An ai spectrogram analysis system and method

Abstract

The application discloses an AI spectrum analysis system and method, and belongs to the technical field of spectrum identification, wherein the system comprises the following modules: a spectrum identification module, which identifies a data carrier and extracts spectrum data required by AI analysis; a management naming module, which completes registration and binding of a user account and acquires naming rules; a data management module, which operates data transmitted by the spectrum identification module, including moving, copying and storing; a spectrum integral judgment module, which performs integral processing on spectrum data received by the data management module and compares and judges spectrum data types according to the integral processing result; and a running control module, which controls the execution process of the spectrum identification module, the management naming module, the data management module and the spectrum integral judgment module, thereby solving the data ambiguity problem caused by a database in the prior art and calculation errors caused by unclear integral point construction in the integral process.

Description

Technical Field

[0001] This invention belongs to the field of spectrum recognition technology, specifically relating to an AI spectrum analysis system and method. Background Technology

[0002] In the current development of intelligent spectral analysis, other companies have chosen large models as the core tool of their intelligent spectral recognition systems, attempting to improve analysis efficiency through the deep learning capabilities of these models. However, from the perspective of the core needs and compliance standards of the food and drug testing industry, the application of large models has key drawbacks that are difficult to avoid. This technology, with its innovative algorithm structure based on standards and integrating expert experience and scientific principles, achieves a precise breakthrough in addressing industry pain points, becoming the optimal solution that meets national standards and ISO requirements. The core requirement of national and ISO standards for testing and inspection results is "absolute accuracy." This baseline standard directly determines the feasibility and applicability of technical solutions. Although large-scale models can improve the accuracy of analysis through training with massive amounts of data, due to the inherent limitations of the technology, their databases always suffer from data ambiguity, leading to inherent uncertainties in the analysis results.

[0003] Even though the large-scale model achieves an industry-leading accuracy of 99.9%, the number of erroneous reports due to model uncertainties remains significant. The leakage of such erroneous reports would expose regulatory loopholes and create a crisis of market trust, which is unacceptable to government regulators and third-party testing companies. More importantly, to investigate these potential erroneous reports, companies and regulatory agencies need to invest substantial additional manpower for secondary manual verification. This not only fails to improve efficiency but also significantly increases human resource costs, completely contradicting the original purpose of applying intelligent technologies. Summary of the Invention

[0004] To address the shortcomings of existing technologies, such as insufficient accuracy in spectral analysis of large models and slow speed of manual spectral recognition, this invention proposes an AI-based spectral analysis system. Includes the following modules: Spectrum recognition module: Identifies the data carrier and extracts the spectrum data required for AI analysis; Naming Management Module: Completes user account registration and binding, retrieves naming rules, and obtains user registration information; Data management module: Performs operations on the data transmitted from the spectrum recognition module, including moving, copying, and storing; Spectrum Integration Judgment Module: Performs integration processing on the spectrum data received by the data management module, and compares and judges the data type of the spectrum based on the integration processing result; The operation control module controls the execution process of the spectrum recognition module, the management naming module, the data management module, and the spectrum integration judgment module.

[0005] Preferred data carriers include PDF, Excel, and internal instrument data.

[0006] Ideally, settings should be configured for the standard, the sample, the parallel sample, the spiked sample, the blank matrix, the spiked blank matrix, the blank sample matrix, and the quality control sample.

[0007] Preferably, the settings for standard products include: fixed prefix, side length numbers, fixed suffix, fixed length numbers, fixed identifiers, and special characters; the settings for samples include: fixed prefix, side length numbers, fixed suffix, fixed length numbers, fixed identifiers, and special characters.

[0008] Preferably, the data management module can perform the copying of spectral data between the computer and the removable storage medium, including: copying the analysis data stored in the removable storage medium to the computer, copying the spectral database on the computer to the removable storage medium, and storing the data transferred by the spectral recognition module.

[0009] Ideally, the spectrum integration judgment module classifies the samples according to the naming rules obtained by the naming system. Based on the sample type, it compares the integrated data with the original data to detect problems or contents that need to be modified in the sample and proposes corresponding modification suggestions.

[0010] A preferred approach is to process the integral in the spectrum integral judgment module as follows: S1: Convert the information contained in the data carrier into spectral information; S2: Based on the collected spectral information, determine whether the spectrum is a single-peak or multi-peak spectrum; S3: Perform baseline integration and tailing peak processing on the unimodal graph, and perform segmentation integration on the bimodal graph; S4: Compare the area of ​​the integrated result and output the final integrated result.

[0011] Preferably, when the removable storage medium is connected to the computer, the operation control module calls the registration information generated by the management and naming module; when the removable storage medium is connected to the computer and exits the spectrum recognition module, management and naming module, and spectrum integration judgment module, the data management module is automatically called to perform a data copying operation, copying the data information generated by the spectrum recognition module, management and naming module, and spectrum integration judgment module and saving it to the removable storage medium; and the data management module is automatically called to delete the spectrum data stored by the spectrum recognition module on time according to a custom time limit.

[0012] An AI spectral analysis method, characterized by the following steps: Step 1: Determine the type of compound; Step 2: Integrate the spectrum according to the compound type; Step 3: Test the peak curve of the standard sample; Step 4: Test the peak curves of the spiked sample and the test sample. Step 5: Output the results.

[0013] Preferably, step one includes: A1: Identify whether the compound is a multi-peaked compound or a single-peaked compound. If it is a single-peaked compound, proceed to step A2; if it is a multi-peaked compound, output an error. A2: Determine if there is a loss of scale on the horizontal axis of the spectrum of the single-peak compound. If there is, proceed to step A3; otherwise, proceed to step A4. A3: Find the RT of any standard in the single-peak compound spectrum. If it exists, proceed to step A5; otherwise, report an error. A4: Read the horizontal and vertical axis scales of the single-peak compound spectrum; A5: Find the unique scale mark of the point and the x-axis to form a scale.

[0014] Preferably, in step three, high-purity chemicals are selected to prepare a standard correction curve for the compounds in the food to be tested, which is used to correct the subsequent samples and spiked samples. After the identification is completed, the standard is first calculated and judged to determine the shape and position of the peaks and the relative error between the peaks of the standard.

[0015] Preferably, in step four, for the sample with added standard (i.e., the spiked sample), the peak curve is compared with that of the standard to check if the spiked sample has a peak. If there is no peak, the peak time obtained from the standard is used to search around this time. If a peak is found, it is integrated and the ion ratio and other relevant judgment criteria are explored to confirm whether this peak is the target peak. For the sample to be tested, it is necessary to determine whether the sample contains the target compound, and if so, how much. First, the original spectrum of the sample is searched to see if there is a peak. If there is a peak, it is judged by the ion ratio and other relevant criteria to see if this peak is normal. If there is no peak, the peak time of the standard and the peak time of the spiked sample are searched.

[0016] Compared with the prior art, the technical solution of the present invention has the following advantages / benefits: 1. This invention transforms the requirements for key aspects of spectral analysis, such as integration points, peak area calculation, and baseline correction, into explicit mathematical models and logical rules. This ensures that every point requiring calculation or integration has a clear standard basis, eliminating fuzzy definitions or subjective judgment space, thereby improving the accuracy of integration calculations.

[0017] 2. In the integral calculation, the key parameters, such as the definition of the peak's starting point, apex, and ending point, as well as the calculation of peak width and determination of half-peak height, were precisely located. Different types of spectral peak shapes were classified, summarized, and their features were extracted. Finally, an integral calculation system covering all scenarios and types was formed, ensuring the accuracy of analysis in scenarios without standards.

[0018] 3. A ruler retrieval strategy was established. Due to factors such as PDF quality and scaling when printing PDFs, the original ruler may be lost. Since AI recognition requires the ruler for calculations, we used technical means to retrieve the ruler, thereby increasing the accuracy of recognition. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the modules of an AI spectral analysis system according to the present invention.

[0021] Figure 2 This is a schematic diagram of regular peak shapes in the spectral recognition of the present invention.

[0022] Figure 3 This is a schematic diagram of irregular peak shapes in the spectral identification process of this invention.

[0023] Figure 4 This is a schematic diagram of overlapping peaks in the spectral identification process of this invention.

[0024] Figure 5 This is a schematic diagram of special peak shapes in the spectral identification of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions in the embodiments of this invention are described clearly and completely below. Obviously, the described embodiments are only a part of the embodiments of this invention, not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention. Therefore, the detailed description of the embodiments of this invention provided below is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the invention.

[0026] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it need not be further defined and explained in subsequent figures. Example 1

[0027] like Figure 1 As shown, an AI spectral analysis system includes the following modules: Spectrum recognition module: Identifies the data carrier and extracts the spectrum data required for AI analysis; Naming Management Module: Completes user account registration and binding, retrieves naming rules, and obtains user registration information; Data management module: Performs operations on the data transmitted from the spectrum recognition module, including moving, copying, and storing; Spectrum Integration Judgment Module: Performs integration processing on the spectrum data received by the data management module, and compares and judges the data type of the spectrum based on the integration processing result; The operation control module controls the execution process of the spectrum recognition module, the management naming module, the data management module, and the spectrum integration judgment module.

[0028] The naming rules in the management naming module include: setting the standards, setting the samples, setting the parallel samples, setting the spiked samples, setting the blank matrix, setting the spiked blank matrix, setting the blank sample matrix, and setting the quality control samples.

[0029] Setting parameters for standard products includes: fixed prefix, side length numbers, fixed suffix, fixed length numbers, fixed identifiers, and special characters; setting parameters for samples includes: fixed prefix, side length numbers, fixed suffix, fixed length numbers, fixed identifiers, and special characters.

[0030] These are used to construct standard correction curves, which are used to correct subsequent samples and spiked samples. After the settings are completed, calculations and discriminations are first performed on the standards. First, the shape and position of the peaks and the relative errors between the standard peaks are determined. Then, the number of qualified standards required by national standards, the regression value of the standard correction curve, the error of the peak elution time of the standard samples are determined, and it is determined whether the standards may have positive results.

[0031] These settings enable accurate calibration of the integration curve generated by the spectral equipment, and more accurately capture the characteristics of the spectral curve during the integration process. This greatly improves the accuracy of the integration of the spectral curve and enhances the accuracy of subsequent spectral analysis.

[0032] The data management module can perform the copying of spectral data between the computer and the removable storage medium, including: copying the analysis data stored in the removable storage medium to the computer, copying the spectral database on the computer to the removable storage medium, and storing the data transferred by the spectral recognition module.

[0033] The spectrum integration judgment module classifies the samples according to the naming rules system, compares the integrated data with the original data based on the sample type, detects the problems in the samples and the content that needs to be modified, and proposes corresponding modification suggestions.

[0034] The process of processing integrals in the spectrum integral judgment module is as follows: S1: Convert the information contained in the data carrier into spectral information; S2: Based on the collected spectral information, determine whether the spectrum is a single-peak or multi-peak spectrum. If it is a single-peak spectrum, then proceed as follows: Figure 2 As shown, if it is a bimodal graph, then as Figure 3 As shown.

[0035] S3: Perform baseline integration and tailing peak processing on the unimodal graph, and perform segmentation integration on the bimodal graph; S4: Compare the area of ​​the integrated result and output the final integrated result.

[0036] If the following occurs Figure 4 , Figure 5 If the peak shape is irregular, an error will be reported directly.

[0037] When the removable storage medium is connected to the computer, the operation control module calls the registration information generated by the management and naming module; when the removable storage medium is connected to the computer and exits the spectrum recognition module, management and naming module, and spectrum integration judgment module, the data management module is automatically called to perform a data copy operation, copying the data information generated by the spectrum recognition module, management and naming module, and spectrum integration judgment module and saving it to the removable storage medium; according to a custom time limit, the data management module is automatically called to delete the spectrum data stored by the spectrum recognition module on time.

[0038] Example 2: An AI spectral analysis method, comprising the following steps: Step 1: Determine the type of compound; Step 2: Integrate the spectrum according to the compound type; Step 3: Test the peak curve of the standard sample; Step 4: Test the peak curves of the spiked sample and the test sample. Step 5: Output the results.

[0039] Step one includes: A1: Identify whether the compound is a multi-peaked compound or a single-peaked compound. If it is a single-peaked compound, proceed to step A2; if it is a multi-peaked compound, output an error. A2: Determine if there is a loss of scale on the horizontal axis of the spectrum of the single-peak compound. If there is, proceed to step A3; otherwise, proceed to step A4. A3: Find the elution time of any standard in the single-peak compound spectrum, i.e., the sample peak time. If it exists, proceed to step A5; otherwise, report an error. A4: Read the horizontal and vertical axis scales of the single-peak compound spectrum; A5: Find the unique scale mark of the point and the x-axis to form a scale.

[0040] In step three, high-purity chemicals are selected to prepare a standard correction curve for the compounds in the food to be tested, which is used to correct the subsequent samples and spiked samples. After the identification is completed, the standard is first calculated and judged to determine the shape and position of the peaks and the relative error between the peaks of the standard.

[0041] In step four, the spiked sample normally has a numerical value. The spiked sample's peak curve is compared with that of the standard to ensure its accuracy and completeness of integration. Then, check if the spiked sample has a peak. If there is no peak, search around the peak time obtained from the standard. If a peak is found, integrate it and explore the ion ratio and other relevant criteria to confirm whether it is the target peak. For the sample to be tested, it is necessary to determine whether the sample contains the target compound, and if so, how much. First, check if the sample has a peak in the original spectrum. If a peak is found, determine whether it is normal based on the ion ratio and other relevant criteria. If there is no peak, search around the peak time of the standard and the spiked sample.

[0042] The above are merely preferred embodiments of the present invention. It should be noted that the above preferred embodiments should not be considered as limitations on the present invention, and the scope of protection of the present invention should be determined by the scope defined in the claims. For those skilled in the art, several improvements and modifications can be made without departing from the spirit and scope of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An AI spectral analysis system, characterized in that, Includes the following modules: Spectrum recognition module: Identifies the data carrier and extracts the spectrum data required for AI analysis; Naming Management Module: Completes user account registration and binding, retrieves naming rules, and obtains user registration information; Data management module: Performs at least the following actions on the data transmitted from the spectrum recognition module: moving, copying, and storing; Spectrum Integration Judgment Module: Performs integration processing on the spectrum data received by the data management module, and compares and judges the data type of the spectrum based on the integration processing result; The operation control module controls the execution process of the spectrum recognition module, the management and naming module, the data management module, and the spectrum integration judgment module.

2. The AI ​​spectral analysis system according to claim 1, characterized in that, The naming rules in the management naming module include: setting the standards, setting the samples, setting the parallel samples, setting the spiked samples, setting the blank matrix, setting the spiked blank matrix, setting the blank sample matrix, and setting the quality control samples.

3. The AI ​​spectral analysis system according to claim 2, characterized in that, The settings for the standard product include: fixed prefix, side length number, fixed suffix, fixed length number, fixed identifier, and special characters; the settings for the sample include: fixed prefix, side length number, fixed suffix, fixed length number, fixed identifier, and special characters.

4. The AI ​​spectral analysis system according to claim 1, characterized in that, The data management module is capable of copying spectral data between a computer and a removable storage medium, including: copying the analysis data stored in the removable storage medium to the computer, copying the spectral database on the computer to the removable storage medium, and storing the data transmitted by the spectral recognition module.

5. The AI ​​spectral analysis system according to claim 1, characterized in that, The spectrum integration judgment module classifies the samples according to the naming rules system, compares the integrated data with the original data based on the sample type, detects the problems in the samples and the content that needs to be modified, and proposes corresponding modification suggestions.

6. The AI ​​spectral analysis system according to claim 5, characterized in that, The integration process in the spectrum integration judgment module is as follows: S1: Convert the information contained in the data carrier into spectral information; S2: Based on the collected spectral information, determine whether the spectrum is a single-peak or multi-peak spectrum; S3: Perform baseline integration and tailing peak processing on the unimodal graph, and perform segmentation integration on the bimodal graph; S4: Compare the area of ​​the integrated result and output the final integrated result.

7. The AI ​​spectral analysis system according to claim 1, characterized in that, When the removable storage medium is connected to the computer, the operation control module calls the registration information generated by the management and naming module; when the removable storage medium is connected to the computer and exits the spectrum recognition module, management and naming module, and spectrum integration judgment module, the data management module is automatically called to perform a data copying operation, copying the data information generated by the spectrum recognition module, management and naming module, and spectrum integration judgment module and saving it to the removable storage medium; according to a custom time limit, the data management module is automatically called to delete the spectrum data stored by the spectrum recognition module on time.

8. An AI spectral analysis method, characterized in that, Includes the following steps: Step 1: Determine the type of compound; Step 2: Integrate the spectrum according to the compound type; Step 3: Test the peak curve of the standard sample; Step 4: Test the peak curves of the spiked sample and the sample to be tested; Step 5: Output the results.

9. The AI ​​spectral analysis method according to claim 8, characterized in that, Step one includes: A1: Identify whether the compound is a multi-peaked compound or a single-peaked compound. If it is a single-peaked compound, proceed to step A2; if it is a multi-peaked compound, output an error. A2: Determine if there is a loss of scale on the horizontal axis of the spectrum of the single-peak compound. If there is, proceed to step A3; otherwise, proceed to step A4. A3: Find the sample elution time of any standard in the single-peak compound spectrum. If it exists, proceed to step A5; otherwise, report an error. A4: Read the horizontal and vertical axis scales of the single-peak compound spectrum; A5: Find the unique scale mark of the point and the x-axis to form a scale.

10. The AI ​​spectral analysis method according to claim 8, characterized in that, In step three, high-purity chemicals are selected to prepare a standard correction curve for the compounds in the food to be tested, which is used to correct subsequent samples and spiked samples. After identification, calculations and discrimination are first performed on the standard to determine the peak shape and position, as well as the relative error between standard peaks. In step four, for the sample with added standard (i.e., the spiked sample), the peak curve is compared with that of the standard to check if the spiked sample has a peak. If there is no peak, the peak time obtained from the standard is used to search around this time. If a peak is found, it is integrated, and the ion ratio and other relevant judgment criteria are explored to confirm whether this peak is the target peak. For the sample to be tested, it is necessary to determine whether the sample contains the target compound, and if so, how much. First, the original spectrum of the sample is searched for peaks. If there are peaks, the ion ratio and other relevant judgments are used to determine whether the peak is normal. If there are no peaks, the peak times of the standard and the spiked sample are searched.

Images