Tobacco tar component detection method and system based on spectral analysis

Abstract

The invention discloses a tobacco tar component detection method and system based on spectral analysis, and relates to the technical field of component detection.The tobacco tar component detection method comprises the steps that a to-be-detected tobacco tar sample is obtained, a spectrum collection device is activated to emit an optical signal within a preset wavelength range to the tobacco tar sample, and original spectral data is obtained; preprocessing the original spectral data to obtain preprocessed spectral data; according to a pre-trained component identification model, identifying the preprocessed spectral data to obtain a tobacco tar sample detection result; and comparing the tobacco tar sample detection result with a preset target component concentration threshold value to obtain a tobacco tar component detection result. The technical problem that in the prior art, the tobacco tar component detection result is unstable is solved, the near infrared spectrum data of the tobacco tar sample is collected, the multi-model integrated component recognition algorithm and the threshold comparison mechanism are combined, automatic recognition and judgment of multiple target components in the sample are achieved, and the detection accuracy is improved. The technical effect of improving the detection accuracy of tobacco tar components is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of component detection, and particularly to a method and system for detecting e-liquid components based on spectral analysis. Background Art

[0002] With the continuous improvement of the market's regulatory requirements for e-liquid quality, e-liquid component detection technology has received extensive attention in production quality control, safety monitoring, and formula preparation. However, with the diversification of product types and the complexity of formula structures, traditional e-liquid detection methods are difficult to accurately detect e-liquid components. Summary of the Invention

[0003] This application provides a method and system for detecting e-liquid components based on spectral analysis, which is used to solve the technical problem of unstable e-liquid component detection results in the prior art.

[0004] In view of the above problems, this application provides a method and system for detecting e-liquid components based on spectral analysis.

[0005] In the first aspect of this application, a method for detecting e-liquid components based on spectral analysis is provided. The method includes:

[0006] Obtain an e-liquid sample to be detected, activate a spectral acquisition device to emit an optical signal within a preset wavelength range to the e-liquid sample to obtain original spectral data; preprocess the original spectral data to obtain preprocessed spectral data; according to a pre-trained component recognition model, recognize the preprocessed spectral data to obtain an e-liquid sample detection result; compare the e-liquid sample detection result with a preset target component concentration threshold to obtain an e-liquid component detection result.

[0007] In the second aspect of this application, a system for detecting e-liquid components based on spectral analysis is provided. The system includes:

[0008] An original spectral data acquisition module, configured to obtain an e-liquid sample to be detected, activate a spectral acquisition device to emit an optical signal within a preset wavelength range to the e-liquid sample to obtain original spectral data; a preprocessing module, configured to preprocess the original spectral data to obtain preprocessed spectral data; a component recognition module, configured to recognize the preprocessed spectral data according to a pre-trained component recognition model to obtain an e-liquid sample detection result; a detection result acquisition module, configured to compare the e-liquid sample detection result with a preset target component concentration threshold to obtain an e-liquid component detection result.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] This application obtains a vape oil sample to be detected, activates a spectral acquisition device to emit an optical signal within a preset wavelength range to the vape oil sample, and obtains original spectral data; preprocesses the original spectral data to obtain preprocessed spectral data; according to a pre-trained component recognition model, recognizes the preprocessed spectral data to obtain a detection result of the vape oil sample; compares the detection result of the vape oil sample with a preset target component concentration threshold to obtain a detection result of the vape oil components. The present invention solves the technical problem of unstable detection results of vape oil components in the prior art. By collecting near-infrared spectral data of the vape oil sample, combining a component recognition algorithm with multi-model integration and a threshold comparison mechanism, it realizes the automatic recognition and judgment of multiple target components in the sample, achieving the technical effect of improving the detection accuracy of vape oil components. Description of the Drawings

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0012] Figure 1 It is a schematic flowchart of a method for detecting vape oil components based on spectral analysis provided by an embodiment of this application;

[0013] Figure 2 It is a schematic structural diagram of a system for detecting vape oil components based on spectral analysis provided by an embodiment of this application.

[0014] Explanation of reference numerals: The original spectral data acquisition module 11, the preprocessing module 12, the component recognition module 13, the detection result acquisition module 14. Detailed Embodiments

[0015] This application provides a method and system for detecting vape oil components based on spectral analysis. Aiming at solving the technical problem of unstable detection results of vape oil components in the prior art, by collecting near-infrared spectral data of the vape oil sample, combining a component recognition algorithm with multi-model integration and a threshold comparison mechanism, it realizes the automatic recognition and judgment of multiple target components in the sample, achieving the technical effect of improving the detection accuracy of vape oil components.

[0016] The following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the drawings in the embodiments of this application. Obviously, the described embodiments are only some embodiments of this application, rather than all embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application.

[0017] It should be noted that any variations of the terms "including" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules not clearly listed or inherent to these processes, methods, products, or devices.

[0018] Example 1, as Figure 1 shown, the present application provides a method for detecting the components of e-liquid based on spectral analysis. The method includes:

[0019] Step S100: Obtain an e-liquid sample to be detected, activate a spectral acquisition device to emit an optical signal within a preset wavelength range to the e-liquid sample, and obtain original spectral data.

[0020] In the embodiment of the present application, first, an e-liquid sample prepared in advance is obtained, the sample is placed in the detection area of the spectral acquisition device, the spectral acquisition device is activated, a near-infrared optical signal within the range of 800 nm to 2500 nm is emitted by the light source unit inside the spectral acquisition device to irradiate the sample, the reflected optical signal after absorption or scattering by the e-liquid sample is received by a detector and converted into an electrical signal, and original spectral data containing the spectral response characteristics of the e-liquid components is obtained.

[0021] Furthermore, in the method provided by the embodiment of the application, obtaining an e-liquid sample to be detected, activating a spectral acquisition device to emit an optical signal within a preset wavelength range to the e-liquid sample, and obtaining original spectral data further includes:

[0022] Place the e-liquid sample in the detection area of the spectral acquisition device, and use the light source unit in the spectral acquisition device to emit a near-infrared light signal covering the range of 800 nm to 2500 nm to irradiate the e-liquid sample; receive the reflected optical signal after absorption or scattering by the e-liquid sample through the detector in the spectral acquisition device, and generate the original spectral data.

[0023] In the embodiment of the present application, first, the e-liquid sample is placed in the detection area of the spectral acquisition device, and then the light source unit in the spectral acquisition device is started to emit a near-infrared light signal covering the range of 800 nm to 2500 nm, and this wavelength range covers the characteristic absorption regions corresponding to common organic components (such as nicotine, glycerol, propylene glycol, and flavorings) in e-liquid.

[0024] After irradiation, the reflected optical signal after the action of the e-liquid sample is received through the detector in the spectral acquisition device, and the received optical signal is converted into a corresponding electrical signal to obtain original spectral data. The original spectral data refers to a two-dimensional spectral map with wavelength as the abscissa and absorption intensity or reflectivity as the ordinate.

[0025] Step S200: Preprocess the original spectral data to obtain preprocessed spectral data.

[0026] In the embodiments of the present application, when smoothing the original spectral data, a moving average method is adopted to remove high-frequency noise and make the spectral curve smoother. Then, baseline correction is performed to eliminate the overall shift of the curve caused by equipment drift or sample background by deducting the background signal or setting reference points. Next, normalization processing is carried out, and maximum normalization is used to unify the spectral data of different samples into the same numerical range. Through the above steps, preprocessed spectral data with lower noise, clearer features, and stronger comparability is obtained.

[0027] Step S300: Identify the preprocessed spectral data according to the pre-trained component identification model to obtain the detection result of the e-liquid sample.

[0028] In the embodiments of the present application, the preprocessed spectral data is input into a pre-trained component identification model. The preprocessed spectral data is processed by multiple sub-identification models in the component identification model to obtain the predicted component concentration results output by each sub-model. Then, the predicted values of each target component are sorted out, a component concentration prediction matrix is constructed, and the mean value of each component prediction result in the matrix is calculated. Finally, the concentration values of each target component are obtained as the detection result of the e-liquid sample.

[0029] Furthermore, in the method provided by the embodiments of the application, identifying the preprocessed spectral data according to the pre-trained component identification model to obtain the detection result of the e-liquid sample further includes:

[0030] Input the preprocessed spectral data into multiple sub-identification models in the pre-trained component identification model to obtain a sequence of predicted component concentration values output by each sub-model; based on the sequence of predicted component concentration values, extract multiple predicted concentration values corresponding to each target component to construct a component concentration prediction matrix; calculate the mean value of the multiple predicted values of each target component in the component concentration prediction matrix to obtain the final detected concentration result of each target component as the detection result of the e-liquid sample.

[0031] In the embodiments of the present application, first, the preprocessed spectral data is input into multiple sub-identification models in the pre-trained component identification model. These sub-models are constructed using different machine learning algorithms, including support vector machines, random forests, etc. Each sub-model is modeled based on a large number of e-liquid samples with known component labels during the training phase and has the ability to predict the component concentration of new sample spectral data. Through the parallel identification processing of multiple sub-identification models, the predicted concentration values of each target component (such as nicotine, glycerol, propylene glycol, etc.) of each model are obtained, forming a set of model output results, that is, a sequence of predicted component concentration values.

[0032] Next, the predicted component concentration value sequence is sorted and classified according to the target components, the predicted values of each component in each sub-model are extracted, and a component concentration prediction matrix is constructed. Each row of this matrix corresponds to a target component, and each column corresponds to the prediction result of a sub-model.

[0033] After that, the average values of the multiple predicted values of each target component in the component concentration prediction matrix are calculated to obtain the final concentration detection result of each target component, and this result is used as the detection result of the e-liquid sample.

[0034] Furthermore, in the method provided by the application embodiment, constructing the component recognition model further includes:

[0035] Obtain an e-liquid sample data set with component concentration information, collect the spectral data of each sample within a preset wavelength range respectively to form a labeled data set, where the labeled data set includes the spectral features of multiple e-liquid samples and the corresponding target component concentration labels; based on the labeled data set, use a variety of different machine learning algorithms for training respectively to construct multiple sub-recognition models; perform an integration process on the multiple sub-recognition models to fuse the multiple sub-recognition models into a component recognition model.

[0036] In the embodiment of the present application, first, an e-liquid sample data set is obtained from the historical database, where the historical database stores a large number of e-liquid sample records that have been experimentally detected, including the spectral data collected for each sample within the near-infrared wavelength range (for example, 800nm to 2500nm), and the concentration detection results of each target component (such as nicotine, glycerol, propylene glycol, and flavor) in the sample. Subsequently, the spectral data is paired with the corresponding component concentration values one by one to construct the labeled data set required for supervised learning.

[0037] Next, based on the labeled data set, a variety of different machine learning algorithms are used for training respectively, including support vector machines and random forest regression, etc., to construct the corresponding prediction models, so as to obtain multiple sub-recognition models, and each sub-recognition model can independently complete the prediction task of the concentration of each target component in the e-liquid sample.

[0038] After that, an integration process is performed on the multiple sub-recognition models, that is, the multiple sub-recognition models are combined into a unified component recognition model through an ensemble learning strategy. This process uses a simple averaging method to comprehensively combine the recognition results of multiple models, so as to realize the recognition of the multi-component concentration of the e-liquid sample.

[0039] Step S400: Compare the e-liquid sample detection result with a preset target component concentration threshold to obtain the e-liquid component detection result.

[0040] In the embodiments of the present application, first, the predicted concentration values of each target component in the e-liquid sample test results are extracted, including the concentration values of components such as nicotine, glycerin, propylene glycol, and flavor. Subsequently, the corresponding preset upper and lower concentration limits in the target component threshold library are called, and the predicted concentration of each component is compared with its corresponding threshold. According to the comparison result, a judgment label indicating whether the standard is met is generated. Finally, the predicted concentration value of each target component is combined with the corresponding judgment label to form an e-liquid component test result including quantitative test results and qualitative judgment information.

[0041] Further, in the method provided by the application embodiments, when comparing the e-liquid sample test results with the preset target component concentration thresholds to obtain the e-liquid component test results, it further includes:

[0042] Extract the predicted component concentration values of the e-liquid sample test results, and call the corresponding preset target component concentration thresholds stored in the target component threshold library; compare the predicted component concentration values with the preset target component concentration thresholds, and generate a judgment label based on the comparison result; combine the predicted component concentration values with the corresponding judgment labels to form the e-liquid component test results including quantitative test data and qualitative judgment results.

[0043] Further, the method provided by the application embodiments further includes:

[0044] The preset target component concentration thresholds include the upper and lower concentration limits corresponding to nicotine, glycerin, propylene glycol, and flavor.

[0045] In the embodiments of the present application, first, the concentration values corresponding to components such as nicotine, glycerin, propylene glycol, and flavor are extracted from the e-liquid sample test results. Then, the upper and lower concentration limits corresponding to the above target components are called from the pre-established target component threshold library.

[0046] Subsequently, the predicted concentration value of each component is compared with its corresponding preset threshold, and a judgment result is generated by comparing whether it exceeds the upper and lower limit ranges. If the predicted value is within the threshold range, a "qualified" label is generated; if it exceeds, it is marked as "unqualified" or "exceeding the standard".

[0047] Finally, the predicted concentration value of each component is combined with the corresponding judgment label and output to form a structured e-liquid component test result, including quantitative test data (i.e., concentration value) and qualitative judgment results (i.e., judgment label).

[0048] Further, in the method provided by the application embodiments, after obtaining the e-liquid component test results, it further includes:

[0049] Visualize the detection results of the e-liquid components; the visualization process includes displaying the predicted component concentration values and their corresponding determination labels in the form of charts, color markings, or dashboards.

[0050] In the embodiment of the present application, to more intuitively display the detection results of the e-liquid components, the detection results are visualized. Specifically, through visualization, the predicted concentration values of each target component and their corresponding determination labels are displayed in a graphical form. The display methods include presenting the concentration values in the form of bar charts, line charts, or radar charts, etc., for showing the relative relationships between components. The determination labels are represented in the form of color markings, such as using green to represent "qualified" and red to represent "exceeding the standard"; or adopting the form of a dashboard, and the position of the component within the set range is reflected by a pointer or scale. Through the above methods, a structured graphical display of the e-liquid component detection results is achieved.

[0051] In the embodiment of the present application, in summary, the embodiment of the present application has at least the following technical effects:

[0052] The present application obtains an e-liquid sample to be detected, activates a spectral acquisition device to emit an optical signal within a preset wavelength range to the e-liquid sample, and obtains original spectral data; preprocesses the original spectral data to obtain preprocessed spectral data; according to a pre-trained component recognition model, recognizes the preprocessed spectral data to obtain the detection results of the e-liquid sample; compares the detection results of the e-liquid sample with a preset target component concentration threshold to obtain the e-liquid component detection results. The present invention solves the technical problem of unstable e-liquid component detection results in the prior art. By collecting near-infrared spectral data of the e-liquid sample, combining a component recognition algorithm integrating multiple models and a threshold comparison mechanism, automatic recognition and judgment of multiple target components in the sample are realized, achieving the technical effect of improving the accuracy of e-liquid component detection.

[0053] Embodiment 2, based on the same inventive concept as the e-liquid component detection method based on spectral analysis in the foregoing embodiment, as Figure 2 shown, the present application provides an e-liquid component detection system based on spectral analysis. The system in the embodiment of the present application and the method embodiment are based on the same inventive concept. Among them, the system includes:

[0054] The original spectral data acquisition module 11 is used to obtain the e-liquid sample to be detected, activate the spectral acquisition device to emit an optical signal within a preset wavelength range to the e-liquid sample, and obtain the original spectral data; the preprocessing module 12 is used to preprocess the original spectral data to obtain the preprocessed spectral data; the component identification module 13 is used to identify the preprocessed spectral data according to the pre-trained component identification model to obtain the e-liquid sample detection result; the detection result acquisition module 14 is used to compare the e-liquid sample detection result with a preset target component concentration threshold to obtain the e-liquid component detection result.

[0055] Furthermore, the system is also used to implement the following functions:

[0056] Place the e-liquid sample in the detection area of the spectral acquisition device, and use the light source unit in the spectral acquisition device to emit a near-infrared band optical signal covering 800nm to 2500nm to irradiate the e-liquid sample; receive the reflected optical signal absorbed or scattered by the e-liquid sample through the detector in the spectral acquisition device to generate the original spectral data.

[0057] Furthermore, the system is also used to implement the following functions:

[0058] Input the preprocessed spectral data into multiple sub-identification models in the pre-trained component identification model to obtain a sequence of component concentration prediction values output by each sub-model; based on the sequence of component concentration prediction values, extract multiple concentration prediction values corresponding to each target component, and construct a component concentration prediction matrix; calculate the mean value of the multiple prediction values of each target component in the component concentration prediction matrix to obtain the final detection concentration result of each target component as the e-liquid sample detection result.

[0059] Furthermore, the system is also used to implement the following functions:

[0060] Obtain an e-liquid sample dataset with component concentration information, collect the spectral data of each sample within a preset wavelength range respectively to form a labeled dataset, and the labeled dataset includes the spectral characteristics of multiple e-liquid samples and the corresponding target component concentration labels; based on the labeled dataset, use multiple different machine learning algorithms for training respectively to construct multiple sub-identification models; perform an integration process on the multiple sub-identification models to fuse the multiple sub-identification models into a component identification model.

[0061] Furthermore, the system is also used to implement the following functions:

[0062] Extract the predicted values of the component concentrations in the e-liquid sample test results, and call the preset target component concentration thresholds corresponding to those stored in the target component threshold library; compare the predicted values of the component concentrations with the preset target component concentration thresholds, and generate a determination label based on the comparison result; combine the predicted values of the component concentrations with the corresponding determination labels to form the e-liquid component test results including quantitative test data and qualitative judgment results.

[0063] Further, the system is also used to implement the following functions:

[0064] The preset target component concentration thresholds include the upper and lower limits of the concentrations corresponding to nicotine, glycerin, propylene glycol, and flavorings.

[0065] Further, the system is also used to implement the following functions:

[0066] Perform visualization processing on the e-liquid component test results; the visualization processing includes displaying the predicted values of the component concentrations and their corresponding determination labels in the form of charts, color markings, or dashboards.

[0067] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above describes specific embodiments of this specification. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0068] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

[0069] This specification and the drawings are only exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A method for detecting smoke oil components based on spectral analysis, characterized in that: The method comprises: Obtaining a to-be-tested e-liquid sample, activating a spectrum acquisition device to emit a light signal within a preset wavelength range to the e-liquid sample, and obtaining raw spectrum data; Preprocessing the original spectral data to obtain preprocessed spectral data; According to the pre-trained component recognition model, the pre-processed spectral data is recognized to obtain the detection result of the e-liquid sample; The detection result of the e-liquid sample is compared with a preset target component concentration threshold to obtain the e-liquid component detection result.

2. The method for detecting smoke oil components based on spectral analysis according to claim 1, characterized in that: Obtain a smoke oil sample to be tested, activate a spectrum acquisition device to emit a light signal in a preset wavelength range to the smoke oil sample, and obtain raw spectrum data, including: Placing the e-liquid sample in the detection area of ​​a spectrum collection device, and using a light source unit in the spectrum collection device to emit a near-infrared light signal covering 800nm ​​to 2500nm to illuminate the e-liquid sample; The detector in the spectrum acquisition device receives the reflected light signal absorbed or scattered by the e-liquid sample to generate the original spectrum data.

3. The method for detecting smoke oil components based on spectral analysis according to claim 1, characterized in that: According to the pre-trained component recognition model, the pre-processed spectral data is recognized to obtain the e-liquid sample detection result, including: Inputting the preprocessed spectral data into a plurality of sub-recognition models in a pre-trained component recognition model to obtain a component concentration prediction value sequence output by each sub-model; Based on the component concentration prediction value sequence, extract multiple concentration prediction values ​​corresponding to each target component to construct a component concentration prediction matrix; The multiple predicted values ​​of each target component in the component concentration prediction matrix are averaged to obtain a final detection concentration result of each target component as the e-liquid sample detection result.

4. The method for detecting smoke oil components based on spectral analysis according to claim 3, characterized in that: Construct a component identification model, including: Acquire a data set of e-liquid samples with component concentration information, collect spectral data of each sample within a preset band range, and form a labeled data set, wherein the labeled data set includes spectral features of multiple e-liquid samples and corresponding target component concentration labels; Based on the labeled data set, a plurality of different machine learning algorithms are used for training to construct a plurality of sub-recognition models; The multiple sub-recognition models are integrated to fuse the multiple sub-recognition models into a component recognition model.

5. The method for detecting smoke oil components based on spectral analysis according to claim 1, characterized in that: The detection result of the e-liquid sample is compared with a preset target component concentration threshold to obtain the e-liquid component detection result, including: Extracting the predicted value of the component concentration of the e-liquid sample test result, and calling the corresponding preset target component concentration threshold stored in the target component threshold library; Comparing the predicted component concentration value with the preset target component concentration threshold, and generating a determination label according to the comparison result; The component concentration prediction value is combined with the corresponding judgment label to form the e-liquid component detection result including quantitative detection data and qualitative judgment result.

6. The method for detecting smoke oil components based on spectral analysis according to claim 5, characterized in that: The preset target component concentration thresholds include upper and lower concentration limits corresponding to nicotine, glycerin, propylene glycol and flavors.

7. The method for detecting smoke oil components based on spectral analysis according to claim 5, characterized in that: After obtaining the results of the e-liquid composition test, including: Performing visualization on the detection result of the e-liquid components; The visualization processing includes displaying the predicted component concentration values ​​and their corresponding determination labels in the form of charts, color markings or dashboards.

8. The smoke oil component detection system based on spectral analysis is characterized by: The system comprises: The original spectrum data acquisition module is used to acquire a smoke oil sample to be tested, activate the spectrum acquisition device to emit a light signal within a preset wavelength range to the smoke oil sample, and obtain the original spectrum data; A preprocessing module, used for preprocessing the raw spectral data to obtain preprocessed spectral data; A component identification module, used to identify the pre-processed spectral data according to a pre-trained component identification model to obtain a detection result of the e-liquid sample; The detection result acquisition module is used to compare the detection result of the e-liquid sample with a preset target component concentration threshold to obtain the e-liquid component detection result.

Images