A method and system for identifying types of atmospheric organic matter

Abstract

The present invention provides a method and system for identifying the types of atmospheric organic compounds, and relates to the technical field of organic identification. The method includes: collecting atmospheric samples; performing chromatographic separation on the atmospheric samples to obtain a plurality of unknown organic compounds and the retention time of each unknown organic compound; then ionizing the atmospheric samples to obtain characterizing ions of each unknown organic compound; combining the Schrodinger equation and wavelet transform to generate digital signals about the characterizing ions; pre-splitting the digital signals by a non-negative matrix decomposition algorithm; matching the organic compound types of each unknown organic compound based on a standard database and the split digital signals; combining a deep learning model to perform molecular structure inference on the unknown organic compounds to be inferred that failed to match; and outputting the obtained organic compound types as organic type identification results of atmospheric samples. The accuracy of organic type identification is effectively improved, the co-elution problem is avoided, and the scope of identification application is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of organic matter identification, and in particular to a method and system for identifying the types of atmospheric organic matter. Background Art

[0002] Atmospheric Organic Compounds (AOCs) refer to organic compounds present in the atmosphere, including volatile organic compounds (VOCs), semi-volatile organic compounds (SVOCs) and particulate organic matter (such as organic aerosols). Their sources include both natural processes (such as plant emissions and biological decomposition) and human activities (such as burning fossil fuels and industrial emissions).

[0003] Identifying the types of atmospheric organic matter is crucial to studying their sources, chemical reaction mechanisms, and impacts on air quality and climate. Differences in the types of organic matter directly affect how they participate in photochemical reactions and the types of pollutants they generate. For example, some volatile organic compounds are key precursors of ozone and secondary aerosols, while others have less impact on the environment. Only by accurately identifying the types of organic matter can we clarify the sources of major pollutants, provide a scientific basis for targeted air pollution prevention and control measures, and avoid blind governance and waste of resources. Only through accurate identification of organic matter types can we effectively assess environmental risks and effectively respond to environmental pollution problems.

[0004] However, the existing organic species identification process often relies too much on the fixed matching model of the mass spectrometer, which cannot solve the co-elution problem. That is, during the chromatographic separation process, there may be organic compounds with the same retention time, and the mass spectrometer will analyze and match them as an organic compound during the analysis process, resulting in large analysis deviations and easy matching failures. In addition, the fixed matching mode of the mass spectrometer is difficult to identify unconventional or complex organic compounds, and has great limitations. Summary of the invention

[0005] In order to solve the existing technical problems that in the process of organic species identification, the conventional technology often relies too much on the fixed matching model of the mass spectrometer and cannot solve the co-elution problem, that is, there may be organic compounds with the same retention time in the chromatographic separation process, and the mass spectrometer will analyze and match them as an organic compound during the analysis process, resulting in large analysis deviations and easily leading to matching failures. In addition, the fixed matching mode of the mass spectrometer is difficult to identify unconventional or complex organic compounds, and has great limitations. The present invention provides a method and system for identifying atmospheric organic species.

[0006] The technical solution provided by the embodiment of the present invention is as follows:

[0007] First aspect

[0008] An embodiment of the present invention provides a method for identifying types of atmospheric organic matter, comprising:

[0009] S1: Collect atmospheric samples;

[0010] S2: Perform chromatographic separation on the atmospheric samples to obtain a variety of unknown organic compounds and the retention time of each unknown organic compound;

[0011] S3: ionizing each unknown organic compound to obtain a characteristic ion of each unknown organic compound, wherein the characteristic ion includes a molecular ion and a fragment ion;

[0012] S4: combining the Schrödinger equation and wavelet transform to generate a digital signal about the characterizing ions, i.e., a mass spectrum in a digitized form, wherein the horizontal axis of the mass spectrum represents the ion mass-to-charge ratio of the unknown compound corresponding to the characterizing ions, and the vertical axis of the mass spectrum represents the ion abundance of the unknown compound corresponding to the characterizing ions;

[0013] S5: Combined with the retention time, the digital signal is pre-split by the non-negative matrix factorization algorithm to obtain digital signals representing different organic compounds;

[0014] S6: Match the organic compound types of each unknown organic compound based on the standard database and the split digital signals;

[0015] S7: Combine the deep learning model to infer the molecular structure of the unknown organic compound to be inferred that fails to match, and obtain the organic compound type of the unknown organic compound to be inferred;

[0016] S8: Output the obtained organic compound type as the organic matter type identification result of the atmospheric sample.

[0017] Second aspect

[0018] An embodiment of the present invention provides a system for identifying types of atmospheric organic matter, comprising:

[0019] processor;

[0020] A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method for identifying the type of atmospheric organic matter as described in the first aspect is implemented.

[0021] The third aspect

[0022] An embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method for identifying the type of atmospheric organic matter as described in the first aspect is implemented.

[0023] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0024] In the present invention, the atmospheric sample is first subjected to chromatographic separation and ionization, and the characterization ions of each unknown organic compound are generated, and the organic compounds in the complex sample can be effectively separated and converted into measurable signals, and a clear compound information basis is provided for subsequent mass spectrometry analysis, so as to avoid aliasing signals, improve the accuracy and reliability of identification, and provide a data basis for subsequent mass spectrometry analysis. Afterwards, in combination with the Schrodinger equation and wavelet transform, a mass spectrogram in the form of a digital signal, i.e., a digitized form, is generated about the characterization ion, and it is possible to accurately denoise and retain key characteristic signals, so as to provide a high-precision data basis for organic species identification, and significantly improve the accuracy and efficiency of identification. Based on the obtained high-precision digital signal, in combination with the retention time, the digital signal at the same retention time in the chromatographic separation process is pre-split by a non-negative matrix decomposition algorithm, and the digital signal at the same retention time is split by a non-negative matrix decomposition algorithm, which can effectively solve the co-elution problem, further distinguish overlapping signals, and ensure that the signals of different organic compounds are independent, which lays the foundation for improving the recognition accuracy and success rate, and greatly increases the organic identification accuracy and success rate. For organic matter that has not been successfully matched, the molecular structure of the unknown organic compounds to be inferred that failed to match is inferred by combining a deep learning model, so as to achieve the identification of new or complex organic matter, enhance the scope of application and accuracy of the identification, get rid of the limitations of the fixed matching mode of the mass spectrometer, and further increase the reliability, accuracy and scope of application of organic matter identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0026] Figure 1 A schematic diagram of a flow chart of a method for identifying types of atmospheric organic matter provided by an embodiment of the present invention;

[0027] Figure 2 A schematic diagram of the structure of a system for identifying types of atmospheric organic matter provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0029] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0030] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0031] Reference Manual Attached Figure 1 , showing a flow chart of a method for identifying types of atmospheric organic matter provided by an embodiment of the present invention.

[0032] The embodiment of the present invention provides a method for identifying the type of atmospheric organic matter, which can be implemented by an atmospheric organic matter type identification device, which can be a terminal or a server. The processing flow of the atmospheric organic matter type identification method can include the following steps:

[0033] S1: Collect atmospheric samples.

[0034] Among them, atmospheric samples refer to samples collected from the atmospheric environment that contain suspended particles, gases, aerosols and other components. The purpose is to capture organic compounds in them for analysis. They can be collected through professional measurement materials or measurement equipment.

[0035] S2: Perform chromatographic separation on the atmospheric samples to obtain a variety of unknown organic compounds and the retention time of each unknown organic compound.

[0036] Among them, chromatographic separation is a method of separating different components in a complex mixture through the distribution differences between compounds in the sample and the stationary phase and mobile phase. Common chromatographic techniques include gas chromatography (GC) and liquid chromatography (LC). Retention time refers to the time required for a compound to be detected by the detector from the injection of the sample into the chromatographic column. It is a specific characteristic value for each compound in chromatographic separation and is used for compound identification. After chromatographic separation of atmospheric samples, each organic compound enters the detector in turn according to its different retention time, thereby effectively separating and recording their time characteristics, providing clear chemical distribution information for subsequent analysis and avoiding aliasing.

[0037] In a possible implementation manner, S2 is specifically:

[0038] The atmospheric samples are subjected to chromatographic separation by using a chromatographic coupling technique comprising gas chromatography and liquid chromatography to obtain a plurality of unknown organic compounds and retention times corresponding to each unknown organic compound.

[0039] It should be noted that by separating atmospheric samples using gas chromatography and liquid chromatography, organic compounds in complex mixtures can be separated and the retention time of each compound can be recorded. This method significantly improves separation efficiency, avoids compound overlap and signal aliasing problems, effectively distinguishes isomers, and provides a clear and accurate sample basis for subsequent ionization and mass spectrometry analysis.

[0040] S3: Ionize each unknown organic compound separately to obtain a characteristic ion of each unknown organic compound.

[0041] Among them, the characterizing ions include molecular ions and fragment ions.

[0042] Among them, the characterizing ion is the ionized compound signal obtained by mass spectrometry, which can reflect the characteristic information of the compound, including its molecular weight, structure and chemical composition. Molecular ions are molecules that are charged as a whole after the compound is ionized, usually maintaining the integrity of the original molecule, and are used to infer the molecular weight and molecular formula. Fragment ions are partially charged molecular fragments generated by the cleavage of molecular ions during the ionization process, which are used to infer the structural information inside the molecule. Characterizing ions are generated by ionization, in which molecular ions provide molecular weight information of the compound and fragment ions provide structural information, providing key data for subsequent analysis and accurate identification of organic compounds.

[0043] In a possible implementation, S3 is specifically:

[0044] Each unknown organic compound is ionized by EI, ESI, MALDI or APCI to obtain a characteristic ion of each unknown organic compound.

[0045] It should be noted that unknown organic compounds are converted into characterizing ions through ionization techniques such as EI (Electron Ionization), ESI (Electrospray Ionization), MALDI (Matrix-Assisted Laser Desorption Ionization) or APCI (Atmospheric Pressure Chemical Ionization). Molecular ions provide molecular weight information of compounds, which helps to infer molecular formulas, while fragment ions reveal molecular structural characteristics through fragmentation patterns, providing data support for subsequent mass spectrometry analysis and compound identification.

[0046] S4: Combining the Schrödinger equation and wavelet transform, a digital signal representing the ions, i.e., a mass spectrum in digitized form, is generated.

[0047] The horizontal axis of the mass spectrum represents the ion mass-to-charge ratio of the unknown compound corresponding to the characteristic ion, and the vertical axis of the mass spectrum represents the ion abundance of the unknown compound corresponding to the characteristic ion.

[0048] Among them, the Schrödinger equation is the core equation in quantum mechanics, which is used to describe the change of the wave function of particles and is used to simulate the smooth diffusion and noise suppression of signals in step S4. Wavelet transform is a signal processing tool that analyzes local characteristics by decomposing the signal into multiple frequency components. Digital signal refers to the digitally processed electrical signal output by the mass spectrometer, which contains the mass-to-charge ratio and abundance information of the target compound. The mass-to-charge ratio is the ratio of the mass of an ion to its charge, and the mass spectrometer separates and detects ions by this value. The ion abundance is the number intensity of ions of a certain mass-to-charge ratio detected, reflecting the concentration or existence ratio of the ions. By combining the Schrödinger equation and the wavelet transform algorithm to process mass spectrometry data, the signal characterizing the ions is denoised to generate a digital mass spectrum, in which the mass-to-charge ratio and ion abundance provide the mass and existence ratio information of the compound respectively, laying a high-quality data foundation for the accurate analysis of organic matter.

[0049] In a possible implementation, S4 specifically includes:

[0050] S401: Obtaining the ion mass-to-charge ratio and ion abundance of the characteristic ions through a mass spectrometer.

[0051] Specifically, the charged ions generated after ionization, i.e., the characteristic ions, are accelerated by an accelerating electric field and enter the mass analyzer of the mass spectrometer. The acceleration voltage and field strength control the movement speed of the ions to ensure that the ions enter the mass analysis area with a controllable trajectory. According to the different mass-to-charge ratios of the ions, the ion groups are separated. Commonly used mass analyzers include quadrupole mass spectrometer (QMS): ions with different mass-to-charge ratios are controlled to pass through by a dynamic electric field. Time-of-flight mass spectrometer (TOF): separation based on differences in ion flight time. Magnetic mass spectrometer: separation based on the relationship between the degree of magnetic field deflection and the mass-to-charge ratio of the ions, Orbitrap or FT-ICR: separation of high-resolution ions by capturing the frequency signal generated by the movement of ions. The separated ions arrive at the detector of the mass spectrometer (such as an electron multiplier tube, photomultiplier tube or ion collector). The detector records the number of each ion that arrives, i.e., the ion abundance, and converts it into an electrical signal.

[0052] S402: Digitally process the ion mass-to-charge ratio and ion abundance to generate digital signals representing the ions.

[0053] Specifically, the electrical signal is converted into a digital record of mass-to-charge ratio and intensity, i.e., a digital signal, by analog-to-digital conversion (ADC). Each digital signal corresponds to a mass-to-charge ratio and abundance value.

[0054] S403: De-noising the digital signal by combining the Schrodinger equation and wavelet transform.

[0055] Specifically, the digital signal is decomposed into multi-scale coefficients, and the evolution of these coefficients is analogized to the dynamic change of the wave function for signal screening, filtering out background noise and non-target signals, and enhancing the target compound signal.

[0056] In one possible implementation, S403 specifically includes:

[0057] S4031: Perform wavelet transform on the digital signal to obtain the wavelet coefficients of the digital signal at different scales:

[0058]

[0059] in, t Represents the time index, Indicates that the time index is t The digital signal Indicates about t The scale j On k wavelet basis functions, Indicates about k The scale j The wavelet coefficients on .

[0060] It should be noted that by decomposing the signal into wavelet coefficients of different scales through wavelet transform, the local characteristics and detail changes of the signal can be captured, while the key features in the complex signal can be effectively extracted, thereby improving the flexibility of signal processing.

[0061] S4032: Using the wavelet coefficients as the wave function of the Schrödinger equation, establish the Schrödinger equation that describes the noisy state of the wavelet coefficients:

[0062]

[0063] in, Indicates that it corresponds to About time index t and corresponds to The digital signal value s The wave function of represents the wave function at the initial moment, h is Planck's constant, i represents the imaginary unit, represents partial derivatives, represents the potential energy term used to describe the noise distribution and intensity, Describes the wave function The kinetic energy term of the smooth diffusion characteristic, represents an adjustable parameter that controls the strength of noise suppression, Represents the wavelet coefficient threshold used to distinguish noise from digital signals.

[0064] Specifically, the wavelet coefficients are modeled as the wave function of the Schrödinger equation, and the dynamic change characteristics of the wave function are used to accurately suppress noise, which can simultaneously retain the details and structural information of the target signal and enhance the denoising effect.

[0065] It should be noted that those skilled in the art can set the size of the wavelet coefficient threshold according to actual needs, and the present invention is not limited here. Optionally, the setting of the wavelet coefficient threshold is determined by estimating the noise level (such as standard deviation), and combining the soft threshold or hard threshold method to balance noise suppression and signal fidelity. The threshold can be adaptively adjusted according to the characteristics of the target signal to ensure that key signal information is retained during denoising, while effectively filtering out background noise, thereby optimizing signal quality.

[0066] S4033: Solving the Schrödinger equation by iteratively updating the wave function until the wave function change value of the adjacent time index is less than the preset wave function change value:

[0067]

[0068] in, Represents the time step that controls the strength of the denoising process.

[0069] It should be noted that those skilled in the art can set the size of the preset wave function change value according to actual needs, and the present invention is not limited thereto.

[0070] Specifically, the Schrödinger equation can be solved by the finite difference method. By iteratively solving the Schrödinger equation, the wave function is continuously optimized until the signal converges stably, thereby gradually removing the noise and gradually improving the signal quality, ensuring the accuracy of the final result.

[0071] S4034: Perform inverse wavelet transform on the wavelet coefficients obtained after denoising to reconstruct the denoised digital signal:

[0072]

[0073] in, and Represent the denoised digital signal and wavelet coefficients respectively, Indicates that the wave function change value of the adjacent time index is less than the time index corresponding to the preset wave function change value.

[0074] It can be understood that reconstructing the signal by performing inverse wavelet transform on the denoised wavelet coefficients can effectively restore the original form of the signal, while significantly reducing noise interference and retaining the integrity and availability of the target signal.

[0075] Among them, this adjacent time index is the time index corresponding to the wave function change value being less than the preset wave function change value, and the newer time index among the two adjacent time indexes is taken, that is, the time index when the situation that the wave function change value of the adjacent time index is less than the preset wave function change value occurs.

[0076] It should be noted that the mass-to-charge ratio and abundance information of the characterizing ions are measured by a mass spectrometer, recorded and converted into digital signals, and then the digital signals are denoised by combining the Schrödinger equation and wavelet transform. The digital signals are denoised by combining the Schrödinger equation and wavelet transform. The multi-scale analysis capability of wavelet transform can effectively decompose complex signals and extract local features. At the same time, the Schrödinger equation simulates the dynamic smooth diffusion process of the signal, which helps to filter out non-target signals and noise and retain key target compound signals. This method significantly improves the clarity and quality of the signal, ensures that the mass spectrum accurately reflects the characteristics of the target compound, and provides a reliable data basis for subsequent identification, especially when dealing with complex background signals.

[0077] S5: Combined with the retention time, the digital signal is pre-split using a non-negative matrix factorization algorithm to obtain digital signals representing different organic compounds.

[0078] Among them, non-negative matrix decomposition is a linear dimensionality reduction method used to decompose a non-negative matrix into the product of two or more non-negative matrices. By pre-splitting complex aliasing signals, multiple co-eluting compound signals are decomposed into independent digital signals, providing reliable data support for the accurate identification of organic matter.

[0079] It is understandable that if multiple compounds are not completely separated by the chromatographic system (such as organic compounds with the same retention time), they will enter the mass spectrometer at the same time, that is, after chromatographic separation, more than one unknown compound enters the chromatograph, but the chromatograph interprets them as compounds in the same position, resulting in signal aliasing and difficulty in compound identification. This method solves the co-elution problem that often occurs during chromatographic separation, avoids the problem of large deviations or unrecognizable organic identification results caused by pattern aliasing, and can effectively improve the accuracy of organic identification.

[0080] Specifically, after the molecules in the sample are ionized, the charged ions are accelerated, separated and reach the detector in the mass spectrometer. The mass-to-charge ratio and ion abundance of each characteristic ion are recorded by the detector of the mass spectrometer to form a digital signal. The digital signal is decomposed by the non-negative matrix decomposition algorithm, which can take into account both accuracy and automation, complete the automated decomposition of the digital signal, and effectively avoid the co-elution problem.

[0081] In a possible implementation, S5 specifically includes:

[0082] S501: Generate a data matrix about the digital signal, wherein the rows of the data matrix represent mass-to-charge ratio sampling points, and the columns of the data matrix represent retention time sampling points.

[0083] Specifically, each row represents a fixed mass-to-charge ratio value. These values ​​are the sampling points (discretized mass-to-charge ratio range) of the mass spectrometer during measurement, for example Each column corresponds to a fixed retention time point. The retention time is the time range of chromatographic separation, which represents the time when the compounds in the sample enter the mass spectrometer after being separated by the chromatographic column. Each matrix element represents the abundance (intensity) of the ion signal.

[0084] S502: Decomposing the data matrix into a basis matrix representing the mass-to-charge ratio of organic compounds and a weight matrix representing the abundance of organic compounds at different retention times:

[0085]

[0086] in, and Respectively represent the data matrix P The decomposed basis matrix and weight matrix.

[0087] It should be noted that the data matrix is ​​decomposed into a basis matrix and a weight matrix. The basis matrix represents the mass-to-charge ratio distribution of the compound, and the weight matrix reflects the abundance of the compound at different retention times. This decomposition method can decompose complex signals into multiple independent components, providing high-quality data for accurate identification of compounds and subsequent analysis.

[0088] S503: Construct an optimization objective function for the basis matrix and the weight matrix:

[0089]

[0090] in, Indicates about , The optimization objective function value of represents the square of the Frobenius norm, represents the L1 norm, and They represent balanced sparsity and retention time constraints The first regularization parameter and the second regularization parameter, Indicates about A 2 retention time constraint function, Represents compound r The abundance distribution over retention time, Restricted compounds rThe weight function of the abundance distribution, Represents compound r The target distribution on the retention time axis is a unimodal width distribution. Indicates control compound r Adjustable parameter for single peak width, and Respectively represent the balance locality and unimodality The third and fourth regularization parameters of .

[0091] in, The colon in represents the range selector, which means selecting the matrix A 2 for all the elements of the column of compound r.

[0092] It should be noted that the constructed optimization objective function, by introducing regularization constraints such as sparsity, locality, and unimodality, ensures that the decomposed basis matrix and weight matrix can truly reflect the physical and chemical properties of the compound, effectively distinguish aliased signals, and improve the accuracy of decomposition.

[0093] S504: With the goal of minimizing the objective function, the basis matrix and the weight matrix are updated respectively by the alternating least squares method until the termination condition is reached:

[0094] fixed ;

[0095] fixed ;

[0096] Termination conditions: ;

[0097] in, represents element-wise multiplication, represents the update symbol, and the subscript T represents the transposition. express The gradient of Represents a constant that avoids the denominator being 0, Indicates about The reconstruction error threshold.

[0098] Among them, Alternating Least Squares (ALS) is an optimization algorithm that can be used for matrix decomposition problems. Its core idea is to alternately update parts of the matrix by fixing one variable and optimizing another variable until it finally converges to a solution that meets the conditions.

[0099] It can be understood that the goal is to minimize the objective function, and the basis matrix and weight matrix are updated by the alternating least squares method to continuously optimize the matrix decomposition result until the error meets the set threshold. This optimization method ensures that the decomposition result is accurate while avoiding reconstruction errors.

[0100] It should be noted that those skilled in the art can set the reconstruction error threshold according to actual needs, and the present invention does not limit this.

[0101] S505: Extract multiple single peaks based on the updated basis matrix and the updated weight matrix.

[0102] It should be noted that the updated basis matrix and weight matrix constrain the peak types belonging to different organic compounds within different single peaks through the introduced sparsity and peak shape constraints, or it can be directly understood as regularly amplifying the peak type and ignoring the peaks with smaller peak intensities, so that the complex peak type states present different single peak states after amplification. At this time, each single peak corresponds to an organic compound, which effectively solves the signal aliasing problem and ensures that the distribution shape of the compound signal, that is, the signal of the compound is usually a single Gaussian-shaped peak on the retention time axis.

[0103] S506: Reconstruct each single peak to obtain a digital signal representing the corresponding organic compound.

[0104] Specifically, the reconstruction formula is ,in They respectively represent the basis matrix and weight matrix corresponding to the extracted single peak data, and reconstruct the single peak to be extracted for refinement to restore its true peak state.

[0105] In actual application, the whole process starts with generating a data matrix representing the relationship between mass-to-charge ratio and retention time, extracting the basis matrix and weight matrix through matrix decomposition, constructing an optimization objective function and introducing regularization constraints to ensure the accuracy and physical meaning of the decomposition. Then, the matrix is ​​iteratively optimized through the alternating least squares method to extract the single-peak signal corresponding to a single compound, and finally these single-peak signals are reconstructed to restore the true signal state of each compound, effectively solving the signal aliasing problem and achieving accurate compound identification.

[0106] S6: Match the organic compound types of each unknown organic compound based on the standard database and the decomposed digital signals.

[0107] It should be noted that the types of unknown organic compounds are identified by matching the split digital signals with standard databases (such as NIST and MassBank). The matching process uses multi-dimensional features to ensure the accuracy and reliability of the identification results and provide clear chemical substance information for subsequent analysis.

[0108] In one possible implementation, the standard database includes NIST, MassBank, ubChem or ChemSpider.

[0109] S6 specifically includes:

[0110] S601: extracting multi-dimensional signal features of each split digital signal, wherein the multi-dimensional signal features include molecular ion peaks, fragment ion peaks, mass-to-charge ratios, and retention times.

[0111] Among them, the molecular ion peak is the peak in the mass spectrum that represents the entire molecular ion (usually a charged molecule formed after a molecule loses or gains an electron during the ionization process), which provides molecular weight information of the compound and is used to infer the molecular formula. The fragment ion peak is the peak of the charged molecular fragment generated by the breaking of chemical bonds during the ionization process of the molecular ion. These peaks reflect the internal structural characteristics of the molecule and are used to infer the bond connection relationship of the compound.

[0112] S602: traverse the multi-dimensional signal features in the standard database to match the organic compound types of each split digital signal.

[0113] It should be noted that by extracting multidimensional features such as molecular ion peaks, fragment ion peaks, mass-to-charge ratios and retention times of the split digital signals and matching them in the standard database, the types of unknown organic compounds can be accurately identified.

[0114] S7: Combine the deep learning model to infer the molecular structure of the unknown organic compounds to be inferred that failed to match, and obtain the organic compound type of the unknown organic compound to be inferred.

[0115] Specifically, existing mass spectrometers are often based on fixed rules or linear logic, and it is difficult to process unconventional or complex mass spectrometry data. The deep learning model is a data analysis method based on artificial neural networks, which has a high degree of nonlinear fitting ability and automatic feature extraction ability. By inputting a large amount of training data, the model can learn complex patterns and features for inferring or classifying unknown data. In mass spectrometry analysis, deep learning models can be used for molecular formula prediction, fragmentation pattern analysis, and complex sample inference. Using deep learning models to analyze digital signals that fail to match, and combining compound characteristic parameters to infer molecular structure, it overcomes the traditional method's dependence on databases and the limitations of fixed rules, and can identify unknown organic compounds with complex mass spectrometry features or unconventional fragmentation patterns, greatly expanding the scope of application and accuracy of identification.

[0116] This matching algorithm not only relies on known standard databases (such as NIST, MassBank), but can also infer the organic compound types of unknown organic compounds that failed to match based on several parameters, avoiding the limitations of the traditional method of relying on standard databases, and can effectively deal with the large differences in fragmentation patterns produced by different ionization methods (such as EI, ESI, MALDI), and analyze samples across ionization methods. It can also avoid the mistaken matching of unknown substances in samples with known substances in the database due to similar mass spectral features.

[0117] In a possible implementation, S7 specifically includes:

[0118] S701: Determine molecular formula generation constraints:

[0119]

[0120] in, represents the set of candidate molecular formulas, Represents a candidate chemical formula f The mass of ppm represents the mass error threshold. Represents the molecular formula used to restrict the rationality of the candidate molecular formula f The ring double bond equivalent of M The precursor ion mass of the unknown organic compound to be inferred is the sum of the unknown organic compound to be inferred and the adduct.

[0121] Among them, the ring double bond equivalent is an important parameter in the molecular structure of a compound, which is used to indicate the total number of double bonds and rings in a molecule and to limit the rationality of the molecular formula.

[0122] S702: According to the digital signal of the unknown organic compound to be inferred, traverse each chemical element based on the mass decomposition algorithm, and generate candidate molecular formulas of each unknown organic compound to be inferred under the constraint of molecular formula generation.

[0123] S703: extracting multiple single peaks from the digital signal corresponding to the unknown organic compound to be inferred, and traversing each chemical element based on a mass decomposition algorithm to generate candidate sub-molecular formulas of each single peak under molecular formula generation constraints.

[0124] S704: Calculate the peak matching error of each candidate sub-molecular formula:

[0125]

[0126] in, Indicates i The peak matching error of a single peak, Respectively represent i Single peak The measurement quality of iCandidate sub-formula The quality and i The adducts attached to the candidate sub-formulas Quality, Indicates taking the absolute value.

[0127] It should be noted that mass spectrometers work by detecting the mass (mass-to-charge ratio) of charged molecules, while the molecules themselves are usually neutral. In order to make the molecules charged for detection, some means are used in mass spectrometry experiments to allow the molecules to combine with adducts to make them positively or negatively charged precursor ions.

[0128] S705: Combine the feature information of each candidate sub-molecular formula in the unknown organic compound to be inferred and perform joint encoding to generate a feature vector of the unknown organic compound to be inferred:

[0129]

[0130] in, Indicates i The feature vector of the candidate sub-molecular formula, Enc represents the encoding operation, and c Respectively represent i The single peak intensity and the one-hot encoding of the measuring instrument corresponding to each candidate sub-molecular formula, Represents the total number of single peaks of unknown organic compounds to be inferred.

[0131] S706: Input the feature vector into the pre-trained deep learning model, and output the molecular formula score of the candidate molecular formula corresponding to the unknown organic compound to be inferred:

[0132]

[0133] in, Represents candidate molecular formula f The molecular formula score of represents a pre-trained multi-head attention model used to model the relationship between each candidate sub-molecular formula and the candidate molecular formula, and MLP represents a multi-layer perceptron.

[0134] S707: Convert formula scores to formula probabilities:

[0135]

[0136] in, j Indicates the number of candidate molecular formulas. e represents a natural constant, Indicates based on The candidate molecular formula f The probability of the molecular formula.

[0137] S708: Taking the candidate molecular formula corresponding to the maximum molecular formula probability as the target molecular formula of the unknown organic compound to be inferred, and outputting the target molecular formula as the organic compound type to be inferred.

[0138] Specifically, by setting molecular formula generation constraints (including mass error and ring double bond equivalent) to limit the reasonable range of candidate molecular formulas, the mass decomposition algorithm is used to generate candidate molecular formulas that meet the constraints. Next, the single peak data in the digital signal is extracted, and candidate sub-molecular formulas are generated for each single peak. Then, the peak matching error of the candidate sub-molecular formula is calculated according to the mass-to-charge ratio to ensure the accuracy of the molecular formula. After that, the feature information of the candidate sub-molecular formula is encoded to generate a feature vector, and the feature vector is input into the pre-trained deep learning model to calculate the score of the candidate molecular formula. The molecular formula score is converted into a probability distribution, and finally the molecular formula with the highest probability is selected as the target molecular formula and output, thereby achieving high-precision identification of the type of organic compound to be inferred.

[0139] S8: Output the obtained organic compound type as the organic matter type identification result of the atmospheric sample.

[0140] In actual application, first, complex atmospheric samples containing organic compounds are collected, and the compounds are separated and retained by chromatographic separation technology. Then, each compound is ionized to generate ion signals that characterize its mass and structural information. The Schrödinger equation and wavelet transform are combined to denoise the signal and generate a digital mass spectrum. The non-negative matrix decomposition algorithm is used to decompose the aliased signal into independent compound signals. By matching with the standard database, the types of known organic compounds are identified, and the molecular structure of the unmatched compounds is inferred through a deep learning model. Finally, all the identified organic species are comprehensively output to ensure high-precision and wide-coverage recognition results.

[0141] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0142] In the present invention, first chromatographic separation and ionization are carried out to the atmospheric sample, and the characterization ions of each unknown organic compound are generated, and the organic compounds in the complex sample can be effectively separated and converted into measurable signals, and a clear compound information basis is provided for subsequent mass spectrometry analysis, so as to avoid aliasing signals, improve the accuracy and reliability of identification, and provide data basis for subsequent mass spectrometry analysis. Afterwards, in combination with Schrödinger equation and wavelet transform, a mass spectrogram in the form of digital signal, i.e., digital form, about characterizing ions is generated, and accurate denoising and key characteristic signals can be retained, so as to provide a high-precision data basis for organic species identification, and the accuracy and efficiency of identification are significantly improved. Based on the obtained high-precision digital signal, in combination with retention time, the digital signal at the same retention time in the chromatographic separation process is pre-split by a non-negative matrix decomposition algorithm, and the digital signal at the same retention time is split by a non-negative matrix decomposition algorithm, and the co-elution problem can be effectively solved, and overlapping signals are further distinguished, and it is ensured that the signals of different organic compounds are independent, and the foundation is laid for improving recognition accuracy and success rate, and the organic identification accuracy and success rate are greatly increased. For organic matter that has not been successfully matched, the molecular structure of the unknown organic compounds to be inferred that failed to match is inferred by combining a deep learning model, so as to achieve the identification of new or complex organic matter, enhance the scope of application and accuracy of the identification, get rid of the limitations of the fixed matching mode of the mass spectrometer, and further increase the reliability, accuracy and scope of application of organic matter identification.

[0143] Reference Manual Attached Figure 2 , showing a schematic structural diagram of an atmospheric organic matter type identification system provided by the present invention.

[0144] The present invention further provides an atmospheric organic matter type identification system 20, which is applied to the above-mentioned atmospheric organic matter type identification method, and comprises:

[0145] Processor 201.

[0146] The memory 202 stores computer-readable instructions. When the computer-readable instructions are executed by the processor 201, the method for identifying the type of atmospheric organic matter in the method embodiment is implemented.

[0147] The atmospheric organic matter type identification system 20 provided by the present invention can execute the above-mentioned atmospheric organic matter type identification method and achieve the same or similar technical effects. To avoid repetition, the present invention will not elaborate on it.

[0148] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0149] In the present invention, first chromatographic separation and ionization are carried out to the atmospheric sample, and the characterization ions of each unknown organic compound are generated, and the organic compounds in the complex sample can be effectively separated and converted into measurable signals, and a clear compound information basis is provided for subsequent mass spectrometry analysis, so as to avoid aliasing signals, improve the accuracy and reliability of identification, and provide data basis for subsequent mass spectrometry analysis. Afterwards, in combination with Schrödinger equation and wavelet transform, a mass spectrogram in the form of digital signal, i.e., digital form, about characterizing ions is generated, and accurate denoising and key characteristic signals can be retained, so as to provide a high-precision data basis for organic species identification, and the accuracy and efficiency of identification are significantly improved. Based on the obtained high-precision digital signal, in combination with retention time, the digital signal at the same retention time in the chromatographic separation process is pre-split by a non-negative matrix decomposition algorithm, and the digital signal at the same retention time is split by a non-negative matrix decomposition algorithm, and the co-elution problem can be effectively solved, and overlapping signals are further distinguished, and it is ensured that the signals of different organic compounds are independent, and the foundation is laid for improving recognition accuracy and success rate, and the organic identification accuracy and success rate are greatly increased. For organic matter that has not been successfully matched, the molecular structure of the unknown organic compounds to be inferred that failed to match is inferred by combining a deep learning model, so as to achieve the identification of new or complex organic matter, enhance the scope of application and accuracy of the identification, get rid of the limitations of the fixed matching mode of the mass spectrometer, and further increase the reliability, accuracy and scope of application of organic matter identification.

[0150] It should be understood that the processor in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0151] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0152] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When a computer instruction or computer program is loaded or executed on a computer, a process or function according to an embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0153] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0154] In the present invention, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0155] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0156] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0157] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0158] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0159] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0160] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0161] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0162] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, a method for identifying types of atmospheric organic matter as described in the method embodiment is implemented.

[0163] A computer-readable storage medium provided by the present invention can implement the steps and effects of the atmospheric organic matter type identification method of the above method embodiment. To avoid repetition, the present invention will not go into details.

[0164] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0165] In the present invention, first chromatographic separation and ionization are carried out to the atmospheric sample, and the characterization ions of each unknown organic compound are generated, and the organic compounds in the complex sample can be effectively separated and converted into measurable signals, and a clear compound information basis is provided for subsequent mass spectrometry analysis, so as to avoid aliasing signals, improve the accuracy and reliability of identification, and provide data basis for subsequent mass spectrometry analysis. Afterwards, in combination with Schrödinger equation and wavelet transform, a mass spectrogram in the form of digital signal, i.e., digital form, about characterizing ions is generated, and accurate denoising and key characteristic signals can be retained, so as to provide a high-precision data basis for organic species identification, and the accuracy and efficiency of identification are significantly improved. Based on the obtained high-precision digital signal, in combination with retention time, the digital signal at the same retention time in the chromatographic separation process is pre-split by a non-negative matrix decomposition algorithm, and the digital signal at the same retention time is split by a non-negative matrix decomposition algorithm, and the co-elution problem can be effectively solved, and overlapping signals are further distinguished, and it is ensured that the signals of different organic compounds are independent, and the foundation is laid for improving recognition accuracy and success rate, and the organic identification accuracy and success rate are greatly increased. For organic matter that has not been successfully matched, the molecular structure of the unknown organic compounds to be inferred that failed to match is inferred by combining a deep learning model, so as to achieve the identification of new or complex organic matter, enhance the scope of application and accuracy of the identification, get rid of the limitations of the fixed matching mode of the mass spectrometer, and further increase the reliability, accuracy and scope of application of organic matter identification.

[0166] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0167] There are a few points to note:

[0168] (1) The drawings of the embodiments of the present invention only involve structures related to the embodiments of the present invention. Other structures may refer to conventional designs.

[0169] (2) For the sake of clarity, in the drawings used to describe the embodiments of the present invention, the thickness of layers or regions is exaggerated or reduced, that is, these drawings are not drawn according to the actual scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or there may be intermediate elements.

[0170] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to obtain new embodiments.

[0171] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for identifying types of atmospheric organic matter, characterized in that: include: S1: Collect atmospheric samples; S2: performing chromatographic separation on the atmospheric sample to obtain a plurality of unknown organic compounds and the retention time of each unknown organic compound; S3: ionizing each unknown organic compound to obtain a characterizing ion of each unknown organic compound, wherein the characterizing ion includes a molecular ion and a fragment ion; S4: combining the Schrödinger equation and wavelet transform to generate a digital signal about the characterizing ion, that is, a mass spectrum in a digitized form, wherein the horizontal axis of the mass spectrum represents the ion mass-to-charge ratio of the unknown compound corresponding to the characterizing ion, and the vertical axis of the mass spectrum represents the ion abundance of the unknown compound corresponding to the characterizing ion; S5: combining the retention time, pre-splitting the digital signal by a non-negative matrix decomposition algorithm to obtain digital signals representing different organic compounds; S6: Match the organic compound types of each unknown organic compound based on the standard database and the split digital signals; S7: Combine the deep learning model to infer the molecular structure of the unknown organic compound to be inferred that fails to match, and obtain the organic compound type of the unknown organic compound to be inferred; S8: Outputting the obtained organic compound type as the organic matter type identification result of the atmospheric sample; Wherein, the S4 specifically includes: S401: Acquire the ion mass-to-charge ratio and ion abundance of the characterizing ions by means of a mass spectrometer; S402: Digitally process the ion mass-to-charge ratio and ion abundance to generate a digital signal representing the ion; S403: De-noising the digital signal by combining Schrödinger equation and wavelet transform; Wherein, the S403 specifically includes: S4031: Perform wavelet transform on the digital signal respectively to obtain wavelet coefficients of the digital signal at different scales: in, t Represents the time index, Indicates that the time index is t The digital signal Indicates about t The scale j On k wavelet basis functions, Indicates about k The scale j The wavelet coefficients on ; S4032: Using the wavelet coefficients as the wave function of the Schrödinger equation, a Schrödinger equation is established to describe the noisy state of the wavelet coefficients: in, Indicates that it corresponds to About time index t and corresponds to The digital signal value s The wave function of represents the wave function at the initial moment, h is Planck's constant, i represents the imaginary unit, represents partial derivatives, represents the potential energy term used to describe the noise distribution and intensity, Describes the wave function The kinetic energy term of the smooth diffusion characteristic, represents an adjustable parameter that controls the strength of noise suppression, represents the wavelet coefficient threshold used to distinguish noise from digital signals; S4033: Solving the Schrödinger equation by iteratively updating the wave function until the wave function change value of the adjacent time index is less than a preset wave function change value: in, represents the time step that controls the strength of the denoising process; S4034: Perform inverse wavelet transform on the wavelet coefficients obtained after denoising to reconstruct the denoised digital signal: in, and represent the denoised digital signal and wavelet coefficients respectively, Indicates that the wave function change value of the adjacent time index is less than the time index corresponding to the preset wave function change value; Wherein, the S5 specifically includes: S501: Generate a data matrix about the digital signal, wherein the rows of the data matrix represent mass-to-charge ratio sampling points, and the columns of the data matrix represent retention time sampling points; S502: Decomposing the data matrix into a basis matrix representing the mass-to-charge ratio of organic compounds and a weight matrix representing the abundance of organic compounds at different retention times: in, and Respectively represent the data matrix P Decomposed basis matrix and weight matrix; S503: Constructing an optimization objective function for the basis matrix and the weight matrix: in, Indicates about The optimization objective function value of represents the square of the Frobenius norm, represents the L1 norm, and They represent balanced sparsity and retention time constraints The first regularization parameter and the second regularization parameter, Indicates about A 2 retention time constraint function, Represents compound r The abundance distribution over retention time, Restricted compounds r The weight function of the abundance distribution, Represents compound r The target distribution on the retention time axis is a unimodal width distribution. Indicates control compound r Adjustable parameter for single peak width, and Respectively represent the balance locality and unimodality The third and fourth regularization parameters of ; S504: With the goal of minimizing the objective function, the basis matrix and the weight matrix are updated respectively by the alternating least squares method until a termination condition is reached: in, represents element-wise multiplication, represents the update symbol, and the subscript T represents the transposition. express The gradient of Represents a constant that avoids the denominator being 0, Indicates about The reconstruction error threshold of S505: extracting multiple single peaks based on the updated basis matrix and the updated weight matrix; S506: reconstructing each single peak to obtain a digital signal representing the corresponding organic compound; Wherein, the S7 specifically includes: S701: Determine molecular formula generation constraints: in, represents the set of candidate molecular formulas, Represents a candidate chemical formula f The mass of ppm represents the mass error threshold. Represents the molecular formula used to restrict the rationality of the candidate molecular formula f The ring double bond equivalent of M The precursor ion mass of the unknown organic compound to be inferred is the sum of the unknown organic compound to be inferred and the adduct; S702: According to the digital signal of the unknown organic compound to be inferred, traverse each chemical element based on a mass decomposition algorithm, and generate candidate molecular formulas of each unknown organic compound to be inferred under the molecular formula generation constraint; S703: extracting multiple single peaks from the digital signal corresponding to the unknown organic compound to be inferred, and traversing each chemical element based on a mass decomposition algorithm to generate candidate sub-molecular formulas of each single peak under the molecular formula generation constraint; S704: Calculate the peak matching error of each candidate sub-molecular formula: in, Indicates i The peak matching error of a single peak, Respectively represent i Single peak The measurement quality of i Candidate sub-formula The quality and i The adducts attached to the candidate sub-formulas Quality, Indicates taking the absolute value; S705: Combine the feature information of each candidate sub-molecular formula in the unknown organic compound to be inferred and perform joint encoding to generate a feature vector of the unknown organic compound to be inferred: in, Indicates i The feature vector of the candidate sub-molecular formula, Enc represents the encoding operation, and c Respectively represent i The single peak intensity and the one-hot encoding of the measuring instrument corresponding to each candidate sub-molecular formula, Represents the total number of single peaks of the unknown organic compound to be inferred; S706: Input the feature vector into a pre-trained deep learning model, and output a molecular formula score of the candidate molecular formula corresponding to the unknown organic compound to be inferred: in, Represents candidate molecular formula f The molecular formula score of represents a pre-trained multi-head attention model used to model the relationship between each candidate sub-molecular formula and the candidate molecular formula, and MLP represents a multi-layer perceptron; S707: Convert the molecular formula score into a molecular formula probability: in, j Indicates the number of candidate molecular formulas. e represents a natural constant, Indicates based on The candidate molecular formula f The probability of the molecular formula; S708: Taking the candidate molecular formula corresponding to the maximum molecular formula probability as the target molecular formula of the unknown organic compound to be inferred, and outputting the target molecular formula as the organic compound type of the unknown organic compound to be inferred.

2. The method for identifying the types of atmospheric organic matter according to claim 1, characterized in that: The S2 is specifically: The atmospheric sample is subjected to chromatographic separation by a chromatographic coupling technique of gas chromatography and liquid chromatography to obtain a plurality of unknown organic compounds and retention times corresponding to each unknown organic compound.

3. The method for identifying the types of atmospheric organic matter according to claim 1, characterized in that: The S3 is specifically: Each unknown organic compound is ionized by EI, ESI, MALDI or APCI to obtain a characteristic ion of each unknown organic compound.

4. The method for identifying the types of atmospheric organic matter according to claim 1, characterized in that: The standard databases include NIST, MassBank, ubChem or ChemSpider; The S6 specifically includes: S601: extracting multi-dimensional signal features of each split digital signal, wherein the multi-dimensional signal features include molecular ion peaks, fragment ion peaks, mass-to-charge ratios, and retention times; S602: Traversing the multi-dimensional signal features in the standard database to match the types of organic compounds of each split digital signal.

5. A system for identifying types of atmospheric organic matter, characterized in that: include: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method for identifying the type of atmospheric organic matter according to any one of claims 1 to 4 is implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for identifying the type of atmospheric organic matter according to any one of claims 1 to 4 is implemented.

Images