System and method for optimizing extraction of active ingredients from American ginseng based on data analysis

Abstract

The present invention belongs to the technical field of data analysis, and specifically relates to a system and method for optimizing the extraction of effective components of American ginseng based on data analysis, which comprises the following steps: obtaining solution samples containing effective components of American ginseng extracted under different extraction conditions, passing the solution samples through a liquid chromatography separation device to obtain a chromatogram of the sample, and extracting the content of effective components of the chromatogram after pretreatment; putting the effective component content of each solution sample and the control parameters when the solution sample is extracted into a set, obtaining a training sample from each set, and the label of the training sample is the control parameter; enhancing the training sample according to the functional relationship between the effective component content and the control parameter; determining a label attention matrix according to the functional relationship, and completing the training of a neural network model using the label attention matrix and the training sample; obtaining the target effective component content, and then inputting it into the trained neural network model after encoding to obtain the control parameter.

Description

Technical Field

[0001] The present invention belongs to the technical field of data analysis, and specifically relates to a system and method for optimizing the extraction of effective components of American ginseng based on data analysis. Background Art

[0002] American ginseng is a traditional Chinese herbal medicine, which is widely used in the field of traditional Chinese medicine due to its rich medicinal ingredients. The active ingredients in American ginseng include saponins, polysaccharides and volatile oils, which have multiple biological activities such as anti-fatigue, anti-inflammatory and immunity enhancement, and have important pharmacological value. However, the content of these active ingredients in American ginseng is low, and the direct use effect is limited. Through scientific extraction methods, these active ingredients can be separated and purified to improve their bioavailability and efficacy, and can provide high-purity raw materials for subsequent preparation development and clinical application, promoting the modernization and industrialization of traditional Chinese medicine.

[0003] Common methods for extracting American ginseng include solvent extraction, ultrasonic extraction and microwave extraction. These methods can extract the effective ingredients of American ginseng, such as saponins, to a certain extent, but they all rely on manual experience to set the extraction conditions, such as temperature, solvent concentration, etc. There is no way to quantitatively judge whether the extraction is efficient. Moreover, saponins are divided into many types. Even under the same conditions, the extraction rates of different types of saponins are different. Since the optimal extraction conditions for each effective ingredient are unknown, it is impossible to accurately control the extraction of American ginseng effective ingredients. Summary of the invention

[0004] The purpose of the present invention is to propose a method for optimizing the extraction of effective components of American ginseng based on data analysis to solve the problems raised in the above background technology.

[0005] The technical solution of the present invention is: a method for optimizing the extraction of effective components of American ginseng based on data analysis, comprising the following steps:

[0006] Obtain solution samples containing the effective components of American ginseng extracted under different extraction conditions, pass the solution samples through a liquid chromatography separation device to obtain a chromatogram of the sample, and extract the content of the effective components of the chromatogram after pretreatment;

[0007] The effective component content of each solution sample and the control parameters when the solution sample is extracted are put into a set, and a training sample is obtained from each set, and the label of the training sample is the control parameter; the training sample is enhanced according to the functional relationship between the effective component content and the control parameter;

[0008] The label attention matrix is ​​determined according to the functional relationship, and the neural network model is trained using the label attention matrix and training samples; the target active ingredient content is obtained, and then encoded and input into the trained neural network model to obtain control parameters.

[0009] Preferably, the training samples are enhanced according to the functional relationship between the active ingredient content and the control parameter, specifically:

[0010] According to the set data set, a functional relationship is established with the content of each active ingredient as a dependent variable and all control parameters as independent variables, at least one control parameter of any set in the set data set is changed, and inputted into the functional relationship corresponding to the active ingredient to obtain the content of each active ingredient, and the obtained content of each active ingredient and the changed control parameter are put into an enhanced set to obtain a first enhanced set data set; wherein each active ingredient corresponds to a functional relationship;

[0011] According to the set data set, a functional relationship is established with each control parameter as a dependent variable and all active ingredient contents as independent variables, at least one active ingredient content of any set in the set data set is changed, and input into the functional relationship corresponding to the control parameter to obtain each control parameter, and the changed active ingredient content and control parameter are put into an enhanced set to obtain a second enhanced set data set; wherein each control parameter corresponds to a functional relationship;

[0012] If the normalized distance between the set in the first enhanced set data set and the set in the second enhanced set data set is less than a threshold, the average of the two sets whose distance is less than the threshold is taken as the enhanced sample.

[0013] Preferably, the threshold is determined by:

[0014] The distances of all training samples after normalization are calculated, and the threshold is obtained according to the minimum distance.

[0015] Preferably, determining the label attention matrix according to the functional relationship is specifically:

[0016] Obtain the functional relationship with each control parameter as the dependent variable and all active ingredient contents as the independent variable, and calculate the Pearson correlation coefficient r between the jth control variable and the ith active ingredient content in the function corresponding to the jth control variable ij , all the Pearson correlation coefficients constitute a label attention matrix, where i and j are positive integers, and i≤I, j≤J, I is the number of effective components, and J is the number of control parameters.

[0017] Preferably, the training of the neural network model using the label attention matrix and the training samples is specifically as follows:

[0018] The label attention matrix and the effective ingredient content feature matrix of the training sample are used to obtain the feature matrix of each sample, the feature matrix is ​​input into the neural network model, the loss is calculated according to the output and label of the neural network model, and the update of the neural network model is completed.

[0019] On the other hand, the present invention proposes a system for optimizing the extraction of effective components of American ginseng based on data analysis, comprising the following modules:

[0020] The solution sample determination module is used to obtain solution samples containing the effective components of American ginseng extracted under different extraction conditions, pass the solution samples through a liquid chromatography separation device to obtain a chromatogram of the sample, and extract the content of the effective components of the chromatogram after pretreatment;

[0021] The training sample enhancement module puts the effective component content of each solution sample and the control parameters when the solution sample is extracted into a set, obtains a training sample from each set, and the label of the training sample is the control parameter; enhances the training sample according to the functional relationship between the effective component content and the control parameter;

[0022] The control optimization module is used to determine the label attention matrix according to the functional relationship, and use the label attention matrix and training samples to complete the training of the neural network model; obtain the target active ingredient content, and then input it into the trained neural network model after encoding to obtain the control parameters.

[0023] Preferably, the training samples are enhanced according to the functional relationship between the active ingredient content and the control parameter, specifically:

[0024] According to the set data set, a functional relationship is established with the content of each active ingredient as a dependent variable and all control parameters as independent variables, at least one control parameter of any set in the set data set is changed, and inputted into the functional relationship corresponding to the active ingredient to obtain the content of each active ingredient, and the obtained content of each active ingredient and the changed control parameter are put into an enhanced set to obtain a first enhanced set data set; wherein each active ingredient corresponds to a functional relationship;

[0025] According to the set data set, a functional relationship is established with each control parameter as a dependent variable and all active ingredient contents as independent variables, at least one active ingredient content of any set in the set data set is changed, and input into the functional relationship corresponding to the control parameter to obtain each control parameter, and the changed active ingredient content and control parameter are put into an enhanced set to obtain a second enhanced set data set; wherein each control parameter corresponds to a functional relationship;

[0026] If the normalized distance between the set in the first enhanced set data set and the set in the second enhanced set data set is less than a threshold, the average of the two sets whose distance is less than the threshold is taken as the enhanced sample.

[0027] Preferably, the threshold is determined by:

[0028] The distances of all training samples after normalization are calculated, and the threshold is obtained according to the minimum distance.

[0029] Preferably, determining the label attention matrix according to the functional relationship is specifically:

[0030] Obtain the functional relationship with each control parameter as the dependent variable and all active ingredient contents as the independent variable, and calculate the Pearson correlation coefficient r between the jth control variable and the ith active ingredient content in the function corresponding to the jth control variable ij , all the Pearson correlation coefficients constitute a label attention matrix, where i and j are positive integers, and i≤I, j≤J, I is the number of effective components, and J is the number of control parameters.

[0031] Preferably, the training of the neural network model using the label attention matrix and the training samples is specifically as follows:

[0032] The label attention matrix and the effective ingredient content feature matrix of the training sample are used to obtain the feature matrix of each sample, the feature matrix is ​​input into the neural network model, the loss is calculated according to the output and label of the neural network model, and the update of the neural network model is completed.

[0033] In addition, the present invention also provides a computer-readable storage medium, on which a computer program is stored, and when executed by a processor, the method described above is implemented.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] Aiming at the problem that it is difficult to obtain ideal control conditions in the process of extracting the effective components of American ginseng, the present invention enhances the training samples according to the functional relationship between the effective components and the control parameters in the training samples, determines the label attention matrix according to the functional relationship, and uses the label attention matrix and the training samples to complete the training of the neural network model, thereby improving the understanding ability and accuracy of the model. The control parameters can be obtained through the neural network model according to the effective components extracted by the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of Embodiment 1;

[0037] Figure 2 This is the liquid phase separation chromatogram of American ginseng;

[0038] Figure 3 This is a structural diagram of the second embodiment. DETAILED DESCRIPTION

[0039] In this article, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0040] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] In the first embodiment of the present invention, the present invention proposes a method for optimizing the extraction of effective components of American ginseng based on data analysis, comprising the following steps:

[0042] S1, obtaining solution samples containing the effective components of American ginseng extracted under different extraction conditions, passing the solution samples through a liquid chromatography separation device to obtain a chromatogram of the sample, and extracting the content of the effective components of the chromatogram after pretreatment;

[0043] In the process of extracting American ginseng, the extraction method, solvent type, solvent concentration, extraction temperature, extraction time, solid-liquid ratio, and even particle size will affect the final extraction effect. Even under the same extraction conditions, the extraction effects on different active ingredients are different, where active ingredients include but are not limited to saponins, polysaccharides, volatile oils, etc., where saponins include Rb1, Rd, Re, Rg1, Rc, Rb2, etc. By changing the control parameters, solution samples containing American ginseng active ingredients extracted under different conditions can be obtained. The type and content of the active ingredients can be measured by liquid chromatography. After obtaining the chromatogram, the content of each active ingredient is extracted. In one embodiment, the maximum value within the range of the active ingredient position in the recipe diagram is used as the content of each active ingredient. Figure 2 A chromatogram of a solution sample is shown.

[0044] S2, putting the effective component content of each solution sample and the control parameters when the solution sample is extracted into a set, obtaining a training sample from each set, and the label of the training sample is the control parameter; enhancing the training sample according to the functional relationship between the effective component content and the control parameter;

[0045] After obtaining the active ingredient content of each solution sample, a training sample is constructed according to the active ingredient content and the control parameters. In order to facilitate data management, the active ingredient content and the control parameters are put into a set. For example, a sample is [ac1, ac2, ac3, ac4, ct1, ct2, ct3], where ac1-ac4 are the active ingredient contents, and ct1-ct3 are the corresponding control parameters. In the training sample, the label is the control parameter. In this way, when the user wants to obtain the target active ingredient content, the control parameter can be obtained by inputting the desired active ingredient content into the model according to the present invention.

[0046] Since extracting the solution sample of American ginseng and measuring the chromatogram are time-consuming and material-intensive, and the number of training samples is limited, in order to improve the model training effect, the training samples are enhanced. Specifically, the training samples are enhanced according to the functional relationship between the active ingredient content and the control parameters, as follows:

[0047] According to the set data set, a functional relationship is established with the content of each active ingredient as a dependent variable and all control parameters as independent variables, at least one control parameter of any set in the set data set is changed, and input into the functional relationship corresponding to the active ingredient to obtain the content of each active ingredient, and the obtained content of each active ingredient and the changed control parameter are put into the enhanced set to obtain a first enhanced set data set; wherein each active ingredient corresponds to a functional relationship;

[0048] The set data set includes multiple sets, i.e., samples. The sets in the set data set are used to establish a functional relationship with the content of each active ingredient as the dependent variable and the control parameter as the independent variable. The number of functional relationships is the same as the number of active ingredients. For example, ac1=f1(ct1, ct2, ct3). Similarly, the functional relationships corresponding to ac2, ac3, and ac4 can be obtained. When one or more of the control parameters are changed, ac1-ac4 will also change accordingly. The changed ct1-ct3 and ac1-ac4 obtained through the functional relationship are put into a set. When ct1-ct3 is changed multiple times, multiple sets can be obtained. These sets constitute the first enhanced set data set.

[0049] According to the set data set, a functional relationship is established with each control parameter as a dependent variable and all active ingredient contents as independent variables, at least one active ingredient content of any set in the set data set is changed, and input into the functional relationship corresponding to the control parameter to obtain each control parameter, and the obtained changed active ingredient content and control parameter are put into the enhanced set to obtain a second enhanced set data set; wherein each control parameter corresponds to a functional relationship;

[0050] By taking all active ingredient contents as independent variables and each control parameter as dependent variable, multiple functions can be obtained in the same way as the process of obtaining the first enhanced set data set. By changing at least one of the active ingredient contents, the second enhanced set data set can be obtained. Changing the active ingredient content and the control parameters are both based on the samples of S1, and need to be changed within a reasonable range, such as the temperature within a reasonable range. More specifically, the changed active ingredient content and control parameters are within the maximum and minimum ranges of the active ingredients and control parameters corresponding to all samples in S1.

[0051] If the normalized distance between the set in the first enhanced set data set and the set in the second enhanced set data set is less than a threshold, the average of the two sets whose distance is less than the threshold is taken as the enhanced sample.

[0052] Since different parameters have different units, in one embodiment, the sets in the first enhanced set data set and the sets in the second enhanced set data set are normalized. In a preferred embodiment, each parameter of the same type is normalized or the elements at the same position in the set are normalized, and the same parameter in all sets in the first enhanced set data set and the second enhanced set data set is normalized.

[0053] After normalization, two sets whose distance is less than a threshold are calculated in the first enhanced set data set and the second enhanced set data set. One of the two sets is located in the first enhanced set data set, and the other is located in the second enhanced set data set. If it is less than the threshold, the average value of the two sets is used as the enhanced sample.

[0054] In another embodiment, the threshold is determined by calculating the distance of all training samples after normalization, and obtaining the threshold according to the minimum distance. In the training samples, the minimum distance represents the degree of distinction between two samples. In a more specific embodiment, the threshold is obtained according to the minimum distance, specifically, the product of the minimum distance and the weight is used as the threshold, and the weight is less than 1, preferably, the weight is 0.4.

[0055] S3, determine the label attention matrix according to the functional relationship, use the label attention matrix and training samples to complete the training of the neural network model; obtain the target effective ingredient content, and then input it into the trained neural network model after encoding to obtain the control parameters.

[0056] After obtaining the enhanced training samples, the neural network model is trained. Due to the limited training samples, even after enhancement, the number of samples is still small compared to some classification and prediction models, and improving the model's expressiveness requires a larger model and training samples. In order to overcome this problem, the present invention adopts a label attention mechanism to prevent the overfitting of the model and enable the model to pay attention to label-related information. First, the label attention matrix is ​​determined according to the functional relationship, which is specifically:

[0057] Obtain the functional relationship with each control parameter as the dependent variable and all active ingredient contents as the independent variable, and calculate the Pearson correlation coefficient r between the jth control variable and the ith active ingredient content in the function corresponding to the jth control variable ij , all Pearson correlation coefficients constitute the label attention matrix, where i and j are positive integers, and i≤I, j≤J, I is the number of effective components, and J is the number of control parameters.

[0058] Still taking the above data as an example, if there are three control parameters, there will be three functional relationships. Each control variable corresponds to a function, and the independent variable of the function is the content of all effective ingredients. Assuming there are 4 effective ingredients, there are 4 variables. Calculate the Pearson correlation coefficient r between the jth control variable and the ith effective ingredient. ij , the larger the Pearson correlation coefficient, the greater the correlation between the control variable j and the effective component i. In this way, the label attention matrix can be obtained. If there are 4 effective components and 3 control parameters, a 3×4 matrix will be obtained.

[0059] Then, the label attention matrix and training samples are used to complete the training of the neural network model, specifically:

[0060] The label attention matrix and the effective ingredient content feature matrix of the training samples are used to obtain the feature matrix of each sample, the feature matrix is ​​input into the neural network model, the loss is calculated according to the output and label of the neural network model, and the update of the neural network model is completed.

[0061] Among them, the effective ingredient content feature matrix is ​​the effective ingredient vector in the sample, or the feature matrix of the effective ingredient vector in the sample. For example, the effective ingredient content in the training sample is word embedded to obtain the effective ingredient content embedding matrix of each training sample, and the embedding matrix is ​​used as the effective ingredient content feature matrix.

[0062] The feature matrix of each sample is obtained by using the label attention matrix and the feature matrix of the effective ingredient content of the training sample, including but not limited to calculating the cosine similarity and weighting of the label attention matrix and the feature matrix of the effective ingredient content of the training sample. For example, there are 4 effective components constituting a vector, and the label attention matrix is ​​3×4. The vector and the label attention matrix or the transpose of the label attention matrix are multiplied to obtain the feature matrix of the sample. Of course, other methods of the label attention mechanism can also be used, which are not specifically limited here. The feature matrix is ​​input into the neural network model, and the control parameters output by the model are compared with the control parameters in the sample to obtain the loss, thereby completing the update of the neural network model, that is, training. Neural network models include but are not limited to CNN, RNN, LSTM, MLP, etc.

[0063] By utilizing the technical solution of the present invention, the user can obtain control parameters by inputting the desired content of effective ingredients. For example, if the user wants to efficiently extract Rb1, the sample with the highest Rb1 is found, and the content of one or more effective ingredients including Rb1 is further increased in this sample. The obtained control parameters are control parameters that are beneficial to increasing the Rb1 content.

[0064] In a second embodiment of the present invention, Figure 3 As shown, the present invention proposes a system for optimizing the extraction of effective components of American ginseng based on data analysis, comprising the following modules:

[0065] The solution sample determination module is used to obtain solution samples containing the effective components of American ginseng extracted under different extraction conditions, pass the solution samples through a liquid chromatography separation device to obtain a chromatogram of the sample, and extract the content of the effective components of the chromatogram after pretreatment;

[0066] The training sample enhancement module puts the effective component content of each solution sample and the control parameters when the solution sample is extracted into a set, and obtains a training sample from each set. The label of the training sample is the control parameter; the training sample is enhanced according to the functional relationship between the effective component content and the control parameter;

[0067] The control optimization module is used to determine the label attention matrix according to the function relationship, and use the label attention matrix and training samples to complete the training of the neural network model; obtain the target effective ingredient content, and then input it into the trained neural network model after encoding to obtain the control parameters.

[0068] Preferably, the training samples are enhanced according to the functional relationship between the active ingredient content and the control parameter, specifically:

[0069] According to the set data set, a functional relationship is established with the content of each active ingredient as a dependent variable and all control parameters as independent variables, at least one control parameter of any set in the set data set is changed, and inputted into the functional relationship corresponding to the active ingredient to obtain the content of each active ingredient, and the obtained content of each active ingredient and the changed control parameter are put into an enhanced set to obtain a first enhanced set data set; wherein each active ingredient corresponds to a functional relationship;

[0070] According to the set data set, a functional relationship is established with each control parameter as a dependent variable and all active ingredient contents as independent variables, at least one active ingredient content of any set in the set data set is changed, and input into the functional relationship corresponding to the control parameter to obtain each control parameter, and the changed active ingredient content and control parameter are put into an enhanced set to obtain a second enhanced set data set; wherein each control parameter corresponds to a functional relationship;

[0071] If the normalized distance between the set in the first enhanced set data set and the set in the second enhanced set data set is less than a threshold, the average of the two sets whose distance is less than the threshold is taken as the enhanced sample.

[0072] Preferably, the threshold is determined in the following manner:

[0073] The distances of all training samples after normalization are calculated, and the threshold is obtained according to the minimum distance.

[0074] Preferably, determining the label attention matrix according to the functional relationship is specifically:

[0075] Obtain the functional relationship with each control parameter as the dependent variable and all active ingredient contents as the independent variable, and calculate the Pearson correlation coefficient r between the jth control variable and the ith active ingredient content in the function corresponding to the jth control variable ij , all the Pearson correlation coefficients constitute a label attention matrix, where i and j are positive integers, and i≤I, j≤J, I is the number of effective components, and J is the number of control parameters.

[0076] Preferably, the training of the neural network model using the label attention matrix and the training samples is specifically as follows:

[0077] The label attention matrix and the effective ingredient content feature matrix of the training sample are used to obtain the feature matrix of each sample, the feature matrix is ​​input into the neural network model, the loss is calculated according to the output and label of the neural network model, and the update of the neural network model is completed.

[0078] In a third embodiment of the present invention, the present invention further proposes a computer-readable storage medium, on which a computer program is stored, and when executed by a processor, the method described in the first embodiment is implemented.

[0079] In a fourth embodiment of the present invention, the present invention further proposes a computer device, comprising at least a memory and a processor, wherein a computer program is stored in the memory, and when executed by the processor, the method described in the first embodiment is implemented.

[0080] Through the description of the above implementation methods, technicians in this field can clearly understand that each implementation method can be implemented by adding a necessary general hardware platform, and of course can also be implemented by combining hardware and software. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a computer product, and the present invention can be in the form of a computer program product 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 codes.

[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it, and other embodiments may also be used. Although the present invention has been described in detail with reference to the aforementioned embodiments, ordinary technicians in this field should understand that they can still modify the technical solutions recorded in the aforementioned embodiments, or replace some of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the extraction of effective components from American ginseng based on data analysis, characterized in that: The following steps are involved: Obtain solution samples containing the effective components of American ginseng extracted under different extraction conditions, pass the solution samples through a liquid chromatography separation device to obtain a chromatogram of the sample, and extract the content of the effective components of the chromatogram after pretreatment; The effective component content of each solution sample and the control parameters when the solution sample is extracted are put into a set, and a training sample is obtained from each set, and the label of the training sample is the control parameter; the training sample is enhanced according to the functional relationship between the effective component content and the control parameter; Determine the label attention matrix according to the functional relationship, and use the label attention matrix and training samples to complete the training of the neural network model; obtain the target active ingredient content, and then input it into the trained neural network model after encoding to obtain the control parameters; The training samples are enhanced according to the functional relationship between the active ingredient content and the control parameters, specifically: According to the set data set, a functional relationship is established with the content of each active ingredient as a dependent variable and all control parameters as independent variables, at least one control parameter of any set in the set data set is changed, and inputted into the functional relationship corresponding to the active ingredient to obtain the content of each active ingredient, and the obtained content of each active ingredient and the changed control parameter are put into an enhanced set to obtain a first enhanced set data set; wherein each active ingredient corresponds to a functional relationship; According to the set data set, a functional relationship is established with each control parameter as a dependent variable and all active ingredient contents as independent variables, at least one active ingredient content of any set in the set data set is changed, and input into the functional relationship corresponding to the control parameter to obtain each control parameter, and the changed active ingredient content and control parameter are put into an enhanced set to obtain a second enhanced set data set; wherein each control parameter corresponds to a functional relationship; If the normalized distance between the set in the first enhanced set data set and the set in the second enhanced set data set is less than a threshold, the average of the two sets whose distance is less than the threshold is taken as the enhanced sample; The label attention matrix is ​​determined according to the functional relationship, specifically: Obtain the functional relationship with each control parameter as the dependent variable and all active ingredient contents as the independent variable, and calculate the Pearson correlation coefficient r between the jth control variable and the ith active ingredient content in the function corresponding to the jth control variable ij , all the Pearson correlation coefficients constitute a label attention matrix, where i and j are positive integers, and i≤I, j≤J, I is the number of effective components, and J is the number of control parameters.

2. The method for optimizing the extraction of effective components of American ginseng based on data analysis as claimed in claim 1, characterized in that: The threshold is determined in the following manner: The distances of all training samples after normalization are calculated, and the threshold is obtained according to the minimum distance.

3. The method for optimizing the extraction of effective components of American ginseng based on data analysis as claimed in claim 1, characterized in that: The label attention matrix and training samples are used to complete the training of the neural network model, specifically: The label attention matrix and the effective ingredient content feature matrix of the training sample are used to obtain the feature matrix of each sample, the feature matrix is ​​input into the neural network model, the loss is calculated according to the output and label of the neural network model, and the update of the neural network model is completed.

4. A data analysis-based optimization system for extracting active ingredients from American ginseng, characterized in that: Includes the following modules: The solution sample determination module is used to obtain solution samples containing the effective components of American ginseng extracted under different extraction conditions, pass the solution samples through a liquid chromatography separation device to obtain a chromatogram of the sample, and extract the content of the effective components of the chromatogram after pretreatment; The training sample enhancement module puts the effective component content of each solution sample and the control parameters when the solution sample is extracted into a set, obtains a training sample from each set, and the label of the training sample is the control parameter; enhances the training sample according to the functional relationship between the effective component content and the control parameter; A control optimization module is used to determine the label attention matrix according to the functional relationship, and use the label attention matrix and training samples to complete the training of the neural network model; obtain the target active ingredient content, and then input it into the trained neural network model after encoding to obtain the control parameters; The training samples are enhanced according to the functional relationship between the active ingredient content and the control parameters, specifically: According to the set data set, a functional relationship is established with the content of each active ingredient as a dependent variable and all control parameters as independent variables, at least one control parameter of any set in the set data set is changed, and inputted into the functional relationship corresponding to the active ingredient to obtain the content of each active ingredient, and the obtained content of each active ingredient and the changed control parameter are put into an enhanced set to obtain a first enhanced set data set; wherein each active ingredient corresponds to a functional relationship; According to the set data set, a functional relationship is established with each control parameter as a dependent variable and all active ingredient contents as independent variables, at least one active ingredient content of any set in the set data set is changed, and input into the functional relationship corresponding to the control parameter to obtain each control parameter, and the changed active ingredient content and control parameter are put into an enhanced set to obtain a second enhanced set data set; wherein each control parameter corresponds to a functional relationship; If the normalized distance between the set in the first enhanced set data set and the set in the second enhanced set data set is less than a threshold, the average of the two sets whose distance is less than the threshold is taken as the enhanced sample; The label attention matrix is ​​determined according to the functional relationship, specifically: Obtain the functional relationship with each control parameter as the dependent variable and all active ingredient contents as the independent variable, and calculate the Pearson correlation coefficient r between the jth control variable and the ith active ingredient content in the function corresponding to the jth control variable ij , all the Pearson correlation coefficients constitute a label attention matrix, where i and j are positive integers, and i≤I, j≤J, I is the number of effective components, and J is the number of control parameters.

5. The American ginseng active ingredient extraction optimization system based on data analysis as claimed in claim 4, characterized in that: The label attention matrix and training samples are used to complete the training of the neural network model, specifically: The label attention matrix and the effective ingredient content feature matrix of the training sample are used to obtain the feature matrix of each sample, the feature matrix is ​​input into the neural network model, the loss is calculated according to the output and label of the neural network model, and the update of the neural network model is completed.

6. A computer storage device having a computer program stored thereon, characterized in that: When executed by a processor, the method according to any one of claims 1 to 3 is implemented.

Images