with the national standard HJ164-2020 Technical Specification for Groundwater Environmental Monitoring. For areas where existing groundwater environmental monitoring wells meet the requirements of the technical specification, sampling from existing wells should be prioritized. To preserve suspended particles intact, no filtration should be performed during sample collection, and no chemical preservatives should be added during sample preservation. Samples should be stored in inert containers at low temperature and away from light to prevent particle aggregation or chemical changes.

[0050] (2) Sample Collection from Potential Pollution Sources: Water samples were collected from each of the screened potential pollution sources. For sources that generate or discharge industrial wastewater, comprehensive sampling of the process and end-of-pipe wastewater was conducted; for sources that may be leached or infiltrated by rainwater or surface runoff, samples were collected from rainwater ditches, sump pits, and surface runoff; for sources near landfills, landfill leachate was collected. At least 150 ml of liquid sample was collected in a pre-cleaned glass or PTFE bottle for single-particle mass spectrometry analysis. The requirements were the same as for groundwater samples: no filtration was performed during sample collection, and no chemical preservatives were added during sample storage. Samples were stored in inert containers at low temperatures and away from light to prevent particle aggregation or chemical changes.

[0051] Step S302: Perform single-particle mass spectrometry detection on each source sample to obtain the mass spectrometry information of multiple particles in each source sample.

[0052] Before sample analysis, single-particle aerosol mass spectrometry requires prior particle size and mass calibration, and the laser energy should be stabilized at approximately 0.5 mJ. For example... Figure 4 As shown, the collected water sample requires no pretreatment. An appropriate amount (about 30 ml) of liquid sample is placed in aerosol generator 1. High-purity N2 is introduced into aerosol generator 1 to generate aerosols. The aerosol sample is dried in drying tube 2 and then enters a single-particle aerosol mass spectrometer for detection.

[0053] like Figure 5As shown, aerosols enter the system through the sample inlet 13 and continue to propagate after being focused by the aerodynamic lens 3. Subsequently, the particles pass through the optical path formed by the first diameter-measuring laser 6 and the second diameter-measuring laser 7. The scattered light is received by the first photomultiplier tube 4 and the second photomultiplier tube 5 located on both sides to determine the particle's flight time and thus particle size, while simultaneously obtaining the estimated time for the particles to reach the ionization laser 10. The particles then reach the bottom ionization region, where they are ionized by the analytical / ionization laser 10 to obtain positive and negative ions. These ions are accelerated in the acceleration region 9, reach the reflection region 8, and are reflected to the microchannel plate 11 for detection. Furthermore, throughout the process, the ellipsoidal mirror 12 assists in laser diameter measurement, effectively collecting scattered light and ensuring measurement accuracy. This system is an important tool for analyzing atmospheric particulate matter, bioaerosols, and nanomaterials.

[0054] The principle of particulate matter detection and analysis is as follows: Aerosol particles enter the system through the sample inlet 1. Under multi-stage differential vacuum conditions, different particles have different velocities due to their different sizes. The particles are then focused by the sample introduction system into a collimated particle beam. After leaving the aerodynamic lens, the beam enters the diameter measurement zone. In the diameter measurement zone, the particles pass through two diameter measurement laser beams consecutively, and the scattered light generated is reflected and focused onto photomultiplier tubes (PMTs) for detection. By measuring the time interval between the two PMT signals using a timing circuit, the particle's flight velocity can be calculated, and thus the aerodynamic diameter of the particle can be derived. Additionally, the particle velocity is used to control the ionization laser emitted when the particle reaches the center of the ionization zone, ionizing the particle. After entering the ionization zone, the particles are ionized by a 266nm Nd:YAG ultraviolet pulsed laser, generating positive and negative ions. These ions are then detected by a time-of-flight mass analyzer, allowing for the simultaneous acquisition of positive and negative ion information for the particulate matter. Based on the above analysis, the particle size and chemical composition of a single particle can be obtained simultaneously.

[0055] The instrument will record the following parameters of the ionized particles into the corresponding file, as follows: Table 1 Instrument Recording Information

[0056] The time of flight of the particle between the two diameter-measuring laser beams was used to calculate the particle size; the mass-to-charge ratio (m / z) of the mass spectrometry peaks was used to identify the chemical components in the particles, and the peak area and peak height were used to assess the content of the corresponding components. The data file was converted into an array containing n particles, each particle containing 500-dimensional mass-to-charge ratio data (250 for positive ions and 250 for negative ions) and corresponding information such as particle size, peak area, and peak height.

[0057] Step S303: Construct a pollution source mass spectrometry fingerprint database based on the mass spectrometry information of multiple source samples. The pollution source mass spectrometry fingerprint database includes multiple fingerprint database entries. Each fingerprint database entry corresponds to a source attribution subset. The source attribution subset is jointly determined by a general mass spectrometry category and a potential pollution source, and includes the average mass spectrometry features of all particles in the source attribution subset.

[0058] In some optional implementations, constructing a pollution source mass spectrometry fingerprint database based on mass spectrometry information of multiple source samples includes the following steps S3031 to S3035.

[0059] Step S3031: Select from multiple particles contained in the source sample belonging to the same potential pollution source to obtain the effective source spectrum particles corresponding to each potential pollution source.

[0060] Specifically, selecting effective source spectrum particles corresponding to each potential pollution source from multiple particles contained in a source sample belonging to the same potential pollution source includes the following steps: performing first-level clustering on the mass spectrometry information of multiple particles contained in the source sample belonging to the same potential pollution source to obtain multiple sub-particle categories corresponding to each potential pollution source; filtering the multiple sub-particle categories obtained after the first-level clustering for each potential pollution source, retaining only particle categories whose particle quantity exceeds a preset threshold and have significant chemical characteristics, to obtain the effective source spectrum particles for each potential pollution source.

[0061] Therefore, by performing first-level clustering on the particle mass spectrometry data of the same potential pollution source and screening based on the proportion of quantity and the significance of chemical characteristics, noise particles and non-representative components were effectively eliminated, while effective source spectrum particles with statistical significance and source-specific chemical identifiers were retained. This not only improved the stability and representativeness of pollution source fingerprint features, but also enhanced the reliability and discriminativeness of subsequent fingerprint database construction.

[0062] For example, first-level clustering can employ clustering methods such as Adaptive Resonance Neural Network (ART-2a) and C-means clustering (FCM).

[0063] Step S3032: Merge all valid source spectrum particles into a single particle set, and perform global clustering on the particle set to obtain multiple general mass spectrometry categories; each general mass spectrometry category contains particles from one or more potential pollution sources.

[0064] Specifically, global clustering can employ clustering methods such as the Adaptive Resonant Neural Network (ART-2a) algorithm and the C-means clustering algorithm (FCM).

[0065] Step S3033: For each general mass spectrometry category, based on the potential pollution sources to which each particle in the general mass spectrometry category belongs, the general mass spectrometry category is divided into multiple source attribution subsets, and each source attribution subset corresponds to a combination of a general mass spectrometry category and a potential pollution source.

[0066] Step S3034: Associate and store each source subset with its corresponding potential pollution source and the average mass spectrometry characteristics of all particles in the subset to form a fingerprint database entry.

[0067] Step S3035: Traverse all general mass spectrometry categories and their corresponding source subsets to obtain the pollution source mass spectrometry fingerprint database.

[0068] For example, suppose there are three potential pollution sources A, B, and C, with 100 particles collected from each source. After first-level clustering, multiple subcategories of particles are obtained (e.g., A.1~A.50, B.1~B.50, C.1~C.50). By selecting and retaining categories with high retention rates and significant chemical characteristics (e.g., the top 20 representative particles from each source), a set of 150 valid source spectrum particles is formed. This set is then globally clustered to obtain several general mass spectrometry categories (e.g., category 1, category 2, category 3, etc.). These general categories are then divided into subsets (e.g., category 1-A, category 1-B, category 1-C) based on the original source of each particle, and the average mass spectrometry characteristics of each subset are calculated. Finally, a fingerprint database of entries composed of "general category + pollution source" is constructed, enabling refined characterization and differentiation of particle characteristics from different pollution sources.

[0069] Step S304: Obtain a receptor sample, wherein the receptor sample is a suspended particulate matter sample from the groundwater in the contaminated area.

[0070] Step S305: Perform single-particle mass spectrometry detection on the receptor sample to obtain the mass spectrometry information of each particle in the receptor sample.

[0071] Step S306: Determine the contamination source composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the contamination source mass spectrometry fingerprint database.

[0072] In one optional implementation, determining the contamination source composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the contamination source mass spectrometry fingerprint database includes the following steps: For particle j in the receptor sample, the mass spectrometry information of particle j is compared with the average mass spectrometry feature corresponding to each fingerprint database entry in the contamination source mass spectrometry fingerprint database to calculate the similarity, resulting in multiple source matching similarity scores; each source matching similarity score is compared with a preset similarity threshold; if at least one source matching similarity score is greater than the similarity threshold, particle j is assigned to the potential contamination source with the highest similarity score; if all source matching similarity scores are less than or equal to the similarity threshold, particle j is classified as an unknown source or background source; based on the assignment results of all particles in the receptor sample, the proportion of particles corresponding to each potential contamination source is statistically analyzed to determine the contamination source composition of the receptor sample.

[0073] For example, the number of particles in the receptor sample classified as each potential pollution source can be counted, and their percentage of the total number of particles can be calculated. This quantifies the relative contribution of each pollution source, forming the pollution source composition result of the receptor sample. For instance, if 60% of the particles belong to industrial sources and 30% to agricultural sources, then the pollution is determined to mainly originate from industrial sources.

[0074] Therefore, by quantitatively comparing the average mass spectrometry characteristics of receptor particles with those of each source subset in the fingerprint database, and combining this with a preset threshold, the precise source attribution of particles can be achieved. This not only effectively identifies the main pollution sources, but also reasonably distinguishes unknown or background particles, avoiding misjudgments. On this basis, by statistically analyzing the proportion of particles corresponding to each pollution source, the contribution of different pollution sources to receptor samples can be objectively quantified, significantly improving the accuracy, interpretability, and practicality of groundwater pollution source tracing results.

[0075] In another optional implementation, determining the pollution source composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the pollution source mass spectrometry fingerprint database includes the following steps: training a preset model based on the mass spectrometry information of each particle in the pollution source mass spectrometry fingerprint database and its corresponding potential pollution source label to obtain a pollution source classification model; converting the mass spectrometry information of each particle in the receptor sample into a feature vector of the same dimension as the training data and inputting it into the pollution source classification model to obtain the potential pollution source category to which each particle in the receptor sample belongs; and based on the classification results of all particles in the receptor sample, calculating the proportion of particles corresponding to each potential pollution source to determine the pollution source composition of the receptor sample.

[0076] Specifically, the mass spectrometry data of each particle in the pollution source mass spectrometry fingerprint database and its corresponding potential pollution source label can be used as training samples. First, the mass spectrometry data is standardized and uniformly converted into a 500-dimensional feature vector. Then, machine learning algorithms such as random forest, gradient boosting machine, or neural network are used to build a classification model. The model parameters are optimized through 5-fold or 10-fold cross-validation to prevent overfitting, thereby obtaining a pollution source classification model with strong generalization ability and high identification accuracy. In the application stage, the mass spectrometry information of each particle in the receptor sample is preprocessed into a 500-dimensional feature vector in the same way and then input into the model, which can automatically output the most likely potential pollution source category to which each particle belongs.

[0077] Therefore, by using labeled data from the pollution source mass spectrometry fingerprint database to train a dedicated pollution source classification model, the mass spectrometry information of receptor particles is automatically mapped to the corresponding pollution source category, realizing a high-throughput and intelligent source apportionment process.

[0078] Figure 6 This is a schematic diagram of the groundwater pollution source tracing method based on single-particle mass spectrometry coupled with machine learning according to the present invention, as shown below. Figure 6 As shown, in the training phase, samples of groundwater from the background area and potential pollution source areas are first collected, and their mass spectra are obtained through particle mass spectrometry analysis. This allows for the construction of a particle mass spectrometry fingerprint database containing characteristic information of various pollution sources. Subsequently, a comparison model based on machine learning is trained using cosine similarity comparison or by using this data to identify differences in mass spectrometry characteristics of groundwater from different sources. In the application phase, polluted groundwater is sampled and analyzed by particle mass spectrometry to generate corresponding mass spectra. These spectra are then compared with the source spectra or input into the pre-trained model for classification, ultimately outputting pollution source tracing results and achieving accurate identification and location of pollution sources.

[0079] Furthermore, after determining the pollution source composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the pollution source mass spectrometry fingerprint database, the following steps are also included: combining hydrogeological conditions and environmental background information to verify the rationality of the pollution source composition, and correcting the pollution source composition based on the verification results.

[0080] Specifically, the source tracing results can be cross-validated by combining multi-dimensional information such as groundwater flow direction, velocity, residence time, spatial distribution of known pollution sources, hydrochemical parameters, land use type, and particulate matter migration and transformation patterns. If some source apportionment results are found to be inconsistent with hydrogeological paths or environmental background, they are judged as unreasonable results and adjusted or eliminated, thereby correcting the composition ratio of pollution sources and making the final conclusion more consistent with the actual pollution migration patterns and site characteristics.

[0081] To illustrate the groundwater pollution source tracing method of single-particle mass spectrometry coupled with machine learning of the present invention more clearly, a specific example is given.

[0082] 1. Background Introduction In a groundwater pollution area of ​​a petrochemical enterprise in Guangdong Province, the main pollutants are benzene compounds and other organic matter. According to the investigation and analysis, the pollution mainly comes from two sources: (1) leakage from the sewage treatment plant, which seeps into the ground; (2) a small amount of water leaks onto the ground during the plant's production, which flows into the rainwater collection pond with rainwater. Damage to the pond leads to the pollutants entering the groundwater. For this area, groundwater samples and potential pollution source samples (rainwater collection pond and sewage treatment plant samples) with different distances from the pollution source and different degrees of pollution were collected to conduct a groundwater pollution source tracing experiment.

[0083] 2. Sampling and Measurement Status Five water samples were collected from different locations, including two samples from potential pollution sources (a wastewater treatment plant and a rainwater collection pond) and three groundwater samples (2A01, GW2, and W7). Among them, monitoring well 2A01, which is close to the rainwater collection pond, had the most severe groundwater pollution. GW2, located downstream of 2A01, showed slight pollution. W7, also downstream and far from 2A01 and the pollution source, showed no pollution in its groundwater sample.

[0084] Table 2. Basic Information Description of Samples at Each Location

[0085] Sample analysis was performed using a single-particle aerosol time-of-flight mass spectrometer (SPAMS0525) manufactured by Guangzhou Hexin Instruments Co., Ltd. Aerosols were generated for each sample using an aerosol generator, dried in a drying tube, and then directly analyzed by the single-particle mass spectrometer. The results showed that, except for sample W7 which had fewer than 300 particles, the other five water samples met the requirements for subsequent analysis.

[0086] Table 3 Particle characteristics of each sample

[0087] 3. Source spectral characteristics (1) Sample spectral characteristics The main components of the rainwater pool orthogonal spectrum include 39 K + , 23 Na + Organic matter ( 27 C5H3, 51 C4H3, +63 C5H3 + ), Fe, sodium salt ( 81 Na2 35 Cl+ , 83 Na2 37 Cl+), negative spectra include nitrates, nitrogen-containing organic compounds, phosphates, chlorides, etc. Characteristic peak mass-to-charge ratios (m / z) include 75, 96, 97, 111, 112, -95, -93, etc. Figure 7 As shown, the positive and negative ion spectra of the rainwater pool samples exhibit typical organic-inorganic mixed characteristics, especially with significant nitrate and phosphate signals in the negative spectrum.

[0088] The main components of the positive spectrum of a wastewater treatment plant include: sodium salt ( 23 Na + , 46 Na 2+ , 81 Na2 35 Cl + , 83 Na2 37 Cl + ), organic matter ( 27 C5H3, 63 C5H3 + , 91 C7H7 + ), metals (Ca, Fe, Mg), negative spectra mainly include oxygen-containing organic compounds ( -45 HCOO - , -59 CH3COO - Nitrogen-containing organic matter ( -26 CN - , -42 CNO - ) etc. For example Figure 8 As shown, the organic fragment peaks in the sample spectrum of the sewage treatment plant are more abundant, and the signal intensity of metal ions is significantly higher than that of the rainwater pool.

[0089] The mass spectrometry characteristics of groundwater samples 2A01 and GW2 are as follows. Groundwater sample W7 did not detect enough particles, and no valid data were generated.

[0090] like Figure 9 As shown, typical components of both rainwater ponds and sewage treatment plants were detected simultaneously in the positive and negative spectra of sample 2A01, such as sea salt characteristic peaks (m / z 62, 63) and nitrogen-containing organic matter (-42, -104), indicating that it may be affected by dual pollution sources.

[0091] like Figure 10 As shown, the organic signal in the GW2 sample spectrum is relatively weak, but the sulfate and phosphate features associated with the rainwater pool are still identifiable, which is consistent with the fact that it is located downstream and has relatively light pollution.

[0092] The comparison results show that the 2A01 mass spectrometry peaks are more diverse, followed by GW2, which is consistent with the fact that 2A01 is more contaminated than GW2.

[0093] Table 4 Comparison of Particle Spectral Characteristics of Groundwater Samples

[0094] (2) Clustering of sample spectral features and construction of source spectra After ART-2a cluster analysis, six main particle types were identified: Na-rich, Na-rich2, Na-Fe-PO3, Na-Fe-PO32, Ca-N, and Ca-N2 (the main difference between the two types, numbered 1 and 2, lies in the fact that the negative spectrum of type 2 shows more components). Figure 11 As shown, the six types of particles exhibit clear distinction in both positive and negative ion spectra, and can serve as source fingerprint features. Furthermore, as... Figure 12 As shown, there are significant differences in the proportion of various types of particles in each sample: the rainwater pool sample is mainly composed of Na-rich and Na-Fe-PO3, while the sewage treatment plant is rich in Ca-N particles. In the 2A01 sample, both types of pollution source characteristic particles have a high proportion, confirming its mixed pollution characteristics.

[0095] (3) Tracing the source and comparing 1) Verify the accuracy of the source resolution algorithm using source spectrum. Particulate matter samples from rainwater ponds and wastewater treatment plants were mixed. Half of the mixture was used as the source spectrum, and the other half was used as the "receptor" from a known source for source spectrum discrimination verification. The verification results showed that the identification accuracy of particles from wastewater treatment plants was 92.2%, and the identification accuracy of particles from rainwater ponds was 81.5%, with an overall identification accuracy of over 80%.

[0096] Table 5. Algorithm accuracy verification results

[0097] 2) Verification of mixing pollution sources in proportion To control experimental conditions, two types of pollution sources were mixed in different proportions for source tracing verification. Samples from rainwater ponds and wastewater treatment plants were mixed in a series of different proportions, and source tracing was performed based on single-particle mass spectrometry data to evaluate the consistency of the tracing results: results with a similarity of 0.8 were selected for comparison. The tracing results showed that the pollution source contribution ratio calculated by the model had a good linear correlation with the actual ratio, and the regression R-value was high. 2 The value reaches 0.9. For example... Figure 13 As shown, the source tracing results of this method show a high correlation with the actual proportion, reflecting the actual changes in the source composition, and can be used to effectively trace pollution sources.

[0098] 3) Tracing the source of groundwater pollution in the recipient area For two groundwater samples, 2A01 and GW2, the similarity between each particulate matter in the groundwater sample and the characteristic particle categories in the source spectrum was calculated using dot product. A similarity threshold of 0.8 was used for comparison to calculate the contribution of each pollution source to the groundwater. The source tracing results showed that the contribution ratios of the wastewater treatment plant and the stormwater pond for the 2A01 water sample were 0.3 and 0.32, respectively, while the contribution ratios of the wastewater treatment plant and the stormwater pond for the GW2 water sample were 0.16 and 0.45, respectively.

[0099] Table 6. Source tracing results of groundwater samples

[0100] Traditional source tracing technologies (such as three-dimensional fluorescence, eDNA, and stable isotope methods) mainly target pollutants in the liquid phase, while ignoring particulate and colloidal pollutant carriers in groundwater. This invention introduces single-particle mass spectrometry (SPMS) into the groundwater system for the first time, which can independently identify the chemical composition of each suspended particle, achieving a leap in identification dimensions "from the whole to the single particle". By capturing the unique source information carried by particulate pollutants or colloids, which are important migration carriers of pollutants, the source of pollution can be accurately traced. In particular, when multiple pollution sources cause groundwater pollution at the same time, different sources can be distinguished from a microscopic perspective, overcoming the shortcomings of traditional groundwater pollution source tracing, which can only distinguish overlapping effects by relying on the overall characteristics of pollution. It has the following beneficial effects: (1) Strong scalability: The SPMS-ML framework of this invention can be flexibly adjusted according to different pollution scenarios: in the heavy metal pollution scenario, the focus is on extracting metal ion peaks and oxide fragment signals; in the organic pollution scenario, the focus is on organic fragment peaks and nitrogen and oxygen functional group characteristics; in the complex pollution scenario, multi-source data adaptive identification can be achieved through model retraining. In addition, the constructed spectral library and model can be continuously updated to achieve long-term dynamic monitoring and self-learning optimization. This feature provides a technical basis for establishing a regional groundwater pollution "fingerprint database". (2) Reliable contribution rate quantification: The calculation of the "contribution rate" of traditional source apportionment methods mostly relies on statistical methods such as linear regression, principal component analysis (PCA) or mixed index models. These methods usually assume that the contributions of each pollution source are linearly additive and that the pollutant characteristics are independently distributed. However, in actual groundwater systems, the chemical signals of different pollution sources often have significant nonlinear coupling and overlap effects, which makes the source contribution ratio calculated by traditional methods lack physical meaning and easily lead to unreasonable results such as high or negative contribution rates. The contribution rate calculation of this method is directly based on the number of particles classified into different pollution sources, with clear physical meaning and more reliable results.

[0101] This embodiment also provides a groundwater pollution tracing device based on suspended solids single-particle mass spectrometry coupled with machine learning. This device is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0102] This embodiment provides a groundwater pollution source tracing device based on single-particle mass spectrometry coupled with machine learning, such as... Figure 14 As shown, it includes: The first acquisition module 1401 is used to acquire multiple source samples, wherein the source samples are collected from potential pollution sources.

[0103] The mass spectrometry information acquisition module 1402 is used to perform single-particle mass spectrometry detection on each source sample to obtain the mass spectrometry information of multiple particles in each source sample.

[0104] The fingerprint database construction module 1403 is used to construct a pollution source mass spectrometry fingerprint database based on the mass spectrometry information of multiple source samples. The pollution source mass spectrometry fingerprint database includes multiple fingerprint database entries, each fingerprint database entry corresponds to a source attribution subset. The source attribution subset is jointly determined by a general mass spectrometry category and a potential pollution source, and contains the average mass spectrometry features of all particles in the source attribution subset.

[0105] The second acquisition module 1404 is used to acquire receptor samples, wherein the receptor samples are suspended particulate matter samples in the groundwater of the contaminated area.

[0106] The mass spectrometry information acquisition module 1405 is also used to perform single-particle mass spectrometry detection on the receptor sample to obtain the mass spectrometry information of each particle in the receptor sample.

[0107] The traceability module 1406 is used to determine the contamination source composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the contamination source mass spectrometry fingerprint database.

[0108] In some optional implementations, the fingerprint database construction module 1403 is specifically used for: selecting from multiple particles contained in source samples belonging to the same potential pollution source to obtain effective source spectrum particles corresponding to each potential pollution source; merging all effective source spectrum particles into a particle set and performing global clustering on the particle set to obtain multiple general mass spectrometry categories; wherein each general mass spectrometry category contains particles from one or more potential pollution sources; for each general mass spectrometry category, according to the potential pollution source to which each particle in the general mass spectrometry category belongs, dividing the general mass spectrometry category into multiple source attribution subsets, each source attribution subset corresponding to a combination of a general mass spectrometry category and a potential pollution source; associating and storing each source attribution subset with its corresponding potential pollution source and the average mass spectrometry features of all particles in the subset to form a fingerprint database entry; traversing all general mass spectrometry categories and their divided source attribution subsets to obtain a pollution source mass spectrometry fingerprint database.

[0109] In some optional implementations, the fingerprint database construction module 1403 is specifically used to: perform a first-level clustering on the mass spectrometry information of multiple particles contained in the source samples belonging to the same potential pollution source, to obtain multiple sub-particle categories corresponding to each potential pollution source; and filter the multiple sub-particle categories obtained after the first-level clustering for each potential pollution source, retaining only the particle categories whose particle number ratio exceeds a preset threshold and have significant chemical characteristics, to obtain the effective source spectrum particles for each potential pollution source.

[0110] In some optional implementations, the source tracing module 1406 is used to: for particle j in the receptor sample, calculate the similarity between the mass spectrometry information of particle j and the average mass spectrometry feature corresponding to each fingerprint entry in the pollution source mass spectrometry fingerprint database, and obtain multiple source matching similarity scores; compare each source matching similarity score with a preset similarity threshold; if at least one source matching similarity score is greater than the similarity threshold, then particle j is assigned to the potential pollution source with the highest similarity score; if all source matching similarity scores are less than or equal to the similarity threshold, then particle j is classified as an unknown source or background source; based on the attribution results of all particles in the receptor sample, calculate the proportion of particles corresponding to each potential pollution source, and determine the pollution source composition of the receptor sample.

[0111] In some optional implementations, the source tracing module 1406 is used to: train a preset model based on the mass spectrometry information of each particle in the pollution source mass spectrometry fingerprint database and its corresponding potential pollution source label to obtain a pollution source classification model; convert the mass spectrometry information of each particle in the receptor sample into a feature vector of the same dimension as the training data and input it into the pollution source classification model to obtain the potential pollution source category to which each particle in the receptor sample belongs; and, based on the classification results of all particles in the receptor sample, calculate the proportion of particles corresponding to each potential pollution source to determine the pollution source composition of the receptor sample.

[0112] In some optional implementations, after determining the pollution source composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the pollution source mass spectrometry fingerprint database, the source tracing module 1406 is used to: combine hydrogeological conditions and environmental background information to verify the rationality of the pollution source composition, and correct the pollution source composition based on the verification results.

[0113] The groundwater pollution tracing device based on suspended solids single-particle mass spectrometry coupled with machine learning provided in this embodiment of the invention can execute the groundwater pollution tracing method based on suspended solids single-particle mass spectrometry coupled with machine learning provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the method. Further functional descriptions of the above modules and units are the same as in the corresponding embodiments described above, and will not be repeated here.

[0114] Figure 15 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0115] The following is a detailed reference. Figure 15 This diagram illustrates a suitable structural schematic for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 1501, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 1502 or a program loaded from memory 1508 into random access memory (RAM) 1503. The RAM 1503 also stores various programs and data required for the operation of the electronic device. The processor 1501, ROM 1502, and RAM 1503 are interconnected via a bus 1504. An input / output (I / O) interface 1505 is also connected to the bus 1504.

[0116] Typically, the following devices can be connected to I / O interface 1505: input devices 1506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 1507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 1508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1509. Communication device 1509 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 15 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0117] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 1509, or installed from a memory 1508, or installed from a ROM 1502. When the computer program is executed by the processor 1501, it performs the functions defined in the groundwater pollution source tracing method of suspended single-particle mass spectrometry coupled with machine learning according to embodiments of the present invention.

[0118] Figure 15 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0119] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the groundwater pollution source tracing method using suspended solids single-particle mass spectrometry coupled with machine learning shown in the above embodiments is implemented.

[0120] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0121] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for tracing groundwater pollution sources using single-particle mass spectrometry coupled with machine learning, characterized in that the method... include: Acquire multiple source samples, wherein the source samples are collected from potential sources of pollution; Single-particle mass spectrometry detection is performed on each of the source samples to obtain mass spectrometry information of multiple particles in each source sample; A pollution source mass spectrometry fingerprint database is constructed based on the mass spectrometry information of multiple source samples. The pollution source mass spectrometry fingerprint database includes multiple fingerprint database entries, each fingerprint database entry corresponds to a source attribution subset, the source attribution subset is determined by a general mass spectrometry category and a potential pollution source, and includes the average mass spectrometry features of all particles in the source attribution subset. Acquire receptor samples, wherein the receptor samples are suspended particulate matter samples from groundwater in the contaminated area; The receptor sample was subjected to single-particle mass spectrometry to obtain the mass spectrometry information of each particle in the receptor sample; The contamination source composition of the receptor sample is determined based on the mass spectrometry information of each particle in the receptor sample and the contamination source mass spectrometry fingerprint library.

2. The method according to claim 1, characterized in that, The construction of the pollution source mass spectrometry fingerprint database based on the mass spectrometry information of multiple source samples includes: Select from multiple particles contained in a source sample belonging to the same potential pollution source to obtain effective source spectrum particles corresponding to each potential pollution source; All valid source spectrum particles are merged into a single particle set, and the particle set is then globally clustered to obtain multiple general mass spectrometry categories; wherein each general mass spectrometry category contains particles from one or more of the potential pollution sources; For each of the general mass spectrometry categories, the general mass spectrometry category is divided into multiple source attribution subsets according to the potential pollution sources to which each particle in the general mass spectrometry category belongs. Each source attribution subset corresponds to a combination of a general mass spectrometry category and a potential pollution source. Each source subset is associated with its corresponding potential pollution source and the average mass spectrometry characteristics of all particles in that subset, and stored to form a fingerprint database entry. By traversing all general mass spectrometry categories and their corresponding source subsets, a pollution source mass spectrometry fingerprint database is obtained.

3. The method according to claim 2, characterized in that, The step of selecting from multiple particles contained in a source sample belonging to the same potential pollution source to obtain effective source spectrum particles corresponding to each potential pollution source includes: First-level clustering is performed on multiple particle mass spectrometry information contained in source samples belonging to the same potential pollution source to obtain multiple sub-particle categories corresponding to each potential pollution source. After the first-level clustering of each potential pollution source, multiple subcategories of particles are screened, and only particle categories with a particle count exceeding a preset threshold and significant chemical characteristics are retained to obtain the effective source spectrum particles for each potential pollution source.

4. The method according to claim 1, characterized in that, The step of determining the contaminant composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the contaminant mass spectrometry fingerprint database includes: For particle j in the receptor sample, the mass spectrometry information of particle j is compared with the average mass spectrometry feature corresponding to each fingerprint entry in the pollution source mass spectrometry fingerprint database to calculate the similarity and obtain multiple source matching similarity scores. Each source match similarity score is compared with a preset similarity threshold; If there is at least one source matching similarity score greater than the similarity threshold, then particle j is assigned to the potential pollution source with the highest similarity score; If the similarity scores of all source matches are less than or equal to the similarity threshold, then the particle j is classified as an unknown source or a background source. Based on the attribution results of all particles in the receptor sample, the proportion of particles corresponding to each potential pollution source is statistically analyzed to determine the pollution source composition of the receptor sample.

5. The method according to claim 1, characterized in that, The step of determining the contaminant composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the contaminant mass spectrometry fingerprint database includes: Based on the mass spectrometry information of each particle in the pollution source mass spectrometry fingerprint database and its corresponding potential pollution source label, the preset model is trained to obtain a pollution source classification model. The mass spectrometry information of each particle in the receptor sample is converted into a feature vector of the same dimension as the training data and input into the pollution source classification model to obtain the potential pollution source category to which each particle in the receptor sample belongs. Based on the attribution results of all particles in the receptor sample, the proportion of particles corresponding to each potential pollution source is statistically analyzed to determine the pollution source composition of the receptor sample.

6. The method according to claim 1, characterized in that, After determining the contaminant composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the contaminant source mass spectrometry fingerprint library, the method further includes: The rationality of the pollution source composition is verified by combining hydrogeological conditions and environmental background information, and the pollution source composition is corrected based on the verification results.

7. A groundwater pollution source tracing device using single-particle mass spectrometry coupled with machine learning, characterized in that, The device includes: The first acquisition module is used to acquire multiple source samples, wherein the source samples are collected from potential pollution sources; The mass spectrometry information acquisition module is used to perform single-particle mass spectrometry detection on each of the source samples to obtain the mass spectrometry information of multiple particles in each of the source samples. The fingerprint database construction module is used to construct a pollution source mass spectrometry fingerprint database based on the mass spectrometry information of multiple source samples. The pollution source mass spectrometry fingerprint database includes multiple fingerprint database entries, each fingerprint database entry corresponds to a source attribution subset, the source attribution subset is determined by a general mass spectrometry category and a potential pollution source, and includes the average mass spectrometry features of all particles in the source attribution subset. The second acquisition module is used to acquire receptor samples, wherein the receptor samples are suspended particulate matter samples in groundwater in the contaminated area. The mass spectrometry information acquisition module is also used to perform single-particle mass spectrometry detection on the receptor sample to obtain the mass spectrometry information of each particle in the receptor sample; The source tracing module is used to determine the source composition of the receptor sample based on the mass spectrometry information of each particle in the receptor sample and the mass spectrometry fingerprint database of the source contaminants.

8. An electronic device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the groundwater pollution source tracing method based on single-particle mass spectrometry coupled with machine learning of any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the groundwater pollution source tracing method based on single-particle mass spectrometry coupled with machine learning of any one of claims 1 to 6.

10. A computer program product, characterized in that, The method includes computer instructions for causing a computer to execute the groundwater pollution source tracing method based on single-particle mass spectrometry coupled with machine learning of any one of claims 1 to 6.

Images