Methods and systems for detecting trace amounts of estrogen-like substances in aquatic environments

Abstract

A method and system for detecting trace amounts of estrogen-like substances in aquatic environments are disclosed. The method acquires a high-performance liquid chromatography (HPLC) chromatogram of the water sample. It employs deep learning-based artificial intelligence technology to mine high-dimensional implicit correlation features of various estrogens within the HPLC chromatogram, and performs mutual correction of estrogen characteristics based on these features. This allows for accurate detection of various estrogens in water samples, thereby reducing estrogen pollution of water resources.

Description

Technical Field

[0001] This application relates to the field of intelligent detection technology, and more specifically, to a method and system for detecting trace amounts of estrogen-like substances in an aquatic environment. Background Technology

[0002] New pollutants, as the name suggests, refer to pollutants newly discovered in the environment, or those previously discovered but not given due attention, which pose potential hazards to human health and the ecological environment. With the improvement of modern water quality monitoring technology, more and more new pollutants are being detected in water bodies.

[0003] Estrogen is a novel environmental pollutant. Due to its large usage and complex environmental behavior and ecological effects, its residual status and ecological risks in the environment have attracted attention in recent years. China is a major producer and user of estrogen, and currently, various other novel pollutants have been detected to varying degrees in the aquatic environment. At present, my country has not yet issued environmental management policies or related emission standards for novel pollutants, nor has it established unified detection standards and methods. Furthermore, there is a lack of research and risk analysis on estrogen in drinking water sources and other water bodies.

[0004] Therefore, a detection method for trace estrogen-like substances in the aquatic environment is desired. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a method and system for detecting trace amounts of estrogen-like substances in an aquatic environment. The method involves acquiring a high-performance liquid chromatography (HPLC) chromatogram of a water sample; employing deep learning-based artificial intelligence technology to mine high-dimensional implicit correlation features related to various estrogens within the HPLC chromatogram; and performing mutual correction of various estrogen characteristics based on these high-dimensional implicit correlation features. This allows for accurate detection of various estrogens in water samples, thereby reducing estrogen pollution of water resources.

[0006] In a first aspect, a method for detecting trace amounts of estrogen-like substances in an aquatic environment is provided, comprising: acquiring a high-performance liquid chromatography (HPLC) chromatogram of a water sample to be tested; performing sample window scanning on the HPLC chromatogram to obtain multiple local chromatograms of sampling windows; using a first convolutional neural network model with a spatial attention mechanism to obtain multiple local component chromatographic feature matrices for each of the multiple local chromatograms of sampling windows; aggregating the multiple local component chromatographic feature matrices along the channel dimension into a three-dimensional input tensor and then passing it through a second convolutional neural network model as a feature map to obtain an inter-component correlation chromatographic feature map; performing feature discriminability enhancement along the channel dimension on the inter-component correlation chromatographic feature map to obtain a decoded feature map; and passing the decoded feature map through a first decoder to obtain a first decoded value, wherein the first decoded value represents the content of ethinylestradiol in the water sample to be tested.

[0007] In the above-mentioned method for detecting trace amounts of estrogen-like substances in the aquatic environment, the local chromatograms of the multiple sampling windows are respectively used to obtain multiple local component chromatographic feature matrices by employing a first convolutional neural network model with spatial attention mechanism. This includes: each layer of the first convolutional neural network model using spatial attention mechanism performs the following operations on the input data during the forward propagation process: convolution processing is applied to the input data to generate a convolutional feature map; pooling processing is applied to the convolutional feature map to generate a pooled feature map; nonlinear activation is applied to the pooled feature map to generate an activation feature map; the mean value along the channel dimension of each position of the activation feature map is calculated to generate a spatial feature matrix; a softmax-like function value is calculated at each position in the spatial feature matrix to obtain a spatial score matrix; and a positional dot product is calculated between the spatial feature matrix and the spatial score matrix to obtain a feature matrix; wherein, the feature matrix output by the last layer of the first convolutional neural network model using spatial attention mechanism is the multiple local component chromatographic feature matrix.

[0008] In the above-mentioned method for detecting trace amounts of estrogen-like substances in the aquatic environment, the multiple local component chromatographic feature matrices are aggregated along the channel dimension into a three-dimensional input tensor, and then passed through a second convolutional neural network model as a feature map to obtain an inter-component correlation chromatographic feature map. This includes: using each layer of the second convolutional neural network model as a feature map to perform convolution processing, mean pooling processing based on the feature matrix, and nonlinear activation processing on the input data during the forward propagation of the layer, so that the inter-component correlation chromatographic feature map is output by the last layer of the second convolutional neural network model as a feature map, wherein the input of the first layer of the second convolutional neural network model as a feature map is the three-dimensional input tensor.

[0009] In the above-mentioned method for detecting trace amounts of estrogen-like substances in the aquatic environment, the feature discriminability enhancement along the channel dimension of the inter-component correlation chromatographic feature map to obtain a decoded feature map includes: calculating the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map to obtain a weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors; and using the feature values ​​at each position of the weighted feature vector as weights to weight each feature matrix along the channel dimension of the inter-component correlation chromatographic feature map to obtain the classification feature map.

[0010] In the above-mentioned method for detecting trace estrogen-like substances in the aquatic environment, calculating the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map to obtain a weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors includes: calculating the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map using the following formula to obtain the weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors; wherein, the formula is: in, The first feature matrix of each feature matrix representing the intercomponent correlation chromatographic feature map is... Location feature value and These are the mean and variance of the set of eigenvalues ​​at each position of each feature matrix in the chromatographic feature map of the inter-component correlation, respectively. and These are the width and height of each feature matrix of the chromatographic feature map showing the correlation between the components. Represents the logarithmic function with base 2. This indicates the calculation of the natural exponential function value raised to the power of the numerical value. This represents the instance normalization and consistency-related recovery factor among the plurality of instance normalization and consistency-related recovery factors.

[0011] In the above-mentioned method for detecting trace estrogen-like substances in the aquatic environment, the decoded feature map is passed through a first decoder to obtain a first decoded value. The first decoded value represents the content of ethinylestradiol in the water sample to be tested. This method includes: using the decoder to perform decoding regression on the decoded feature map using the following formula to obtain the first decoded value; wherein the formula is: ,in This represents the decoded feature map. This represents the first decoded value. Represents the weight matrix. This represents the bias vector. This represents matrix multiplication.

[0012] Secondly, a detection system for trace estrogen-like substances in an aquatic environment is provided, comprising: a chromatogram acquisition module for acquiring a high-performance liquid chromatogram of a water sample to be tested; a sample window scanning module for scanning the high-performance liquid chromatogram to obtain multiple local chromatograms of sampling windows; a spatial attention module for using a first convolutional neural network model with a spatial attention mechanism to obtain multiple local component chromatographic feature matrices of the multiple sampling window local chromatograms respectively; a feature extraction module for aggregating the multiple local component chromatographic feature matrices along the channel dimension into a three-dimensional input tensor and then passing it through a second convolutional neural network model as a feature map to obtain an inter-component correlation chromatographic feature map; an enhancement module for enhancing the feature discriminability of the inter-component correlation chromatographic feature map along the channel dimension to obtain a decoded feature map; and a decoding module for passing the decoded feature map through a first decoder to obtain a first decoded value, wherein the first decoded value represents the content value of ethinylestradiol in the water sample to be tested.

[0013] In the aforementioned detection system for trace estrogen-like substances in the aquatic environment, the spatial attention module is configured to: each layer of the first convolutional neural network model using the spatial attention mechanism performs the following operations on the input data during the forward propagation of the layer: convolution processing to generate a convolutional feature map; pooling processing to generate a pooled feature map; nonlinear activation to generate an activation feature map; calculating the mean value along the channel dimension of each position of the activation feature map to generate a spatial feature matrix; calculating the softmax-like function value at each position in the spatial feature matrix to obtain a spatial score matrix; and calculating the positional dot product of the spatial feature matrix and the spatial score matrix to obtain a feature matrix; wherein the feature matrix output by the last layer of the first convolutional neural network model using the spatial attention mechanism is the plurality of local component chromatographic feature matrices.

[0014] In the above-mentioned detection system for trace estrogen-like substances in the aquatic environment, the feature extraction module is used to: perform convolution processing, mean pooling processing based on the feature matrix, and nonlinear activation processing on the input data in the forward propagation of the layers of the second convolutional neural network model that serves as the feature map, so that the last layer of the second convolutional neural network model that serves as the feature map outputs the chromatographic feature map of the intercomponent correlation, wherein the input of the first layer of the second convolutional neural network model that serves as the feature map is the three-dimensional input tensor.

[0015] In the above-mentioned detection system for trace estrogen-like substances in the aquatic environment, the enhancement module includes: a recovery factor calculation unit, used to calculate the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map to obtain a weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors; and a weighting unit, used to use the feature values ​​at each position of the weighted feature vector as weights to weight each feature matrix along the channel dimension of the inter-component correlation chromatographic feature map to obtain the classification feature map.

[0016] Compared with existing technologies, the method and system for detecting trace estrogen-like substances in the aquatic environment provided in this application acquire high-performance liquid chromatography (HPLC) chromatograms of the water sample to be tested; it employs deep learning-based artificial intelligence technology to mine high-dimensional implicit correlation features of various estrogens in the HPLC chromatograms of the water sample, and performs mutual correction of various estrogen characteristics based on these high-dimensional implicit correlation features. In this way, various estrogens in the water sample can be accurately detected, thereby reducing estrogen pollution of water resources. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 The image shows the MRM chromatograms of eight estrogens according to embodiments of this application.

[0019] Figure 2 This is a schematic diagram of a method for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application.

[0020] Figure 3 This is a flowchart of a method for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application.

[0021] Figure 4 This is a schematic diagram of the architecture of a method for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application.

[0022] Figure 5 This is a flowchart of step 150 in the method for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application.

[0023] Figure 6 This is a block diagram of a system for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application. Implementation

[0024] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0025] Unless otherwise stated, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application.

[0026] In the embodiments described in this application, it should be noted that, unless otherwise stated and limited, the term "connection" should be interpreted broadly. For example, it can be an electrical connection, or a connection between two internal components. It can be a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above term according to the specific circumstances.

[0027] It should be noted that the terms "first," "second," and "third" used in the embodiments of this application are merely used to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first," "second," and "third" can be interchanged in a specific order or sequence where permitted. It should be understood that the objects distinguished by "first," "second," and "third" can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in an order other than those illustrated or described herein.

[0028] As mentioned above, estrogen is a novel pollutant in the environment. Due to its large usage and complex environmental behavior and ecological effects, its residual status and ecological risks in the environment have attracted attention in recent years. China is a major producer and user of estrogen, and currently, various novel pollutants have been detected to varying degrees in the aquatic environment. At present, my country has not yet issued environmental management policies or related emission standards for novel pollutants, nor has it established unified detection standards and methods. There is also a lack of research and risk analysis on estrogen in drinking water sources and other water bodies. Therefore, a detection scheme for trace amounts of estrogen-like substances in the aquatic environment is desired.

[0029] In one embodiment of this application, a method for determining estrogen in water using solid-phase extraction-high performance liquid chromatography-tandem mass spectrometry (SPE-HPLC-MS / MS) is provided. The principle of this method is as follows: estrogen in a water sample is determined using SPE-HPLC-MS / MS. The water sample is concentrated by adsorption on a solid-phase extraction column, eluted with methanol solution, and then blown down to 0.5 ml with nitrogen. The volume is then adjusted with the mobile phase. Eight estrogens in drinking water and source water are detected using multiple reaction monitoring (MRM) mode of ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS). Qualitative analysis is performed based on retention time and characteristic ion peaks, and quantification is performed using the external standard method.

[0030] In this application, the sample processing is as follows: if the water sample contains suspended solids, it needs to be filtered through a 0.45 μm filter membrane; 1 L of water sample is measured and the pH of the sample is adjusted to <5 using sulfuric acid solution (1+5); enrichment and purification are performed using an HLB solid-phase extraction column. Before loading the sample, the solid-phase extraction column is equilibrated with 10 mL of methanol and 10 mL of pure water respectively. After loading the sample at a flow rate of 3.5 mL / min, it is rinsed with 20 mL of pure water, dried on the column under negative pressure for 10 min, and then eluted with 5 mL of ethyl acetate-methanol solution (ethyl acetate to methanol volume ratio of 9:1). The eluent is collected in a 10 mL collection tube, blown with nitrogen to 0.5 mL, dissolved in the mobile phase solution, thoroughly mixed, and sonicated for 30 s for analysis by ultra-high performance liquid chromatography-tandem mass spectrometry.

[0031] In this method, the liquid chromatography conditions are as follows: mobile phase: A—0.01% ammonia aqueous solution; B—acetonitrile. Column temperature: 30℃; injection volume: 10 μL; flow rate: 0.40 mL / min. Mobile phase gradient and time: gradient elution, 5 min. The gradient elution program is shown in Table 1 below.

[0032]

[0033] The required mass spectrometry conditions for this method include: ionization mode: electrospray ionization positive ion mode (ESI+); ion source temperature: 150℃; capillary voltage: 1.5kV; desolvation gas temperature: 550℃; desolvation gas flow rate: 1000 L / h. The scanning mode is multiple reaction monitoring (MRM). The MRM conditions for the eight estrogens are shown in Table 2 below.

[0034]

[0035] Among them, the MRM chromatograms of 8 estrogens at 10ug / L are shown below. Figure 1 As shown.

[0036] Furthermore, standard series of diethylstilbestrol, estrone, 17β-estradiol, ethinylestradiol, and estriol with concentrations of 1.0, 3.0, 5.0, 7.0, 10, 15, and 20 μg / L were prepared; standard series of 4-tert-octylphenol with concentrations of 0.5, 1.5, 2.5, 3.5, 5.0, 7.5, and 10 μg / L were prepared; and nonylphenol and bisphenol A with concentrations of 0.3, 0.9, 1.5, 2.1, 3.0, 4.5, and 6.0 μg / L were prepared. Under the above experimental conditions, standard curves were constructed by linear regression of peak area A against standard sample concentration c. The linear equations are shown in Table 3 below.

[0037]

[0038] In this application, according to HJ168-2010, when the target substance is not detected in the blank experiment, n (n≧7) parallel determinations are performed on samples with concentrations 2 to 5 times the estimated method detection limit, and the detection limit is calculated. An estrogen standard with a monomer concentration of 5 ng / L (4-tert-octylphenol monomer concentration of 2.5 ng / L; nonylphenol and bisphenol A monomer concentration of 1.5 ng / L) is prepared (added to pure water), and the test is performed under the above conditions. The standard deviation (SD) of the determination results is calculated, and the method detection limit (MDL) and the lower limit of quantification (MQL) are calculated according to the following formula: MDL = t(n-1, 0.99) × SD = 3.143 × SD

[0039] MQL = 4 × MDL. The statistical data of the method detection limits and lower limits of measurement for eight estrogens are shown in Table 4 below.

[0040]

[0041] A solid-phase extraction-high performance liquid chromatography-tandem mass spectrometry (SPE-HPLC-MS / MS) method was used to detect estrogen in water. The method showed good linearity in the range of 1–20 μg / L (0.5–10 μg / L for 4-tert-octylphenol, and 0.3–6 μg / L for nonylphenol and bisphenol A), with a correlation coefficient R ≥ 0.995. The limit of quantitation (LOQ) was 1 g / L (0.5 ng / L for 4-tert-octylphenol, and 0.3 ng / L for nonylphenol and bisphenol A).

[0042] In the survey of estrogen content, solid-phase extraction-high performance liquid chromatography-tandem mass spectrometry (SPE-HPLC-MS / MS) was used to detect and investigate the estrogen content in water samples from Qiandao Lake, Fengkou, Qiantang River, Tiesha River, and the effluent from the water treatment plant. The test results are shown in Tables 5 and 6 below.

[0043]

[0044]

[0045] Specifically, in another embodiment of this application, the estrogens in the water sample are determined using solid-phase extraction-high performance liquid chromatography-tandem mass spectrometry (SPE-HPLC-MS / MS). Specifically, the water sample is concentrated by adsorption on a solid-phase extraction column, eluted with methanol solution, purged with nitrogen to 0.5 ml, and brought to volume with the mobile phase. Eight estrogens in drinking water and source water are detected using multiple reaction monitoring (MRM) mode of ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS). Qualitative analysis is based on retention time and characteristic ion peaks, while quantification is performed using the external standard method.

[0046] Accordingly, in the actual process of detecting estrogen in water samples, the analysis of high-performance liquid chromatography (HPLC) is crucial for detecting the content of eight estrogens in the water sample. Specifically, the eight estrogens are ethinylestradiol, estrone, 17β-estradiol, estriol, 4-tert-octylphenol, nonylphenol, bisphenol A, and diethylstilbestrol. However, considering that each type of estrogen has its own implicit characteristic distribution information in the HPLC chromatogram, and that there are correlations between the implicit features of these estrogens in the HPLC chromatogram, it is difficult to determine the content of each hormone. Therefore, in the technical solution of this application, it is hoped that the mutual correction of the characteristics of each type of estrogen can be performed by analyzing the high-dimensional implicit correlation features between the various estrogens in the HPLC chromatogram, thereby improving the accuracy of detecting the content of each type of estrogen. The difficulty in this process lies in how to fully and accurately extract the high-dimensional implicit correlation feature information of each hormone in the HPLC chromatogram of the water sample, so as to improve the accuracy of detecting the content of each type of estrogen in the water sample.

[0047] In recent years, deep learning and neural networks have been widely applied in fields such as computer vision, natural language processing, and text signal processing. Furthermore, deep learning and neural networks have demonstrated near-human or even superior performance in areas such as image classification, object detection, semantic segmentation, and text translation.

[0048] The development of deep learning and neural networks has provided new ideas and solutions for mining high-dimensional hidden correlation features of various hormones in the high-performance liquid chromatograms of the water sample.

[0049] Specifically, in the technical solution of this application, firstly, a high-performance liquid chromatography (HPLC) chromatogram of the water sample to be tested is obtained. Next, considering the correlation between the implicit characteristic information of various estrogens in the HPLC chromatogram of the water sample to be tested, and that this correlation has different correlation characteristic information at different positions in the HPLC chromatogram, in order to fully extract the correlation characteristic information of various estrogens in the HPLC chromatogram and thus mutually correct to obtain accurate estrogen content values, the technical solution of this application further performs sample window scanning on the HPLC chromatogram to obtain multiple sampling window partial chromatograms. Specifically, in a specific example of this application, multiple sampling window partial chromatograms are obtained by performing sample window scanning of the HPLC chromatogram using a predetermined sliding window.

[0050] Then, for the local chromatograms of each sampling window, considering that they are image data, a convolutional neural network model, which excels in extracting latent features from images, is used to mine the latent features of various estrogens in the local chromatograms of each sampling window. Specifically, considering that the hidden features of various estrogens exhibit different characteristics at different locations in the local chromatograms of the sampling windows during the detection of their content, the actual process of mining the latent features of various estrogens should focus on the high-dimensional latent correlation features of various estrogens at spatial locations in the local chromatograms of each sampling window.

[0051] It is understandable that, given that the attention mechanism can select the focus position and generate more discriminative feature representations, and that the features after adding the attention module will adaptively change as the network deepens, the technical solution of this application processes the multiple sampling window local chromatograms separately using a first convolutional neural network model with a spatial attention mechanism to extract the high-dimensional implicit correlation feature distribution information of the various estrogens focused on in space in each sampling window local chromatogram, thereby obtaining multiple local component chromatographic feature matrices. It is worth mentioning that here, the image features extracted by the spatial attention reflect the weights of spatial dimensional feature differences, used to suppress or enhance features at different spatial locations, thereby extracting the implicit correlation feature information of the content of the various estrogens focused on in the spatial location of each sampling window local chromatogram.

[0052] Furthermore, considering the correlation between the implicit correlation features of the various estrogens in the local chromatograms of each sampling window—that is, each local chromatogram of the sampling window contains correlation features of the content of the various estrogens, and these features are also correlated in the local chromatograms of each sampling window—in order to fully detect the content of the various estrogens in the water sample to be tested, in the technical solution of this application, the multiple local component chromatographic feature matrices are further aggregated into a three-dimensional input tensor along the channel dimension and then subjected to feature mining through a second convolutional neural network model as a feature map to extract the correlation feature distribution information between the implicit correlation features of the various estrogens in the local chromatograms of each sampling window, thereby obtaining a component correlation chromatographic feature map. Specifically, here, the convolution kernel of the second convolutional neural network model is a three-dimensional convolution kernel, which has W (width), H (height) and C (channel dimension). In the technical solution of this application, the channel dimension of the three-dimensional convolution kernel corresponds to the channel dimension of the three-dimensional input tensor. Therefore, when performing three-dimensional convolutional encoding, high-dimensional implicit correlation feature distribution information of the content features of various estrogens between local chromatograms of each sampling window in the three-dimensional input tensor can be extracted.

[0053] Next, the decoded feature map is further decoded and regressed through the first decoder to obtain a first decoded value representing the content of ethinylestradiol in the water sample to be tested. This is done by cross-correcting based on the high-dimensional implicit correlation features of various estrogens in the high-performance liquid chromatogram of the water sample to be tested, thereby improving the accuracy of detecting the content of various estrogens and thus obtaining the content of ethinylestradiol in the water sample to be tested. Specifically, here, there are eight estrogens, namely ethinylestradiol, estrone, 17β-estradiol, estriol, 4-tert-octylphenol, nonylphenol, bisphenol A, and diethylstilbestrol. Correspondingly, in other specific examples of this application, second to eighth decoders can also be constructed to decode and detect the other seven estrogens, which is not limited to this application.

[0054] Specifically, in the technical solution of this application, the feature discriminability enhancement along the channel dimension of the inter-component correlation chromatographic feature map is performed to obtain a decoded feature map. For example, a channel-weighted feature vector can be obtained by performing global mean pooling on each feature matrix arranged along the channel dimension of the inter-component correlation chromatographic feature map, thereby weighting each feature matrix arranged along the channel dimension of the inter-component correlation chromatographic feature map. However, considering that global mean pooling of the feature matrix inevitably loses the discriminative feature spatial distribution information expressed by the feature matrix, it is desirable to optimize the channel-weighted feature vector to recover as much as possible the feature distribution information of the feature vector expressed by the overall distribution correlation between each feature matrix in the inter-component correlation chromatographic feature map, thereby improving the expression effect of the channel-weighted feature vector on the channel-dimensional feature correlation distribution between each feature matrix.

[0055] Based on this, when performing channel-dimensional global mean pooling on the inter-component correlation chromatographic feature maps, the applicant of this application calculates the instance normalization and consistency correlation recovery factor of the global mean pooling for each feature matrix, expressed as:

[0056] and It is a feature set The mean and variance, Characteristic matrix The The eigenvalues ​​of the location, and and These are the width and height of the feature matrix, respectively.

[0057] Here, the instance normalization and consistency-related recovery factor of global mean pooling address the inevitable loss of discriminative feature information in global mean pooling of the feature matrix. Based on instance normalization (IN) of the spatial distribution features of channel instances expressed by the feature matrix, consistency-related features are distilled from the statistical residual features of class probabilities to restore the coherence of consistency-related features in the statistical information to the eigenvalue representation. This achieves the recovery of the channel-weighted feature vector relative to the inter-component correlation chromatographic feature map, which is constrained by the spatial distribution of the feature matrix in terms of channel causality. Thus, by weighting the eigenvalues ​​at corresponding positions of the channel-weighted feature vector with this factor, the channel-weighted feature vector can contain the feature distribution information of the feature vector expressed by the overall distribution correlation between the various feature matrices in the inter-component correlation chromatographic feature map. This improves the expression effect of the channel-weighted feature vector on the channel-dimensional feature correlation distribution between the various feature matrices, thereby correspondingly improving the accuracy of the decoding regression of the decoded feature map. This allows for accurate detection of various estrogens in water samples, thereby reducing estrogen pollution of water resources.

[0058] Figure 2 This is a schematic diagram illustrating a method for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application. Figure 2 As shown, in this application scenario, firstly, a high-performance liquid chromatogram (HPLC) of the water sample to be tested is obtained (e.g., as shown in the figure). Figure 2 (as shown in C); then, the obtained high-performance liquid chromatogram is input to a server deployed with an algorithm for detecting trace amounts of estrogen-like substances in the aquatic environment (e.g., as shown in C). Figure 2 In the S shown, the server is capable of processing the high-performance liquid chromatogram based on an algorithm for detecting trace amounts of estrogen-like substances in the aquatic environment to generate a first decoded value representing the content of ethinylestradiol in the water sample to be tested.

[0059] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0060] In one embodiment of this application, Figure 3 This is a flowchart illustrating a method for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application. Figure 3As shown, a method 100 for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application includes: 110, acquiring a high-performance liquid chromatography (HPLC) chromatogram of a water sample to be tested; 120, performing sample window scanning on the HPLC chromatogram to obtain multiple sampling window local chromatograms; 130, using a first convolutional neural network model with a spatial attention mechanism to obtain multiple local component chromatographic feature matrices for each of the multiple sampling window local chromatograms; 140, aggregating the multiple local component chromatographic feature matrices along the channel dimension into a three-dimensional input tensor and then passing it through a second convolutional neural network model as a feature map to obtain an inter-component correlation chromatographic feature map; 150, performing feature discriminability enhancement along the channel dimension on the inter-component correlation chromatographic feature map to obtain a decoded feature map; and 160, passing the decoded feature map through a first decoder to obtain a first decoded value, wherein the first decoded value represents the content value of ethinylestradiol in the water sample to be tested.

[0061] Figure 4 This is a schematic diagram of the architecture of a method for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application. Figure 4 As shown, in this network architecture, firstly, a high-performance liquid chromatography (HPLC) chromatogram of the water sample to be tested is acquired; then, the HPLC chromatogram is scanned to obtain multiple sampling window local chromatograms; next, the multiple sampling window local chromatograms are respectively processed by a first convolutional neural network model with spatial attention mechanism to obtain multiple local component chromatographic feature matrices; then, the multiple local component chromatographic feature matrices are aggregated along the channel dimension into a three-dimensional input tensor and passed through a second convolutional neural network model as a feature map to obtain an inter-component correlation chromatographic feature map; next, the inter-component correlation chromatographic feature map is enhanced with feature discriminability along the channel dimension to obtain a decoded feature map; and finally, the decoded feature map is passed through a first decoder to obtain a first decoded value, which represents the content value of ethinylestradiol in the water sample to be tested.

[0062] Specifically, in step 110, a high-performance liquid chromatography (HPLC) chromatogram of the water sample to be tested is obtained. In the actual process of estrogen detection in water samples, the analysis of the HPLC chromatogram to detect the content of eight estrogens in the water sample is crucial. Specifically, the eight estrogens are ethinylestradiol, estrone, 17β-estradiol, estriol, 4-tert-octylphenol, nonylphenol, bisphenol A, and diethylstilbestrol. However, considering that each type of estrogen has its own implicit characteristic distribution information in the HPLC chromatogram, and that there are correlations between these implicit features, determining the content of each hormone presents difficulties. Therefore, in the technical solution of this application, it is desirable to improve the accuracy of estrogen content detection by mutually correcting the characteristics of each estrogen through the high-dimensional implicit correlation features between them in the HPLC chromatogram. The challenge in this process lies in how to fully and accurately extract the high-dimensional implicit correlation features of various hormones from the high-performance liquid chromatogram of the water sample, so as to improve the accuracy of detecting the content of various estrogens in the water sample.

[0063] In recent years, deep learning and neural networks have been widely applied in fields such as computer vision, natural language processing, and text signal processing. Furthermore, deep learning and neural networks have demonstrated near-human or even superior performance in areas such as image classification, object detection, semantic segmentation, and text translation.

[0064] The development of deep learning and neural networks has provided new ideas and solutions for mining high-dimensional hidden correlation features of various hormones in the high-performance liquid chromatograms of the water sample.

[0065] Specifically, in the technical solution of this application, firstly, a high-performance liquid chromatogram of the water sample to be tested is obtained.

[0066] Specifically, in step 120, the high-performance liquid chromatography (HPLC) chromatogram is sampled using a sample window scan to obtain multiple partial chromatograms of sampling windows. Next, considering the correlation between the implicit characteristic information of various estrogens in the HPLC chromatogram of the water sample to be tested, and that this correlation has different correlation characteristic information at different positions in the HPLC chromatogram, in order to fully extract the correlation characteristic information of various estrogens in the HPLC chromatogram and thus mutually correct for accurate estrogen content values, the technical solution of this application further performs a sample window scan on the HPLC chromatogram to obtain multiple partial chromatograms of sampling windows. Specifically, in a specific example of this application, the HPLC chromatogram is sampled using a predetermined sliding window to obtain multiple partial chromatograms of sampling windows.

[0067] Specifically, in step 130, the multiple sampling window local chromatograms are processed using a first convolutional neural network model with a spatial attention mechanism to obtain multiple local component chromatographic feature matrices. It should be understood that, given that the attention mechanism can select the focus position, generating more discriminative feature representations, and that the features after adding the attention module will adaptively change as the network deepens, in the technical solution of this application, the multiple sampling window local chromatograms are processed using a first convolutional neural network model with a spatial attention mechanism to extract the high-dimensional implicit correlation feature distribution information of the various estrogens focused on in space in each sampling window local chromatogram, thereby obtaining multiple local component chromatographic feature matrices. It is worth mentioning that here, the image features extracted by the spatial attention reflect the weights of spatial dimensional feature differences, used to suppress or enhance features at different spatial locations, thereby extracting the implicit correlation feature information of the content of the various estrogens focused on in the spatial location of each sampling window local chromatogram.

[0068] The method involves using a first convolutional neural network model with a spatial attention mechanism to obtain multiple local component chromatographic feature matrices from the local chromatograms of the multiple sampling windows. This includes: each layer of the first convolutional neural network model with spatial attention performing the following operations on the input data during the forward propagation process: convolving the input data to generate a convolutional feature map; pooling the convolutional feature map to generate a pooled feature map; applying nonlinear activation to the pooled feature map to generate an activation feature map; calculating the mean value along the channel dimension at each position of the activation feature map to generate a spatial feature matrix; calculating the Softmax-like function value at each position in the spatial feature matrix to obtain a spatial score matrix; and calculating the positional dot product of the spatial feature matrix and the spatial score matrix to obtain a feature matrix. The feature matrix output by the last layer of the first convolutional neural network model with spatial attention is the multiple local component chromatographic feature matrices.

[0069] As you can understand, attention mechanisms are a data processing method in machine learning, widely used in various types of machine learning tasks such as natural language processing, image recognition, and speech recognition. On one hand, attention mechanisms aim to enable the network to automatically learn the areas in an image or text sequence that require attention; on the other hand, attention mechanisms generate a mask through neural network operations, with weights assigned to the values ​​on the mask. Generally, spatial attention mechanisms calculate the average across different channels of the same pixel, then perform convolution and upsampling operations to obtain spatial features, with each pixel in each layer of the spatial features being assigned different weights.

[0070] Specifically, in step 140, the multiple local component chromatographic feature matrices are aggregated along the channel dimension into a three-dimensional input tensor, and then passed through a second convolutional neural network model as a feature map to obtain an inter-component correlation chromatographic feature map. Furthermore, considering the correlation between the implicit correlation feature information regarding the various estrogens in the local chromatograms of each sampling window—that is, each local chromatogram of the sampling window contains correlation feature information regarding the content of the various estrogens—and this feature information is also correlated in the local chromatograms of each sampling window.

[0071] Therefore, in order to fully detect the content of various estrogens in the water sample to be tested, the technical solution of this application further aggregates the multiple local component chromatographic feature matrices along the channel dimension into a three-dimensional input tensor, and then performs feature mining through a second convolutional neural network model as a feature map to extract the correlation feature distribution information between the implicit correlation features of various estrogens in the local chromatograms of each sampling window, thereby obtaining the component correlation chromatographic feature map. In particular, here, the convolution kernel of the second convolutional neural network model is a three-dimensional convolution kernel, which has W (width), H (height) and C (channel dimension). In the technical solution of this application, the channel dimension of the three-dimensional convolution kernel corresponds to the channel dimension of the three-dimensional input tensor. Therefore, when performing three-dimensional convolutional encoding, the high-dimensional implicit correlation feature distribution information of the content features of various estrogens between the local chromatograms of each sampling window can be extracted from the three-dimensional input tensor.

[0072] After aggregating the multiple local component chromatographic feature matrices along the channel dimension into a three-dimensional input tensor, the intercomponent correlation chromatographic feature map is obtained by passing it through a second convolutional neural network model, which serves as the feature map. This includes: using each layer of the second convolutional neural network model as the feature map to perform convolution processing, mean pooling based on the feature matrix, and nonlinear activation processing on the input data during the forward propagation of the layer, so that the intercomponent correlation chromatographic feature map is output by the last layer of the second convolutional neural network model as the feature map, wherein the input of the first layer of the second convolutional neural network model as the feature map is the three-dimensional input tensor.

[0073] As you can understand, a Convolutional Neural Network (CNN) is an artificial neural network with wide applications in fields such as image recognition. A CNN can consist of an input layer, hidden layers, and an output layer. Hidden layers can include convolutional layers, pooling layers, activation layers, and fully connected layers. Each layer performs calculations on the input data and outputs the results to the next layer. The initial input data undergoes multiple layers of computation to arrive at a final result.

[0074] In recent years, deep learning and neural networks have been widely applied in fields such as computer vision, natural language processing, and text signal processing. Furthermore, deep learning and neural networks have demonstrated near-human or even surpassed human-level performance in areas such as image classification, object detection, semantic segmentation, and text translation. The development of deep learning and neural networks provides new approaches and solutions for mining the correlation feature distribution information of various hidden features in the vibration signals of rail transit vehicles.

[0075] Specifically, in step 150, the inter-component correlation chromatographic feature map is enhanced with feature discriminability along the channel dimension to obtain a decoded feature map. Figure 5 The flowchart below shows step 150 of the method for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application. Figure 5 As shown, the method for enhancing the discriminability of the inter-component correlation chromatographic feature map along the channel dimension to obtain a decoded feature map includes: 151, calculating the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map to obtain a weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors; and 152, using the feature values ​​at each position of the weighted feature vector as weights to weight each feature matrix of the inter-component correlation chromatographic feature map along the channel dimension to obtain the classification feature map.

[0076] Specifically, in the technical solution of this application, the feature discriminability enhancement along the channel dimension of the inter-component correlation chromatographic feature map is performed to obtain a decoded feature map. For example, a channel-weighted feature vector can be obtained by performing global mean pooling on each feature matrix arranged along the channel dimension of the inter-component correlation chromatographic feature map, thereby weighting each feature matrix arranged along the channel dimension of the inter-component correlation chromatographic feature map. However, considering that global mean pooling of the feature matrix inevitably loses the discriminative feature spatial distribution information expressed by the feature matrix, it is desirable to optimize the channel-weighted feature vector to recover as much as possible the feature distribution information of the feature vector expressed by the overall distribution correlation between each feature matrix in the inter-component correlation chromatographic feature map, thereby improving the expression effect of the channel-weighted feature vector on the channel-dimensional feature correlation distribution between each feature matrix.

[0077] Based on this, when performing channel-dimensional global mean pooling on the inter-component correlation chromatographic feature map, the applicant of this application calculates the instance normalization and consistency correlation recovery factor of the global mean pooling for each feature matrix. Specifically, the instance normalization and consistency correlation recovery factor of the global mean pooling for each feature matrix of the inter-component correlation chromatographic feature map are calculated using the following formula to obtain the weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors; wherein, the formula is: in, The first feature matrix of each feature matrix representing the intercomponent correlation chromatographic feature map is... Location feature value and These are the mean and variance of the set of eigenvalues ​​at each position of each feature matrix in the chromatographic feature map of the inter-component correlation, respectively. and These are the width and height of each feature matrix of the chromatographic feature map showing the correlation between the components. Represents the logarithmic function with base 2. This indicates the calculation of the natural exponential function value raised to the power of the numerical value. This represents the instance normalization and consistency-related recovery factor among the plurality of instance normalization and consistency-related recovery factors.

[0078] Here, the instance normalization and consistency-related recovery factor of global mean pooling address the inevitable loss of discriminative feature information in global mean pooling of the feature matrix. Based on instance normalization (IN) of the spatial distribution features of channel instances expressed by the feature matrix, consistency-related features are distilled from the statistical residual features of class probabilities to restore the coherence of consistency-related features in the statistical information to the eigenvalue representation. This achieves the recovery of the channel-weighted feature vector relative to the inter-component correlation chromatographic feature map, which is constrained by the spatial distribution of the feature matrix in terms of channel causality. Thus, by weighting the eigenvalues ​​at corresponding positions of the channel-weighted feature vector with this factor, the channel-weighted feature vector can contain the feature distribution information of the feature vector expressed by the overall distribution correlation between the various feature matrices in the inter-component correlation chromatographic feature map. This improves the expression effect of the channel-weighted feature vector on the channel-dimensional feature correlation distribution between the various feature matrices, thereby correspondingly improving the accuracy of the decoding regression of the decoded feature map. This allows for accurate detection of various estrogens in water samples, thereby reducing estrogen pollution of water resources.

[0079] Specifically, in step 160, the decoded feature map is passed through a first decoder to obtain a first decoded value, which represents the content of ethinylestradiol in the water sample to be tested. Next, the decoded feature map is further passed through the first decoder for decoding regression to obtain another first decoded value representing the content of ethinylestradiol in the water sample to be tested. This is done by cross-correcting based on the high-dimensional implicit correlation features of various estrogens in the high-performance liquid chromatogram of the water sample to be tested, thereby improving the accuracy of detecting the content of various estrogens and thus obtaining the content of ethinylestradiol in the water sample to be tested.

[0080] Specifically, the term "estrogens" refers to eight estrogens, namely ethinylestradiol, estrone, 17β-estradiol, estriol, 4-tert-octylphenol, nonylphenol, bisphenol A, and diethylstilbestrol. Correspondingly, in other specific examples of this application, second to eighth decoders can also be constructed to decode and detect the other seven estrogens, which is not limited to this application.

[0081] The first decoding value is obtained by passing the decoded feature map through a first decoder. The first decoding value represents the content of ethinylestradiol in the water sample to be tested. This includes: using the decoder to perform decoding regression on the decoded feature map using the following formula to obtain the first decoding value; wherein the formula is: ,in This represents the decoded feature map. This represents the first decoded value. Represents the weight matrix. This represents the bias vector. This represents matrix multiplication.

[0082] It should be understood that in this application, second to eighth decoders can also be constructed to decode and detect the other seven estrogens, which will not be elaborated upon here.

[0083] In summary, the method 100 for detecting trace amounts of estrogen-like substances in the aquatic environment based on the embodiments of this application is explained. It involves acquiring a high-performance liquid chromatography (HPLC) chromatogram of the water sample to be tested; employing deep learning-based artificial intelligence technology to mine high-dimensional implicit correlation features of various estrogens in the HPLC chromatogram of the water sample; and performing mutual correction of various estrogen characteristics based on these high-dimensional implicit correlation features. This allows for accurate detection of various estrogens in the water sample, thereby reducing estrogen pollution of water resources.

[0084] In one embodiment of this application, Figure 6 This is a block diagram of a system for detecting trace amounts of estrogen-like substances in an aquatic environment according to an embodiment of this application. Figure 6As shown, a detection system 200 for trace estrogen-like substances in an aquatic environment according to an embodiment of this application includes: a chromatogram acquisition module 210 for acquiring a high-performance liquid chromatogram of a water sample to be tested; a sample window scanning module 220 for performing sample window scanning on the high-performance liquid chromatogram to obtain multiple local chromatograms of sampling windows; a spatial attention module 230 for obtaining multiple local component chromatogram feature matrices by using a first convolutional neural network model with a spatial attention mechanism on the multiple local chromatograms of sampling windows respectively; a feature extraction module 240 for aggregating the multiple local component chromatogram feature matrices into a three-dimensional input tensor along the channel dimension and then passing it through a second convolutional neural network model as a feature map to obtain an inter-component correlation chromatogram feature map; an enhancement module 250 for enhancing the feature discriminability of the inter-component correlation chromatogram along the channel dimension to obtain a decoded feature map; and a decoding module 260 for passing the decoded feature map through a first decoder to obtain a first decoded value, wherein the first decoded value represents the content value of ethinylestradiol in the water sample to be tested.

[0085] In a specific example, in the above-mentioned detection system for trace estrogen-like substances in the aquatic environment, the spatial attention module is used to: each layer of the first convolutional neural network model using the spatial attention mechanism performs the following operations on the input data during the forward propagation of the layer: convolution processing on the input data to generate a convolutional feature map; pooling processing on the convolutional feature map to generate a pooled feature map; nonlinear activation on the pooled feature map to generate an activation feature map; calculating the mean value along the channel dimension of each position of the activation feature map to generate a spatial feature matrix; calculating the softmax-like function value at each position in the spatial feature matrix to obtain a spatial score matrix; and calculating the positional dot product of the spatial feature matrix and the spatial score matrix to obtain a feature matrix; wherein the feature matrix output by the last layer of the first convolutional neural network model using the spatial attention mechanism is the plurality of local component chromatographic feature matrices.

[0086] In a specific example, in the above-mentioned detection system for trace estrogen-like substances in the aquatic environment, the feature extraction module is used to: perform convolution processing, mean pooling based on the feature matrix, and nonlinear activation processing on the input data in the forward propagation of the layers of the second convolutional neural network model that serves as the feature map, so that the last layer of the second convolutional neural network model that serves as the feature map outputs the inter-component correlation chromatographic feature map, wherein the input of the first layer of the second convolutional neural network model that serves as the feature map is the three-dimensional input tensor.

[0087] In a specific example, in the above-mentioned detection system for trace estrogen-like substances in the aquatic environment, the enhancement module includes: a recovery factor calculation unit, used to calculate the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map to obtain a weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors; and a weighting unit, used to weight each feature matrix along the channel dimension of the inter-component correlation chromatographic feature map respectively using the feature values ​​at each position of the weighted feature vector as weights to obtain the classification feature map.

[0088] In a specific example, in the above-mentioned detection system for trace estrogen-like substances in the aquatic environment, the recovery factor calculation unit is used to: calculate the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map using the following formula to obtain the weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors; wherein, the formula is: in, The first feature matrix of each feature matrix representing the intercomponent correlation chromatographic feature map is... Location feature value and These are the mean and variance of the set of eigenvalues ​​at each position of each feature matrix in the chromatographic feature map of the inter-component correlation, respectively. and These are the width and height of each feature matrix of the chromatographic feature map showing the correlation between the components. Represents the logarithmic function with base 2. This indicates the calculation of the natural exponential function value raised to the power of the numerical value. This represents the instance normalization and consistency-related recovery factor among the plurality of instance normalization and consistency-related recovery factors.

[0089] In a specific example, in the above-mentioned detection system for trace estrogen-like substances in the aquatic environment, the decoding module is used to: perform decoding regression on the decoding feature map using the decoder according to the following formula to obtain the first decoded value; wherein, the formula is: ,in This represents the decoded feature map. This represents the first decoded value. Represents the weight matrix. This represents the bias vector. This represents matrix multiplication.

[0090] Here, those skilled in the art will understand that the specific functions and operations of each unit and module in the above-described detection system for trace estrogen-like substances in the aquatic environment have been referenced above. Figures 1 to 5 The method for detecting trace amounts of estrogen-like substances in the aquatic environment has been described in detail, and therefore, its repeated description will be omitted.

[0091] As described above, the detection system 100 for trace estrogen-like substances in the aquatic environment according to embodiments of this application can be implemented in various terminal devices, such as servers for detecting trace estrogen-like substances in the aquatic environment. In one example, the detection system 100 for trace estrogen-like substances in the aquatic environment according to embodiments of this application can be integrated into the terminal device as a software module and / or a hardware module. For example, the detection system 100 for trace estrogen-like substances in the aquatic environment can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the detection system 100 for trace estrogen-like substances in the aquatic environment can also be one of many hardware modules of the terminal device.

[0092] Alternatively, in another example, the detection system 100 for trace estrogen-like substances in the aquatic environment and the terminal device can also be separate devices, and the detection system 100 for trace estrogen-like substances in the aquatic environment can be connected to the terminal device via wired and / or wireless networks, and transmit interactive information in accordance with an agreed data format.

[0093] This application also provides a computer program product including instructions that, when executed, cause a device to perform operations corresponding to the methods described above.

[0094] In one embodiment of this application, a computer-readable storage medium is also provided, which stores a computer program that performs the methods described above.

[0095] It should be understood that embodiments of this application may be provided as methods, systems, or computer program products. Therefore, they may take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, they may take the form of computer program products implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] This application describes methods, systems, and computer program products using flowcharts and / or block diagrams. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, generate instructions for implementing the flowcharts and / or block diagrams. Figure 1 One or more processes and / or frames Figure 1 A device that provides the functions specified in one or more boxes.

[0097] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or frames Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or frames Figure 1 The steps of the function specified in one or more boxes.

[0099] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0100] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0101] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0102] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0103] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0104] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for detecting trace amounts of estrogen-like substances in an aquatic environment, characterized in that, include: Obtain the high-performance liquid chromatogram of the water sample to be tested; The high-performance liquid chromatogram is subjected to sample window scanning to obtain multiple partial chromatograms of the sampling window; The local chromatograms of the multiple sampling windows are respectively used to obtain multiple local component chromatographic feature matrices by using a first convolutional neural network model with spatial attention mechanism; The multiple local component chromatographic feature matrices are aggregated along the channel dimension into a three-dimensional input tensor, and then passed through a second convolutional neural network model as a feature map to obtain a chromatographic feature map of inter-component correlation. The feature discriminability enhancement along the channel dimension of the inter-component correlation chromatographic feature map is performed to obtain a decoded feature map; as well as The decoded feature map is passed through a first decoder to obtain a first decoded value, which is used to represent the content of ethinylestradiol in the water sample to be tested. The process of enhancing the feature discriminability of the inter-component correlation chromatographic feature map along the channel dimension to obtain a decoded feature map includes: Calculating the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map to obtain a weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors includes: calculating the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map using the following formula to obtain the weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors; The formula is as follows: ； in, The first feature matrix of each feature matrix representing the intercomponent correlation chromatographic feature map is... Location feature value and These are the mean and variance of the set of eigenvalues ​​at each position of each feature matrix in the chromatographic feature map of the inter-component correlation, respectively. and These are the width and height of each feature matrix of the chromatographic feature map showing the correlation between the components. Represents the logarithmic function with base 2. This indicates the calculation of the natural exponential function value raised to the power of the numerical value. This represents the instance normalization and consistency-related recovery factor in the plurality of instance normalization and consistency-related recovery factors; The feature values ​​at each position of the weighted feature vector are used as weights to weight each feature matrix along the channel dimension of the inter-component correlation chromatographic feature map to obtain the decoded feature map.

2. The method for detecting a trace amount of an estrogenic substance in an aquatic environment according to claim 1, characterized by, The multiple sampling window local chromatograms are respectively processed using a first convolutional neural network model with spatial attention mechanism to obtain multiple local component chromatographic feature matrices, including: each layer of the first convolutional neural network model using spatial attention mechanism processes the input data during the forward propagation of the layer: The input data is processed by convolution to generate convolutional feature maps; The convolutional feature map is subjected to pooling to generate a pooled feature map; The pooled feature map is nonlinearly activated to generate an activated feature map; Calculate the mean value along the channel dimension at each position of the activated feature map to generate a spatial feature matrix; Calculate the softmax function value at each position in the spatial feature matrix to obtain the spatial score matrix; and The feature matrix is ​​obtained by calculating the positional dot product of the spatial feature matrix and the spatial score matrix; The feature matrix output by the last layer of the first convolutional neural network model using spatial attention mechanism is the plurality of local component chromatographic feature matrices.

3. The method for detecting a trace amount of an estrogenic substance in an aquatic environment according to claim 2, characterized by, After aggregating the multiple local component chromatographic feature matrices along the channel dimension into a three-dimensional input tensor, the intercomponent correlation chromatographic feature map is obtained by passing it through a second convolutional neural network model, which serves as the feature map. This includes: using each layer of the second convolutional neural network model as the feature map to perform convolution processing, mean pooling based on the feature matrix, and nonlinear activation processing on the input data during the forward propagation of the layer, so that the intercomponent correlation chromatographic feature map is output by the last layer of the second convolutional neural network model as the feature map, wherein the input of the first layer of the second convolutional neural network model as the feature map is the three-dimensional input tensor.

4. The method for detecting a trace amount of an estrogenic substance in an aquatic environment according to claim 3, wherein The decoding feature map is passed through a first decoder to obtain a first decoding value, the first decoding value being used to represent the content value of ethinylestradiol in the water sample to be detected, including: using the decoder to perform decoding regression on the decoding feature map with the following formula to obtain the first decoding value; The formula is as follows: ,in, This represents the decoded feature map. This represents the first decoded value. Represents the weight matrix. This represents the bias vector. This represents matrix multiplication.

5. A detection system for trace estrogen-like substances in an aquatic environment, characterized in that, include: The chromatogram acquisition module is used to acquire the high-performance liquid chromatogram of the water sample to be tested; The sample window scanning module is used to perform sample window scanning on the high performance liquid chromatogram to obtain multiple sample window local chromatograms; The spatial attention module is used to obtain multiple local component chromatographic feature matrices by applying a first convolutional neural network model with a spatial attention mechanism to the local chromatograms of the multiple sampling windows. The feature extraction module is used to aggregate the multiple local component chromatographic feature matrices along the channel dimension into a three-dimensional input tensor, and then pass it through a second convolutional neural network model as a feature map to obtain a chromatographic feature map of inter-component correlation. The enhancement module is used to enhance the feature discriminability of the inter-component correlation chromatographic feature map along the channel dimension to obtain a decoded feature map; as well as A decoding module is used to pass the decoded feature map through a first decoder to obtain a first decoded value, wherein the first decoded value is used to represent the content of ethinylestradiol in the water sample to be detected; The enhancement module includes: The recovery factor calculation unit is used to calculate the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map to obtain a weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors, including: calculating the global mean pooling instance normalization and consistency correlation recovery factor of each feature matrix of the inter-component correlation chromatographic feature map according to the following formula to obtain the weighted feature vector composed of multiple instance normalization and consistency correlation recovery factors; The formula is as follows: ； in, The first feature matrix of each feature matrix representing the intercomponent correlation chromatographic feature map is... Location feature value and These are the mean and variance of the set of eigenvalues ​​at each position of each feature matrix in the chromatographic feature map of the inter-component correlation, respectively. and These are the width and height of each feature matrix of the chromatographic feature map showing the correlation between the components. Represents the logarithmic function with base 2. This indicates the calculation of the natural exponential function value raised to the power of the numerical value. This represents the instance normalization and consistency-related recovery factor in the plurality of instance normalization and consistency-related recovery factors; A weighting unit is used to weight each feature matrix along the channel dimension of the inter-component correlation chromatographic feature map using the feature values ​​at each position of the weighted feature vector as weights to obtain the decoded feature map.

6. The system for detecting a trace amount of an estrogenic substance in an aquatic environment according to claim 5, wherein The spatial attention module is used to: perform the following on the input data during the forward propagation of each layer of the first convolutional neural network model using the spatial attention mechanism: The input data is processed by convolution to generate convolutional feature maps; The convolutional feature map is subjected to pooling to generate a pooled feature map; The pooled feature map is nonlinearly activated to generate an activated feature map; Calculate the mean value along the channel dimension at each position of the activated feature map to generate a spatial feature matrix; Calculate the softmax function value at each position in the spatial feature matrix to obtain the spatial score matrix; as well as The feature matrix is ​​obtained by calculating the positional dot product of the spatial feature matrix and the spatial score matrix; The feature matrix output by the last layer of the first convolutional neural network model using spatial attention mechanism is the plurality of local component chromatographic feature matrices.

7. The system for detecting a trace amount of an estrogenic substance in an aquatic environment according to claim 6, wherein The feature extraction module is configured to: use each layer of the second convolutional neural network model, which serves as the feature map, to perform convolution processing, mean pooling based on the feature matrix, and nonlinear activation processing on the input data during the forward propagation of the layers, so that the last layer of the second convolutional neural network model, which serves as the feature map, outputs the intercomponent correlation chromatographic feature map, wherein the input of the first layer of the second convolutional neural network model, which serves as the feature map, is the three-dimensional input tensor.

Images