Chemical new material quality detection method and system based on artificial intelligence

Abstract

The invention relates to the technical field of artificial intelligence detection, in particular to a chemical new material quality detection method and system based on artificial intelligence, and the method comprises the following steps: collecting a polyolefin copolymer infrared absorption spectrum, extracting an absorption peak parameter and building a spectrum section index, matching a standard interval to judge a functional group, and identifying structure change classification characteristics. The method comprises the following steps: constructing a spectrum segment structure sequence and label mapping, training an artificial intelligence model and deploying the artificial intelligence model in a system, realizing automatic identification and prediction of a new sample from a spectrum segment structure to functional group composition and a quality state, and outputting a material quality detection result set. Functional group identification is realized by combining standard interval matching, structural difference information is extracted according to absorbance and peak shape change, a mapping relation between a structural sequence and a sample label is established, systematic matching of structural features and quality standards is realized, and a response mode comparison method and an artificial intelligence model are introduced in the detection process. The structural integrity and functional group composition are automatically judged, the closed-loop capability and intelligent recognition precision of the detection process are improved, and the adaptability and detection consistency of complex materials are enhanced.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence detection technology, and in particular to a method and system for quality testing of new chemical materials based on artificial intelligence. Background Technology

[0002] The field of artificial intelligence (AI) testing technology involves the intelligent identification, judgment, and evaluation of materials, products, or systems using AI algorithms. It falls under the category of automated testing for studying or analyzing the physical or chemical properties of materials. This technology mainly includes intelligent identification methods based on models such as machine learning, deep learning, and neural networks. By combining these methods with image processing, spectral analysis, and pattern recognition, it enables the automatic analysis and judgment of test data. AI testing technology has important applications in material quality control, manufacturing process optimization, and automated testing systems. Its systematic characteristics are manifested in its algorithm-driven approach, data acquisition and training as its foundation, and model output as the result form, covering the entire process from sample collection and feature extraction to result determination. Traditional chemical new material quality testing methods and systems refer to manual or semi-automated testing methods used to ensure that the physical properties, chemical stability or other key indicators of various new chemical materials meet the standards during production or application. These traditional tests rely on spectroscopic instruments, chromatographic equipment or mechanical testing devices, combined with specific steps such as chemical analysis, infrared or ultraviolet absorption detection, thermal analysis, and tensile testing to obtain test parameters, which are then interpreted and recorded by professional technicians. These methods require a lot of manual operation and experience-based judgment during implementation, and data processing relies on statistical methods or set threshold judgments, which has limitations in the identification of complex materials and the detection of microscopic defects.

[0003] Existing technologies rely on traditional methods such as spectroscopic instruments, chromatographic equipment, and mechanical testing in material testing. The testing process is mainly manual, requiring professional technicians to manually interpret the test parameters. This approach relies on experience to identify complex chemical structures and microscopic material differences, resulting in a relatively limited information extraction dimension and difficulty in fully revealing the details of material structural changes. When processing multidimensional spectral data, it often uses set thresholds or statistical regularities for judgment, which is easily affected by external interference factors, leading to poor stability of test results. It has significant limitations in terms of identification accuracy and quantitative analysis capabilities. In particular, when dealing with the diverse testing needs of new chemical materials, it lacks a unified and systematic identification path and standardized modeling mechanism. Summary of the Invention

[0004] To address the technical problems existing in the prior art, this invention provides an artificial intelligence-based method for quality testing of new chemical materials, comprising the following steps: To achieve the above objectives, the present invention adopts the following technical solution: a method for quality testing of new chemical materials based on artificial intelligence, comprising the following steps: S1: Collect absorption spectrum data of polyolefin copolymers in the infrared band, extract the wavelength, peak intensity and bandwidth of absorption peaks, sort the spectral information by wavelength, establish a spectral index and generate a spectral response characteristic table. S2: Based on the spectral response characteristic table, extract the peak position and bandwidth of the spectral bands, and combine the standard absorption ranges of aromatic C=C, hydroxyl O–H and hydrocarbon C–H bonds to complete the spectral band matching and functional group assignment, and generate a functional group segment labeling information set. S3: Call the functional group segment annotation information set, identify functional group changes between spectral segments, extract absorbance differences and full width at half maximum (FWHM) changes, combine peak shape trend to identify structural change segments, classify change types, and generate a spectral segment structural difference feature set. S4: Based on the spectral segment structure difference feature set, summarize the functional group type, spectral shape parameters and change trends, construct the spectral segment structure sequence index, and map it with the material category label to generate a spectral input feature sequence file; S5: Based on the spectral input feature sequence file, combined with the functional group distribution and response mode, the structural features and quality standards are matched to determine the structural integrity and functional group composition of the material, and output the quality test result set of the new chemical material. S6: Based on the spectral input feature sequence archive and the label data of existing samples, train an artificial intelligence model to realize automatic identification and prediction from spectral structure to functional group composition and quality status. Deploy the trained model in the detection system, receive the spectral sequence input of new samples, and output the structural integrity and quality level judgment results.

[0005] As a further aspect of the present invention, the spectral response feature table includes a spectral segment index, peak position parameters, peak intensity parameters, bandwidth parameters, and normalized absorbance; the functional group segment annotation information set includes matching interval identifiers, functional group type labels, standard range numbers, and segment annotations; the spectral segment structural difference feature set includes functional group change positions, absorbance differences, half-width at half-maximum (FWHM) changes, and change type classifications; the spectral input feature sequence file includes a structural sequence number, functional group distribution sequence, spectral shape parameter vector, change trend index, and sample category mapping table; and the chemical new material quality testing result set includes structural integrity determination, composition status assessment, quality standard matching conclusions, and sample qualification conclusions.

[0006] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Acquire the absorption spectrum data of the polyolefin copolymer material to be tested in the infrared band, record the correspondence between light intensity and wavelength, detect the change of absorbance at the wavelength, calculate the absorption value based on the relationship between light intensity and absorbance, organize and form the spectral curve data of continuous bands, and generate an absorption spectrum dataset. S102: Based on the absorption spectrum dataset, identify the position of the maximum absorption peak in the band, calculate the peak wavelength and corresponding peak intensity, determine the bandwidth parameter according to the peak width range, arrange the peak parameters in wavelength order, and generate an absorption peak parameter sequence. S103: Based on the absorption peak parameter sequence, extract the band absorbance data, normalize the absorbance, adjust the peak intensity value to a uniform standard range, establish an index list including peak position, peak intensity, bandwidth and normalized absorbance, integrate it into spectral response data, and generate a spectral response feature table.

[0007] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Based on the spectral response feature table, extract the peak wavelength and bandwidth information in the spectral band, record the correspondence between peak position and bandwidth in the spectral band, organize the spectral band structure information according to wavelength order, and generate a peak bandwidth parameter list. S202: Based on the peak bandwidth parameter list, call the standard absorption wavelength range of the aromatic carbon-carbon double bond stretching vibration range, the hydroxyl group stretching vibration range and the carbon-hydrogen single bond stretching vibration range, determine whether the peak position falls into the standard range, obtain the correspondence between the spectral segment and the functional range, and generate the functional range matching judgment result. S203: Based on the functional interval matching determination result, identify the wavelength, bandwidth and functional group type of the spectral segment falling into the interval, construct the matching relationship between functional group category and spectral segment position, integrate the functional group interval information of each spectral segment, and generate a functional group segment annotation information set.

[0008] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Call the functional group segment annotation information set, compare the functional group types in adjacent spectral segments in turn, identify the positions where the functional group types change, record the spectral segment index information corresponding to the changes, and generate a functional group transformation position set; S302: Based on the functional group transformation position set, extract the absorbance difference and half-width variation of adjacent spectral bands, determine whether the dual conditions of absorption difference and peak shape variation are met, mark the spectral band positions that meet the conditions, and generate spectral band response change characteristic values. S303: Based on the characteristic values ​​of the spectral band response change, combine and judge the direction of absorbance difference and the trend of half width at half maximum (WHM) change, verify the structural change type corresponding to the spectral band, organize the matching relationship between structural change and spectral band position, and generate a spectral band structural difference feature set.

[0009] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Based on the spectral segment structure difference feature set, extract the functional group type, absorbance value and half width at half maximum (WHM) parameter corresponding to the spectral segment, sort out the combination relationship between functional group markers and spectral shape parameters in the same spectral segment, classify the combination according to the change trend, and generate spectral segment structure combination trend. S402: Call the spectral segment structure combination trend value, number the functional group type and spectral shape parameter combination index position of the spectral segment, establish a combination index sequence according to the spectral segment arrangement order, and number the continuous structure combination to generate the structure sequence archive index value. S403: Based on the structural sequence archive index value, match the category label content recorded in the original material sample, establish a correspondence table between the structural sequence and the corresponding label, integrate the correspondence table and the spectral segment combination index sequence information, and generate a spectral input feature sequence archive.

[0010] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Call the spectral input feature sequence file, extract the functional group identifiers and response features corresponding to the spectral structure sequence in the material sample, construct the functional group change path by combining the distribution position of the spectral parameters, and generate the functional group change trend coefficient based on the continuity of the response change direction and position. S502: Based on the functional group change trend coefficient, compare the functional group composition sequence and spectral response feature interval defined in the quality standard, identify the difference index that is inconsistent with the standard range, screen the response feature offset items, and generate the functional group matching deviation amount. S503: Based on the functional group matching deviation, combined with the functional group arrangement integrity and spectral structure continuity standards, determine the structural composition state in the sample, record the corresponding response anomaly information and structural stability index of the sample, and generate a set of quality test results for new chemical materials.

[0011] As a further embodiment of the present invention, the polyolefin copolymer is a high molecular weight compound generated by polymerization reaction of olefin monomers and one or more comonomers, including ethylene-propylene copolymers, ethylene-butene copolymers and ethylene-hexene copolymers, and has the characteristics of adjustable crystallinity and high chemical stability. The infrared band range refers to the electromagnetic wave wavelength range used in spectral detection, which is between 2.5 micrometers and 25 micrometers, and is used to measure the molecular vibrational absorption characteristics of new chemical materials under different light energies. The peak intensity refers to the maximum absorbance value at the wavelength corresponding to the absorption peak in the spectrum, and the parameter is obtained directly by a spectrophotometer.

[0012] As a further aspect of the present invention, the standard absorption range is the typical absorption wavelength range of functional groups obtained by statistical analysis based on publicly available infrared or ultraviolet-visible spectral databases, including the stretching vibration absorption region of aromatic carbon-carbon double bonds, which is located in the range of 1,400 to 1,600 wavenumbers. The half-width at half maximum (WHM) variation refers to the change in the width value of the spectral peak at half the maximum absorbance. The structural change segment refers to the spectral region generated between adjacent spectral segments due to changes in the functional group affiliation; The spectral parameters are spectral indices used to describe the morphological characteristics of absorption peaks, including absorbance, full width at half maximum (FWHM), peak position shift, and peak symmetry factor. The quality standard is a quality judgment criterion established based on the structural integrity, compositional uniformity, and spectral distribution of new chemical materials, and a structural comparison template is formed based on standard samples or industry testing specifications. A training dataset is constructed based on sample labels, and a convolutional neural network model is trained to realize the predictive relationship between spectral input feature sequences and quality levels; the model is then deployed in the detection system for new sample quality prediction.

[0013] An artificial intelligence-based quality inspection system for new chemical materials includes: The spectral acquisition and analysis module acquires the absorption spectrum data of polyolefin copolymer materials in the infrared band, extracts the absorption peak positions, absorbance values ​​and half width at half maximum (WHM) parameters of the spectral bands, organizes all spectral band information in wavelength order, and generates a spectral response characteristic table after normalizing the absorbance data. The response feature extraction module calls the spectral response feature table to extract the peak position and bandwidth information of the spectral band, matches it with the wavelength range and the standard functional group stretching vibration range, establishes a correspondence between the matching results and the functional group type, and generates a functional group segment annotation information set. The functional group identification module calls the functional group segment annotation information set to identify the location of changes in functional group type between adjacent spectral segments, extracts the absorbance difference and full width at half maximum (FWHM) changes in the changed segments, determines the location and type of structural change based on the change characteristics of the absorption response parameters, and generates a spectral segment structural difference feature set. The structural difference extraction module calls the spectral segment structural difference feature set, organizes the functional group type, spectral shape parameters and response change trend corresponding to the spectral segment, establishes the mapping relationship between the spectral segment sequence and the material sample category label, and generates a spectral input feature sequence file; The quality status judgment module calls the spectral input feature sequence file, and judges the structural integrity and functional group composition of the sample based on the distribution of functional groups and response change patterns in the spectral structure sequence, generating a set of quality test results for new chemical materials. The artificial intelligence reasoning module calls the trained AI recognition model, inputs the spectral input feature sequence file, and outputs the quality grade prediction, structural integrity score and abnormal risk warning corresponding to the material sample, thereby realizing the intelligent discrimination function of the detection system.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, functional group identification is achieved by normalizing and extracting features from spectral data and combining standard interval matching. Structural difference information is extracted based on absorbance and peak shape changes. A mapping relationship between structural sequences and sample labels is established to achieve systematic matching of structural features and quality standards. A response pattern comparison method is introduced during the detection process to automatically judge structural integrity and functional group composition, thereby improving the closed-loop capability and recognition accuracy of the detection process and enhancing adaptability and detection consistency for complex materials. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention; Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same. In this embodiment of the invention, to enhance the detection system's generalization ability and discrimination efficiency for complex samples, a recognition model based on artificial intelligence algorithms is further constructed. The extracted spectral input feature sequence is used as the input vector of the neural network, and the output is the functional group composition, structural integrity, and quality grade label of the material. Supervised learning is used for training.

[0021] The model structure can be selected as follows: Convolutional Neural Networks (CNNs): Used to automatically extract local peak shape variation features from spectral signals; Transformer structure: suitable for capturing long-range dependencies between spectral segments and improving the ability to express complex functional group structures; Multilayer perceptrons (MLPs) or ensemble models (such as LightGBM) are used for the final classification and regression tasks.

[0022] During training, the model is based on labeled sample data, optimized using the cross-entropy loss function, and continuously updated with weight parameters through the backpropagation algorithm, ultimately obtaining an AI model that can perform inference on new samples.

[0023] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0024] Please see Figure 1 This invention provides a method for quality testing of new chemical materials based on artificial intelligence, comprising the following steps: S1: Obtain the absorption spectrum data of the polyolefin copolymer material to be tested in the infrared band, extract the absorption peak position, peak intensity and bandwidth parameters of the spectral band, organize the spectral band data according to the wavelength order, establish a spectral band index list, and normalize the absorbance data to generate a spectral response characteristic table. Polyolefin copolymers are high molecular weight compounds formed by the polymerization reaction of olefin monomers and one or more comonomers. Common types include ethylene-propylene copolymers, ethylene-butene copolymers, and ethylene-hexene copolymers. They are characterized by adjustable crystallinity and high chemical stability and are used as test objects to characterize new chemical materials samples. The infrared band refers to the range of electromagnetic wavelengths used in spectral detection, which is between 2.5 micrometers and 25 micrometers (i.e., the wavenumber range of 4000 cm⁻¹ to 400 cm⁻¹). It is used to measure the molecular vibrational absorption characteristics of new chemical materials under different light energies. Peak intensity refers to the maximum absorbance value at the wavelength corresponding to the absorption peak in the spectrum. It is used to reflect the material's ability to absorb incident light, and the parameter can be obtained directly by a spectrophotometer. Normalization refers to transforming absorbance data from different spectral bands to a uniform numerical range through linear mapping, in order to maintain the consistency and comparability of spectral signals. S2: Based on the spectral response feature table, extract the peak position and bandwidth information of all spectral segments, and combine the standard absorption ranges of the stretching vibration ranges of aromatic carbon-carbon double bonds, hydroxyl groups, and carbon-hydrogen single bonds to perform interval matching of the spectral segments, determine the functional group type, and generate a set of functional group segment annotation information. The standard absorption range is the range of typical absorption wavelengths of functional groups obtained from publicly available infrared or ultraviolet-visible spectral databases. This includes the stretching vibration absorption region of aromatic carbon-carbon double bonds, which is located in the range of 1,400 to 1,600 wavenumbers. It is used to determine the type of functional group in the spectral band. S3: Call the functional group segment annotation information set, identify the functional group change position between adjacent spectral segments, extract the absorbance difference and half-width variation of the corresponding spectral segments, identify the structural change segments and classify the change type based on the absorbance change trend and peak shape change characteristics, and generate a spectral segment structural difference feature set. The half-width at half-maximum (WHM) variation refers to the change in the width of a spectral peak at half its maximum absorbance, used to describe the broadening and morphological differences of the absorption peak. Structural variation segments refer to spectral regions that arise between adjacent spectral segments due to changes in the functional group affiliation, and are used to reflect differences in the distribution or composition of functional groups in new chemical materials. S4: Based on the spectral segment structure difference feature set, organize the functional group types, spectral parameters and variation trends corresponding to the spectral segments, establish a structural sequence archive index, and establish a mapping relationship based on the category labels of the original material samples to generate spectral input feature sequence archives; Spectral shape parameters are spectral indices used to describe the morphological characteristics of absorption peaks, including absorbance, full width at half maximum (FWHM), peak position shift, and peak symmetry factor, which are used to characterize the shape and distribution of spectral bands. S5: Call the spectral input feature sequence file, and perform matching and identification between structural features and quality standards based on the functional group distribution and response change pattern in the spectral structural sequence. Judge the structural integrity and functional group composition state of the material sample, and generate a set of quality test results for new chemical materials. S6: Based on the spectral input feature sequence archive and the label data of existing samples, train an artificial intelligence model (including deep learning networks such as convolutional neural networks CNN and Transformer) to achieve automatic identification and prediction from spectral structure to functional group composition and quality status. Deploy the trained model in the detection system, receive the spectral sequence input of new samples, and output the structural integrity and quality level judgment results. The quality standard is a quality judgment criterion established based on the structural integrity, compositional uniformity, and spectral distribution of new chemical materials. It can form a structural comparison template based on standard samples or industry testing specifications. A training dataset is constructed based on sample labels, and a convolutional neural network model is trained to realize the predictive relationship between spectral input feature sequences and quality levels; the model is then deployed in the detection system for new sample quality prediction.

[0025] The spectral response characteristic table includes spectral band index, peak position parameter, peak intensity parameter, bandwidth parameter, and normalized absorbance. The functional group segment annotation information set includes matching interval identifier, functional group type label, standard range number, and segment annotation. The spectral band structural difference characteristic set includes functional group change position, absorbance difference, half-width at half-maximum change, and change type classification. The spectral input characteristic sequence file includes structural sequence number, functional group distribution sequence, spectral shape parameter vector, change trend index, and sample category mapping table. The chemical new material quality testing result set includes structural integrity judgment, composition status assessment, quality standard matching conclusion, and sample qualification conclusion.

[0026] Please see Figure 2 The specific steps of S1 are as follows: S101: Acquire the absorption spectrum data of the polyolefin copolymer material to be tested in the infrared band, record the correspondence between light intensity and wavelength, detect the change of absorbance at the wavelength, calculate the absorption value based on the relationship between light intensity and absorbance, organize and form the spectral curve data of continuous bands, and generate an absorption spectrum dataset. To obtain the infrared absorption spectrum data of polyolefin copolymer materials, a Fourier transform infrared spectrometer is required. The sample, in pellet or liquid film form, is placed in the instrument's optical path. The wavelength range is scanned from approximately 4000 cm⁻¹ to 400 cm⁻¹ to collect data on the material's absorption of infrared light at different wavelengths. During the scan, the instrument records the transmitted light intensity at each wavelength and, combined with a reference background scan, obtains the actual transmittance value. By comparing the ratio between the transmitted and reference light intensities, the material's absorption capacity is indirectly calculated. For example, when the transmitted light intensity at a certain wavelength is 55%, and the reference intensity is 100%, the absorbance can be understood as the material's absorption ratio of infrared radiation at that wavelength being 45%. This process is repeated to collect hundreds or thousands of data sets across the entire wavelength range. Spectral curves are then plotted using programming software such as Origin or Python. Connecting wavelength and absorbance points creates a continuous graph, forming an absorption spectrum dataset reflecting the material's infrared response.

[0027] S102: Based on the absorption spectrum dataset, identify the position of the maximum absorption peak in the band, calculate the peak wavelength and corresponding peak intensity, determine the bandwidth parameter according to the peak width range, arrange the peak parameters in wavelength order, and generate an absorption peak parameter sequence. To identify peaks in existing absorption spectrum datasets, it is necessary to sequentially check whether all absorbance points are higher than their adjacent points and meet a set wavelength interval threshold, which is set to 2 to 5 cm⁻¹, to avoid misjudging peak positions due to slight perturbations. For example, if the absorbance of a point in the data is 0.83, and its left and right neighboring points are 0.75 and 0.72 respectively, with an interval of 5 cm⁻¹, it is determined to be a local peak point, and its wavelength is recorded as 2920 cm⁻¹. Subsequently, the absorbance decreasing trend in the surrounding range of this peak, such as ±15 cm⁻¹, is analyzed. The absorption peak bandwidth is set as the wavelength range where the peak height decreases to 50%. For example, if the peak height is 0.80, then the wavelength point corresponding to 0.40 is found. For example, the points where the absorbance decreases to 0.40 on the left and right are 2900 and 2940 cm⁻¹ respectively, so the bandwidth is 40 cm⁻¹. All peak wavelengths, absorbance, and bandwidth data are recorded one by one and arranged in ascending order of wavelength to generate a parameter sequence, such as including multiple bands such as 2920, 2850, and 1460. Each parameter entry contains three data points: peak position, peak intensity, and bandwidth, which are used for subsequent normalization and feature extraction.

[0028] S103: Based on the absorption peak parameter sequence, extract the band absorbance data, normalize the absorbance, adjust the peak intensity value to a uniform standard range, establish an index list including peak position, peak intensity, bandwidth and normalized absorbance, integrate it into spectral response data, and generate a spectral response characteristic table. To normalize the absorbance data, the minimum and maximum values ​​of all peak intensities in the entire parameter sequence are first calculated. The standard normalization range is set to 0 to 1, and linear scaling is used to convert the original values ​​to the standard range. For example, if the minimum peak intensity is 0.25 and the maximum is 1.10, and a peak intensity is 0.70, the normalized value is (0.70-0.25) divided by (1.10-0.25), resulting in 0.53. The normalized intensity of all peaks is calculated using the same method to ensure the values ​​are on a uniform scale. The bandwidth is recorded directly as the actual value, between 10 and 80 cm⁻¹, without scaling. The normalized absorbance data is supplemented by the normalized intensity values, and combined with peak position and bandwidth, a response data item containing four indicators is constructed. All parameters are integrated to generate a table, with each row recording a set of peak position, intensity, bandwidth, and normalized absorbance, fully reflecting the sample's response characteristics to infrared light, facilitating spectral comparison and analysis between different materials. For example, in the LDPE and HDPE copolymer, response data with peak positions at 2916 cm⁻¹, 1464 cm⁻¹ and 719 cm⁻¹ appeared, and after normalization, they showed intensity values ​​of 0.78, 0.52 and 0.34, respectively. The differences in characteristic curves of different material samples were obvious.

[0029] Please see Figure 3 The specific steps of S2 are as follows: S201: Based on the spectral response feature table, extract the peak wavelength and bandwidth information in the spectral band, record the correspondence between peak position and bandwidth under the spectral band, organize the spectral band structure information according to wavelength order, and generate a peak bandwidth parameter list. After generating the spectral response feature table, it is necessary to extract the wavelength and bandwidth information of each absorption peak. First, select the wavelength and corresponding wavelength range of each absorption peak from the feature table to determine the center wavelength of the peak. Bandwidth refers to the wavelength range from the peak value of the absorption peak until the absorbance decreases to a certain percentage of the peak value; bandwidth is defined as the absorbance decreasing to 50%. For each spectral band, organize the peak position and bandwidth data of each absorption peak in ascending order of wavelength. For example, if an absorption peak is located at 2920 cm⁻¹ and has a bandwidth of 40 cm⁻¹, this data item will be recorded. Repeat this process for each sample to compile information for all bands, forming a parameter list of peak position and bandwidth. This organized data provides the foundation for subsequent functional group identification and functional region matching.

[0030] S202: Based on the peak bandwidth parameter list, call the standard absorption wavelength range of the aromatic carbon-carbon double bond stretching vibration range, the hydroxyl group stretching vibration range and the carbon-hydrogen single bond stretching vibration range, determine whether the peak position falls into the standard range, obtain the correspondence between the spectrum and the functional range, and generate the functional range matching judgment result. Based on the peak bandwidth parameter list, it is necessary to determine whether each peak falls within these known functional ranges by comparing the standard absorption wavelength ranges of aromatic carbon-carbon double bond stretching vibrations, hydroxyl group stretching vibrations, and hydrocarbon single bond stretching vibrations. The stretching vibration of aromatic carbon-carbon double bonds is around 1600 cm⁻¹, hydroxyl group stretching vibrations appear in the 3200 to 3600 cm⁻¹ range, and hydrocarbon single bond stretching vibrations are generally between 2800 and 3000 cm⁻¹. Based on these known wavelength ranges, it is determined whether each peak conforms to these standard ranges. For example, if a peak appears at 2850 cm⁻¹ with a bandwidth of 30 cm⁻¹, it is necessary to determine whether this peak belongs to the hydrocarbon single bond stretching vibration range. If the wavelength is within the standard range, it is determined to be a characteristic band of hydrocarbon single bond stretching vibrations. If it is not within the standard range, it may involve characteristics of other functional groups, requiring further analysis. In this way, the spectral data of the sample can be compared with the functional range to obtain the matching status of each spectral band and the functional range, and generate the functional range matching judgment result.

[0031] S203: Based on the functional interval matching determination results, identify the wavelength, bandwidth and functional group type of the spectral segment falling into the interval, construct the matching relationship between functional group category and spectral segment position, integrate the functional group interval information of each spectral segment, and generate a functional group segment annotation information set. Based on the functional range matching results, spectral segments falling within the standard range need to be identified, and their wavelength, bandwidth, and functional group type recorded. For each spectral segment conforming to the standard range, its wavelength and bandwidth information are labeled, and its functional group type is specified. For example, if a peak position is at 2920 cm⁻¹ and conforms to the standard range of C-H single bond stretching vibration, then this spectral segment is labeled as C-H single bond stretching vibration, and its bandwidth is recorded as 40 cm⁻¹. All spectral segments that meet the conditions are labeled with their functional group types one by one, and these data are integrated to form a functional group segment labeling information set. For example, if the sample has absorption peaks at 1800 cm⁻¹ (possibly carbonyl vibration), 2920 cm⁻¹ (possibly CH stretching vibration), and 1250 cm⁻¹ (possibly C–O stretching vibration), then these bands and their corresponding functional group types (such as carbonyl, CH, C–O) are integrated to generate a complete functional group segment labeling information set for subsequent material composition analysis.

[0032] Please see Figure 4 The specific steps of S3 are as follows: S301: Call the functional group segment annotation information set, compare the functional group types in adjacent spectral segments in turn, identify the positions where the functional group types change, record the spectral segment index information corresponding to the changes, and generate a functional group transformation position set; When using functional group segment annotation information sets, it is first necessary to compare the functional group types of adjacent spectral segments to identify the locations where functional group types change. During execution, the functional group types of adjacent spectral segments are compared one by one from the already annotated functional group segment information. If the functional group type of an adjacent spectral segment changes, it is determined that a functional group transformation has occurred. The index information of the spectral segment at the location of the change is recorded to ensure accurate identification of the spectral segment where the transformation occurred. For example, if a spectral segment is located at 1600 cm⁻¹ and annotated as a characteristic band of an aromatic carbon-carbon double bond, and the immediately following spectral segment is located at 1650 cm⁻¹ and annotated as a characteristic band of a carbonyl functional group, then a functional group transformation has occurred at that location. This method can accurately determine the location of functional group transformation and record the index information of the changed spectral segment, providing data support for subsequent analysis.

[0033] S302: Based on the functional group transformation position set, extract the absorbance difference and half-width variation of adjacent spectral bands, determine whether the dual conditions of absorption difference and peak shape variation are met, mark the spectral band positions that meet the conditions, and generate spectral band response change characteristic values. Based on the functional group transformation position set, it is necessary to analyze the absorbance difference and full width at half maximum (FWHM) variation of adjacent spectral bands. First, the absorbance difference between adjacent spectral bands is calculated, i.e., the absorbance values ​​of two bands are compared. For example, if the absorbance of one band is 0.8 and the absorbance of the next band is 0.6, the absorbance difference is 0.2. Next, the FWHM variation is calculated, which is obtained by measuring the wavelength range corresponding to the drop in absorption peak to its half-maximum. If the FWHM of a peak changes from 40 cm⁻¹ to 50 cm⁻¹, the FWHM variation is 10 cm⁻¹. Then, it is necessary to determine whether the absorbance difference and FWHM variation meet the set criteria. For example, if the absorbance difference is greater than 0.1 and the FWHM variation exceeds 10 cm⁻¹, it is determined that the spectral band has a significant response change. The positions of spectral bands that meet these conditions are marked and used as characteristic values ​​of the spectral band response change. These characteristic values ​​help identify the characteristics of the spectral band changes, providing a basis for subsequent analysis.

[0034] S303: Based on the characteristic values ​​of spectral response changes, combine the direction of absorbance difference and the trend of half-width at half-maximum to make a judgment, verify the structural change type corresponding to the spectral segment, organize the matching relationship between structural changes and spectral segment positions, and generate a spectral segment structural difference feature set. Based on the characteristic values ​​of spectral response changes, it is necessary to combine and judge the direction of absorbance difference and the trend of half-width at half-maximum (HWHM) changes to verify the type of structural change corresponding to the spectral segment. The direction of absorbance difference can be either increasing or decreasing, while the trend of HWHM changes can be either widening or narrowing. By comparing the trends of absorbance and HWHM changes, it can be determined whether the spectral segment has undergone structural changes. For example, when the absorbance difference increases and the HWHM widens, it may indicate a change in molecular structure, such as aggregation or molecular rearrangement. If the absorbance difference decreases and the HWHM narrows, it may indicate that the structure is becoming more stable or that the interactions between molecules are weakening. Combining these changes, the type of structural change can be further analyzed, and the type and location of the structural change corresponding to the spectral segment can be compiled. Finally, this information is summarized into a spectral segment structural difference feature set for analyzing changes in material properties and molecular structure.

[0035] Please see Figure 5 The specific steps of S4 are as follows: S401: Based on the spectral segment structure difference feature set, extract the functional group type, absorbance value and half width at half maximum (WHM) parameter corresponding to the spectral segment, sort out the combination relationship between functional group label and spectral shape parameter in the same spectral segment, classify the combination according to the change trend, and generate spectral segment structure combination trend. Based on the spectral segment structural difference feature set, it is necessary to extract the functional group type, absorbance value, and full width at half maximum (FWHM) parameter for each spectral segment, and to organize the combination relationship between functional group labels and spectral shape parameters within the same spectral segment. During execution, firstly, the functional group type for each spectral segment is extracted from the spectral segment structural difference feature set. Possible functional group types include carboxyl, ether, and ester groups. Simultaneously, the absorbance value and FWHM parameter for each spectral segment are obtained. The absorbance value is directly obtained from the spectrometer, representing the light absorption intensity at that wavelength, and is dimensionless. The FWHM is the width of the spectral peak, measured in wavenumbers (cm⁻¹), reflecting the sharpness of the absorption peak. Next, the functional group type and spectral shape parameters need to be combined to form the combination relationship for that spectral segment. For example, a certain spectral segment may simultaneously exhibit characteristic bands of carboxyl and ester functional groups, with an absorbance of 0.75 and a FWHM of 40 cm⁻¹. During processing, these combinations need to be categorized according to their changing trends. When categorizing, trends in absorbance (e.g., absorbance gradually increasing or decreasing) and half-width at half-maximum (FWHM) changes (e.g., FWHM widening or narrowing) can be considered. This method generates the spectral structure combination trend for each spectral band, revealing the variations in functional groups and spectral shape parameters across different spectral bands.

[0036] S402: Call the spectral segment structure combination trend value, number the functional group type and spectral shape parameter combination index position of the spectral segment, establish a combination index sequence according to the spectral segment arrangement order, and number the continuous structure combination to generate the structure sequence archive index value. When calling the spectral segment structure combination trend value, it is necessary to number the index position of the functional group type and spectral shape parameter combination of each segment. First, based on the generated spectral segment structure combination trend value, record the functional group type and its corresponding spectral shape parameter of each segment. For example, a segment may exhibit the characteristics of ester and carboxyl groups, with an absorbance of 0.85 and a full width at half maximum (FWHM) of 35 cm⁻¹. The functional group type and spectral shape parameter of this segment will form a combination, which will be assigned a number in the index. All segments are numbered in the order of arrangement, establishing an index sequence for spectral segment combinations. For example, segment 1 corresponds to the combination of ester and carboxyl groups, and segment 2 may correspond to the combination of ether and hydroxyl groups. Next, for consecutive structural combinations, numbering is required. For example, consecutive spectral segment combinations can be marked as sequence numbers 1, 2, 3, etc., forming a structural sequence. The numbering of this structural sequence will reflect the changes in structural combinations between different segments, generating a structural sequence archive index value to ensure that the structural changes of each spectral segment combination have a clear number and sequence.

[0037] S403: Based on the structural sequence archive index value, match the category label content recorded in the original material sample, establish a correspondence table between structural sequences and corresponding labels, integrate the correspondence table with the spectral segment combination index sequence information, and generate a spectral input feature sequence archive. Based on the archived index values ​​of the structural sequences, they need to be matched with the category labels recorded in the original material samples. First, the archived index values ​​are matched against the category labels in the original material samples to ensure that the changes in each spectral segment combination correspond to the corresponding label. For example, assuming that the structural change corresponding to a certain spectral segment combination is formed by a specific molecule, then the structural sequence of that spectral segment should match the category label of that molecule recorded in the original sample. At this point, a mapping table between structural sequences and category labels is created, listing each structural sequence and its corresponding label. Based on this, the archived index values ​​of the structural sequences and the index sequence information of the spectral segment combinations need to be integrated to generate a spectral input feature sequence archive containing complete information. For example, structural sequence 1 may match the label "sample A," while structural sequence 2 may match the label "sample B." In this way, the structural change information of all spectral segments is integrated with the corresponding label information to generate a complete spectral input feature sequence archive, facilitating subsequent data analysis and processing.

[0038] Please see Figure 6 The specific steps of S5 are as follows: S501: Call the spectral input feature sequence file, extract the functional group identifiers and response features corresponding to the spectral structure sequence in the material sample, construct the functional group change path by combining the distribution position of spectral parameters, and generate the functional group change trend coefficient based on the continuity of response change direction and position. When calling the spectral input feature sequence file, it is first necessary to extract the functional group identifiers and response characteristics corresponding to the spectral segment structure sequences in the material sample, and then construct the functional group change path by combining the distribution position of the spectral parameters. Specifically, the process involves first extracting the spectral segment structure sequence for each sample from the spectral input feature sequence file. These spectral segment structure sequences contain absorbance data at different wavelengths within each spectral segment. Next, by identifying the absorbance values ​​of these spectral segments and combining them with the functional group identifiers of each segment, the response characteristics of each functional group in each spectral segment can be determined. For example, in infrared spectroscopy analysis, if the absorbance peak of a spectral segment appears at a specific wavelength (such as around 1700 cm⁻¹), this is associated with a carbonyl functional group. By analyzing the distribution position of the spectral segment response characteristics, the positional changes of each functional group within the spectral segment can be clearly identified. Subsequently, using the continuity of the direction and position of the response characteristic changes in the spectral segment structure, a functional group change trend coefficient is generated. This coefficient reflects the changes of functional groups in different spectral segments and is quantified based on the trend of functional group changes within the spectral segments. For example, if the absorbance value of a functional group increases linearly with wavelength, its change path will be marked as an upward trend in the spectrum and assigned a corresponding trend coefficient. By combining the directionality and positional continuity of the response characteristics, the change path of each functional group in the material sample can be determined, and its trend coefficient can be calculated, thus obtaining the complete functional group change trend.

[0039] S502: Based on the functional group change trend coefficient, compare the functional group composition sequence and spectral response characteristic interval defined in the quality standard, identify the difference index that is inconsistent with the standard range, screen the response characteristic offset item, and generate the functional group matching deviation. After obtaining the functional group variation trend coefficient, it is necessary to identify discrepancies with the standard range based on the functional group composition sequence and spectral response characteristic range defined in the quality standard. First, according to the functional group composition sequence set in the quality standard, for example, the spectral analysis requirements for a certain material may include the frequency of occurrence and response characteristic range of specific functional groups, such as the response characteristics of alcohol, ester, and ether functional groups should remain consistent within a certain wavelength range. By comparing this with the functional group variation trend in the spectral input feature file, it is possible to identify situations where the response characteristics of certain spectral bands deviate from the standard range in a specific wavelength band. For example, if the absorbance value of a certain spectral band deviates from the specified standard absorbance value within a specific wavelength range (e.g., the absorbance should be 0.8 ± 0.05, but the actual value is 0.75 or 0.85), it can be identified as a response characteristic offset. In this way, response characteristic offsets that do not meet the quality standard can be screened out. Next, these offsets are further quantified by generating a functional group matching deviation. For example, if the functional group characteristics of a certain spectral band change, the deviation can be obtained by calculating the difference between the actual response eigenvalue and the standard response eigenvalue of that spectral band. Assuming the standard value is 0.8 and the actual measured value is 0.75, the deviation is |0.8 - 0.75| = 0.05. By calculating the deviations for all spectral bands, a set of deviations containing all offset terms is obtained for further analysis.

[0040] S503: Based on the functional group matching deviation, combined with the functional group arrangement integrity and spectral structure continuity standards, the structural composition state in the sample is judged, the corresponding response anomaly information and structural stability index of the sample are recorded, and a set of quality test results for new chemical materials is generated. Based on the functional group matching deviation, it is necessary to combine the integrity of the functional group arrangement with the continuity standard of the spectral structure to determine the structural composition of the sample. First, in spectral data, the integrity of the functional group arrangement requires that the spectral segments of the sample exhibit a certain functional group order and structural consistency. For example, in the spectral analysis of some polymers, specific functional groups may be required to be arranged in a certain order; otherwise, it may indicate an abnormality in the sample's synthesis process. Based on this, the structural continuity of the spectral segments in the sample needs to be evaluated to determine whether there are discontinuities or abrupt changes between the segments. For example, if two adjacent spectral segments show significant differences or positional jumps in their functional group identifiers, this may indicate structural instability or problems with the material. In this way, abnormal response information of the sample can be recorded, and its structural stability index can be calculated, reflecting the quality of the sample. If there are abnormalities or discontinuities in the structural arrangement of the sample, the generated structural stability index will be lower than the set standard threshold, thus indicating potential quality problems in the sample. For example, if the response characteristics of some spectral segments deviate significantly from the standard spectral response curve, and this trend is not defined in the quality standard, then the sample can be considered to have an abnormal response. Finally, by combining this information, a quality test result set for the new chemical material is generated to assess whether the new material meets predetermined quality standards. In practical applications, this analytical method can provide a basis for the production and quality control of chemical materials, ensuring the structural and performance stability of the product.

[0041] Please see Figure 7 A new chemical material quality testing system based on artificial intelligence includes: The spectral acquisition and analysis module acquires the absorption spectrum data of polyolefin copolymer materials in the infrared band, extracts the absorption peak positions, absorbance values ​​and half width at half maximum (WHM) parameters of the spectral bands, organizes all spectral band information in wavelength order, and generates a spectral response characteristic table after normalizing the absorbance data. The response feature extraction module calls the spectral response feature table to extract the peak position and bandwidth information of the spectral band. It matches the wavelength range with the standard functional group stretching vibration range, establishes a correspondence between the matching results and the functional group type, and generates a functional group segment annotation information set. The functional group identification module calls the functional group segment annotation information set to identify the location of changes in functional group type between adjacent spectral segments, extracts the absorbance difference and full width at half maximum (FWHM) changes in the changed segments, determines the location and type of structural change based on the change characteristics of the absorption response parameters, and generates a spectral segment structural difference feature set. The structural difference extraction module calls the spectral segment structural difference feature set, organizes the functional group types, spectral parameters and response change trends corresponding to the spectral segments, establishes the mapping relationship between the spectral segment sequence and the material sample category label, and generates a spectral input feature sequence file; The quality status judgment module calls the spectral input feature sequence file, and judges the structural integrity and functional group composition of the sample based on the distribution of functional groups and response change patterns in the spectral structure sequence, generating a set of quality test results for new chemical materials. The artificial intelligence reasoning module calls the trained AI recognition model, inputs the spectral input feature sequence file, and outputs the corresponding quality grade prediction, structural integrity score and abnormal risk warning of the material sample, so as to realize the intelligent discrimination function of the detection system.

[0042] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for quality testing of new chemical materials based on artificial intelligence, characterized in that, Includes the following steps: S1: Collect absorption spectrum data of polyolefin copolymers in the infrared band, extract the wavelength, peak intensity and bandwidth of absorption peaks, sort the spectral information by wavelength, establish a spectral index and generate a spectral response characteristic table. S2: Based on the spectral response characteristic table, extract the peak position and bandwidth of the spectral bands, and combine the standard absorption ranges of aromatic C=C, hydroxyl O–H and hydrocarbon C–H bonds to complete the spectral band matching and functional group assignment, and generate a functional group segment labeling information set. S3: Call the functional group segment annotation information set, identify functional group changes between spectral segments, extract absorbance differences and full width at half maximum (FWHM) changes, combine peak shape trend to identify structural change segments, classify change types, and generate a spectral segment structural difference feature set. S4: Based on the spectral segment structure difference feature set, summarize the functional group type, spectral shape parameters and change trends, construct the spectral segment structure sequence index, and map it with the material category label to generate a spectral input feature sequence file; S5: Based on the spectral input feature sequence file, combined with the functional group distribution and response mode, the structural features and quality standards are matched to determine the structural integrity and functional group composition of the material, and output the quality test result set of the new chemical material. S6: Based on the spectral input feature sequence archive and the label data of existing samples, train an artificial intelligence model to realize automatic identification and prediction from spectral structure to functional group composition and quality status. Deploy the trained model in the detection system, receive the spectral sequence input of new samples, and output the structural integrity and quality level judgment results.

2. The method for quality testing of new chemical materials based on artificial intelligence according to claim 1, characterized in that, The spectral response feature table includes spectral band index, peak position parameter, peak intensity parameter, bandwidth parameter, and normalized absorbance. The functional group segment annotation information set includes matching interval identifier, functional group type label, standard range number, and segment annotation. The spectral band structural difference feature set includes functional group change position, absorbance difference, half-width at half-maximum change, and change type classification. The spectral input feature sequence file includes structural sequence number, functional group distribution sequence, spectral shape parameter vector, change trend index, and sample category mapping table. The chemical new material quality testing result set includes structural integrity judgment, composition status assessment, quality standard matching conclusion, and sample qualification conclusion.

3. The method for quality testing of new chemical materials based on artificial intelligence according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Acquire the absorption spectrum data of the polyolefin copolymer material to be tested in the infrared band, record the correspondence between light intensity and wavelength, detect the change of absorbance at the wavelength, calculate the absorption value based on the relationship between light intensity and absorbance, organize and form the spectral curve data of continuous bands, and generate an absorption spectrum dataset. S102: Based on the absorption spectrum dataset, identify the position of the maximum absorption peak in the band, calculate the peak wavelength and corresponding peak intensity, determine the bandwidth parameter according to the peak width range, arrange the peak parameters in wavelength order, and generate an absorption peak parameter sequence. S103: Based on the absorption peak parameter sequence, extract the band absorbance data, normalize the absorbance, adjust the peak intensity value to a uniform standard range, establish an index list including peak position, peak intensity, bandwidth and normalized absorbance, integrate it into spectral response data, and generate a spectral response feature table.

4. The method for quality testing of new chemical materials based on artificial intelligence according to claim 3, characterized in that, The specific steps of S2 are as follows: S201: Based on the spectral response feature table, extract the peak wavelength and bandwidth information in the spectral band, record the correspondence between peak position and bandwidth in the spectral band, organize the spectral band structure information according to wavelength order, and generate a peak bandwidth parameter list. S202: Based on the peak bandwidth parameter list, call the standard absorption wavelength range of the aromatic carbon-carbon double bond stretching vibration range, the hydroxyl group stretching vibration range and the carbon-hydrogen single bond stretching vibration range, determine whether the peak position falls into the standard range, obtain the correspondence between the spectral segment and the functional range, and generate the functional range matching judgment result. S203: Based on the functional interval matching determination result, identify the wavelength, bandwidth and functional group type of the spectral segment falling into the interval, construct the matching relationship between functional group category and spectral segment position, integrate the functional group interval information of each spectral segment, and generate a functional group segment annotation information set.

5. The method for quality testing of new chemical materials based on artificial intelligence according to claim 4, characterized in that, The specific steps for S3 are as follows: S301: Call the functional group segment annotation information set, compare the functional group types in adjacent spectral segments in turn, identify the positions where the functional group types change, record the spectral segment index information corresponding to the changes, and generate a functional group transformation position set; S302: Based on the functional group transformation position set, extract the absorbance difference and half-width variation of adjacent spectral bands, determine whether the dual conditions of absorption difference and peak shape variation are met, mark the spectral band positions that meet the conditions, and generate spectral band response change characteristic values. S303: Based on the characteristic values ​​of the spectral band response change, combine and judge the direction of absorbance difference and the trend of half width at half maximum (WHM) change, verify the structural change type corresponding to the spectral band, organize the matching relationship between structural change and spectral band position, and generate a spectral band structural difference feature set.

6. The method for quality testing of new chemical materials based on artificial intelligence according to claim 5, characterized in that, The specific steps of S4 are as follows: S401: Based on the spectral segment structure difference feature set, extract the functional group type, absorbance value and half width at half maximum (WHM) parameter corresponding to the spectral segment, sort out the combination relationship between functional group markers and spectral shape parameters in the same spectral segment, classify the combination according to the change trend, and generate spectral segment structure combination trend. S402: Call the spectral segment structure combination trend value, number the functional group type and spectral shape parameter combination index position of the spectral segment, establish a combination index sequence according to the spectral segment arrangement order, and number the continuous structure combination to generate the structure sequence archive index value. S403: Based on the structural sequence archive index value, match the category label content recorded in the original material sample, establish a correspondence table between the structural sequence and the corresponding label, integrate the correspondence table and the spectral segment combination index sequence information, and generate a spectral input feature sequence archive.

7. The method for quality testing of new chemical materials based on artificial intelligence according to claim 6, characterized in that, The specific steps of S5 are as follows: S501: Call the spectral input feature sequence file, extract the functional group identifiers and response features corresponding to the spectral structure sequence in the material sample, construct the functional group change path by combining the distribution position of the spectral parameters, and generate the functional group change trend coefficient based on the continuity of the response change direction and position. S502: Based on the functional group change trend coefficient, compare the functional group composition sequence and spectral response feature interval defined in the quality standard, identify the difference index that is inconsistent with the standard range, screen the response feature offset items, and generate the functional group matching deviation amount. S503: Based on the functional group matching deviation, combined with the functional group arrangement integrity and spectral structure continuity standards, determine the structural composition state in the sample, record the corresponding response anomaly information and structural stability index of the sample, and generate a set of quality test results for new chemical materials.

8. The method for quality testing of new chemical materials based on artificial intelligence according to claim 1, characterized in that, The polyolefin copolymers are high molecular weight compounds generated by polymerization of olefin monomers and one or more comonomers. The types include ethylene-propylene copolymers, ethylene-butene copolymers and ethylene-hexene copolymers, and they have the characteristics of adjustable crystallinity and high chemical stability. The infrared band range refers to the electromagnetic wave wavelength range used in spectral detection, which is between 2.5 micrometers and 25 micrometers, and is used to measure the molecular vibrational absorption characteristics of new chemical materials under different light energies. The peak intensity refers to the maximum absorbance value at the wavelength corresponding to the absorption peak in the spectrum, and the parameter is obtained directly by a spectrophotometer.

9. The method for quality testing of new chemical materials based on artificial intelligence according to claim 1, characterized in that, The standard absorption range is the typical absorption wavelength range of functional groups obtained by statistical analysis of publicly available infrared or ultraviolet-visible spectral databases, including the stretching vibration absorption region of aromatic carbon-carbon double bonds, which is located in the range of 1,400 to 1,600 wavenumbers. The half-width at half maximum (WHM) variation refers to the change in the width value of the spectral peak at half the maximum absorbance. The structural change segment refers to the spectral region generated between adjacent spectral segments due to changes in the functional group affiliation; The spectral parameters are spectral indices used to describe the morphological characteristics of absorption peaks, including absorbance, full width at half maximum (FWHM), peak position shift, and peak symmetry factor. The quality standard is a quality judgment criterion established based on the structural integrity, compositional uniformity, and spectral distribution of new chemical materials, and a structural comparison template is formed based on standard samples or industry testing specifications. A training dataset is constructed based on sample labels, and a convolutional neural network model is trained to realize the predictive relationship between spectral input feature sequences and quality levels. The model is then deployed in the detection system for predicting the quality of new samples.

10. A quality inspection system for new chemical materials based on artificial intelligence, characterized in that, The system is used to implement the artificial intelligence-based quality testing method for new chemical materials as described in any one of claims 1-9, and the system comprises: The spectral acquisition and analysis module acquires the absorption spectrum data of polyolefin copolymer materials in the infrared band, extracts the absorption peak positions, absorbance values ​​and half width at half maximum (WHM) parameters of the spectral bands, organizes all spectral band information in wavelength order, and generates a spectral response characteristic table after normalizing the absorbance data. The response feature extraction module calls the spectral response feature table to extract the peak position and bandwidth information of the spectral band, matches it with the wavelength range and the standard functional group stretching vibration range, establishes a correspondence between the matching results and the functional group type, and generates a functional group segment annotation information set. The functional group identification module calls the functional group segment annotation information set to identify the location of changes in functional group type between adjacent spectral segments, extracts the absorbance difference and full width at half maximum (FWHM) changes in the changed segments, determines the location and type of structural change based on the change characteristics of the absorption response parameters, and generates a spectral segment structural difference feature set. The structural difference extraction module calls the spectral segment structural difference feature set, organizes the functional group type, spectral shape parameters and response change trend corresponding to the spectral segment, establishes the mapping relationship between the spectral segment sequence and the material sample category label, and generates a spectral input feature sequence file; The quality status judgment module calls the spectral input feature sequence file, and judges the structural integrity and functional group composition of the sample based on the distribution of functional groups and response change patterns in the spectral structure sequence, generating a set of quality test results for new chemical materials. The artificial intelligence reasoning module calls the trained AI recognition model, inputs the spectral input feature sequence file, and outputs the quality grade prediction, structural integrity score and abnormal risk warning corresponding to the material sample, thereby realizing the intelligent discrimination function of the detection system.

Images