Infrared spectrum-based rubber auxiliary component analysis and quality control method and system

Abstract

The invention belongs to the technical field of spectrum detection and analysis, and discloses an infrared spectrum-based rubber auxiliary component analysis and quality control method and a system thereof, the method comprises the following steps: using a Fourier transform infrared spectrometer to collect infrared spectrum data of a rubber auxiliary mixture in real time, and preprocessing the infrared spectrum data; based on the preprocessed infrared spectrum data, a partial least square regression method is adopted to construct a multi-component quantitative model for component identification and quantitative analysis; according to a result of component identification and quantitative analysis, generating a content report and quality control information of the rubber auxiliary component; according to the method, the influence of environmental interference on infrared spectrum data can be effectively reduced, and real-time and accurate quantitative analysis and quality control can be carried out on each component in the rubber additive production process.

Description

Technical Field

[0001] The invention belongs to the technical field of spectrum detection and analysis, and in particular relates to a method and system for analyzing and controlling the composition of rubber additives based on infrared spectroscopy. Background Art

[0002] Rubber additives play a key role in rubber production and are usually used to improve rubber's vulcanization, strength, adhesion and anti-aging properties. However, in the production process of rubber additives, effective detection and control of the content of each component can ensure the consistency of the final product quality. In the existing production process, traditional analysis methods mainly rely on chemical titration, gas chromatography (GC), liquid chromatography (HPLC) and other methods. Although these methods have a certain degree of accuracy in the qualitative and quantitative analysis of components, they are difficult to meet the needs of real-time monitoring and rapid feedback in modern production due to cumbersome operations and long cycles.

[0003] Traditional rubber additive testing methods usually require sampling and offline analysis, and the testing cycle is long. This delayed detection will result in abnormal situations in the production process not being discovered and handled in a timely manner, thus affecting the stability of product quality. In addition, due to the complex composition of rubber additives, the characteristic spectra of each component often overlap, which poses a great challenge to traditional analysis methods. It is difficult to effectively extract the characteristic information of each component in a complex mixed spectrum. Especially when faced with infrared spectral data with multiple components and strong collinearity, the accuracy and reliability of traditional methods will be greatly reduced, resulting in a lack of precision and real-time quality control of the production process. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for analyzing and controlling the components of rubber additives based on infrared spectroscopy, which can perform real-time, accurate quantitative analysis and quality control of each component during the production process of rubber additives, reduce the impact of environmental interference on infrared spectral data, and improve production efficiency and consistency of product quality.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: The method for analyzing and controlling the composition of rubber additives based on infrared spectroscopy includes the following steps: Fourier transform infrared spectrometer is used to collect infrared spectrum data of rubber additive mixture in real time, and the infrared spectrum data is preprocessed; Based on the preprocessed infrared spectral data, partial least squares regression method was used to construct a multi-component quantitative model for component identification and quantitative analysis. Based on the results of component identification and quantitative analysis, a content report and quality control information of rubber additive components are generated.

[0006] The following is a further optimization of the above technical solution by the present invention: The steps for preprocessing infrared spectral data include: The collected infrared spectrum data is smoothed by using Savitzky-Golay filter; The infrared spectrum data after smoothing and filtering were subjected to baseline correction using polynomial fitting method. The baseline-corrected infrared spectral data were normalized using standard normal distribution.

[0007] Further optimization: Based on the preprocessed infrared spectrum data, the steps of using partial least squares regression method to construct a multi-component quantitative model for component identification and quantitative analysis include: Get the preprocessed infrared spectrum data matrix ;in: is the sample size, is the dimension of infrared spectrum data; Get the corresponding component content matrix ,in: is the number of components to be analyzed; The infrared spectrum data matrix is ​​transformed into and ingredient content matrix Project to the latent variable space and get the projection matrix and , and calculate the covariance matrix to construct a linear relationship; A partial least squares regression model was established and optimized by minimizing the mean square error to obtain the quantitative analysis results of each additive component. The established partial least squares regression model is used to quantitatively analyze the real-time infrared spectrum data and output the content data of each component of the rubber additive.

[0008] Further optimization: The steps for obtaining the ingredient content matrix include: The rubber additive mixture samples are standardized and prepared so that the content of each component is known; Determine the actual content value of each component by benchmark testing method; The actual content values ​​of each component obtained by measurement are constructed into a component content matrix.

[0009] Further optimization: Based on the results of component identification and quantitative analysis, the steps for generating content reports and quality control information for rubber additive components include: According to the content data of each component obtained by the partial least squares regression model, the content of each component is sorted and recorded to form a component content matrix of each component; Conduct statistical analysis on the content data of each component, calculate the average content and standard deviation of each component, and evaluate its stability; Compare the content data of each component with the preset quality control range to determine whether it is within the specified quality control range; Generate content reports and quality control information for rubber additive components, including the average content of each component, standard deviation, and whether it meets the preset quality control range.

[0010] Further optimization: Based on the pre-processed infrared spectrum data, the steps of using partial least squares regression method to construct a multi-group component quantitative model for component identification and quantitative analysis also include: The orthogonal matching pursuit algorithm is used to perform sparse decomposition of infrared spectrum data and extract the characteristic spectra of each component in rubber additives.

[0011] Further optimization: Based on the pre-processed infrared spectrum data, the steps of using partial least squares regression method to construct a multi-group component quantitative model for component identification and quantitative analysis also include: Dynamic data fusion and environmental compensation are used to denoise and drift correct infrared spectrum data to reduce the impact of environmental changes on analysis results.

[0012] The present invention also provides a rubber additive component analysis and quality control system based on infrared spectroscopy, which is applied to the above-mentioned rubber additive component analysis and quality control method based on infrared spectroscopy. The system comprises: Fourier transform infrared spectrometer, used to collect infrared spectrum data of rubber additive mixture in real time; A data preprocessing module is used to preprocess infrared spectrum data, including smoothing filtering, baseline correction and normalization processing; The component identification and quantitative analysis module is used to construct a multi-component quantitative model for component identification and quantitative analysis of the pre-processed infrared spectrum data using partial least squares regression method; The result generation module is used to generate content reports and quality control information of rubber additive components based on the results of component identification and quantitative analysis.

[0013] Further optimization: the system further includes: The sparse decomposition module is used to perform sparse decomposition on infrared spectrum data using the orthogonal matching pursuit algorithm and extract the characteristic spectra of each component in the rubber additive.

[0014] Further optimization: the system further includes: The environmental compensation module is used to denoise and drift correct infrared spectrum data through dynamic data fusion and environmental compensation to reduce the impact of environmental changes on analysis results.

[0015] The present invention adopts the above technical solution and has the following beneficial effects: 1. In the present invention, Fourier transform infrared spectroscopy (FTIR) technology is combined with partial least squares regression (PLS) technology to achieve high-precision component analysis of each component of the rubber additive; by constructing a multi-component component quantitative model, the content of each component in the rubber additive can be effectively identified and quantified, overcoming the problems of characteristic spectrum overlap and interference signals between components in traditional analysis methods, and greatly improving the accuracy of quantitative analysis.

[0016] 2. The present invention uses a Fourier transform infrared spectrometer installed on the production line, combined with Kalman filtering and dynamic data fusion technology, to achieve real-time monitoring of infrared spectral data and dynamic compensation for environmental changes; through real-time monitoring of environmental factors such as temperature and humidity and denoising and drift correction, the impact of the external environment on the infrared spectral data is effectively reduced, ensuring the stability and consistency of the infrared spectral data, thereby enabling the quality of the entire production process to be continuously controlled.

[0017] 3. The present invention adopts an orthogonal matching pursuit algorithm to perform sparse decomposition on infrared spectral data, decomposes the infrared spectral data in a complex mixture into sparse basis vectors, and extracts independent characteristic spectra of each component; through this decomposition technology, the uncertainty caused by the overlap of characteristic spectra between components is effectively reduced, making the infrared spectral data in subsequent analysis more representative and identifiable, greatly improving the detection efficiency and accuracy of each component.

[0018] 4. After analyzing the content of each component, the present invention generates a detailed content report and quality control information; the content report records the changes of each component at different times, and is evaluated through statistical data such as the average value and standard deviation to generate a complete quality judgment result; this enables abnormal fluctuations in the production process to be discovered and corrected in a timely manner, thereby ensuring the stability and consistency of the rubber additive ingredients.

[0019] 5. The present invention reduces manual intervention and simplifies the operational process of production control through automated infrared spectrum data collection, preprocessing, analysis and report generation; the automation of data collection and analysis steps not only improves detection efficiency, but also reduces the cumbersome manual operations in traditional chemical analysis, making quality control on the production line more convenient and efficient, thereby reducing production costs.

[0020] The present invention is further described below in conjunction with the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a schematic diagram of the method flow of Example 1 of the present invention; Figure 2This is a schematic diagram of the method flow of Example 2 of the present invention; Figure 3 This is a schematic diagram of the method flow of Example 3 of the present invention; Figure 4 This is a schematic diagram of the method flow of Example 4 of the present invention; Figure 5 This is a schematic diagram of the structure of the system in Example 5 of the present invention.

[0022] In the figure: 10-Fourier transform infrared spectrometer; 20-data preprocessing module; 30-component identification and quantitative analysis module; 40-sparse decomposition module; 50-environmental compensation module; 60-result generation module. DETAILED DESCRIPTION

[0023] Example 1: Please refer to the attached Figure 1 The present invention provides a method for analyzing and controlling the components of rubber additives based on infrared spectroscopy. The method uses Fourier transform infrared spectroscopy (FTIR) technology to obtain infrared spectral data of rubber additives, combines partial least squares regression (PLS) method, and constructs a multi-group component quantitative model to achieve identification and quantitative analysis of rubber additive components. The method can not only quantitatively analyze the content of each component of the rubber additive, but also monitor and control the quality fluctuation in the production process to ensure the stable quality of the final product.

[0024] The method for analyzing and controlling the composition of rubber additives based on infrared spectroscopy specifically includes the following steps: S1. Use a Fourier transform infrared spectrometer to collect infrared spectrum data of the rubber additive mixture in real time, and preprocess the infrared spectrum data.

[0025] In this embodiment, the specific implementation of step S1 is as follows: Firstly, the infrared spectral data of the rubber additive mixture was collected in real time using a Fourier transform infrared spectrometer. The Fourier transform infrared spectrometer can generate spectral information related to the sample composition based on the sample's absorption of infrared light at different wavelengths. The technology measures the molecular vibration characteristics of different components to reflect the molecular structure and component distribution of the sample. The working principle of the Fourier transform infrared spectrometer is closely related to infrared absorption spectroscopy, and can capture the characteristic absorption peaks of various functional groups in the molecules, thereby realizing the quantitative analysis of each component in the rubber additive mixture.

[0026] In practical applications, the Fourier transform infrared spectrometer needs to be linked with the rubber additive production line; therefore, the Fourier transform infrared spectrometer is first installed on the rubber additive production line to ensure that it can work synchronously with the production line and obtain the infrared spectrum data of the sample in real time; in order to ensure the accuracy and real-time nature of the infrared spectrum data, the installation position of the Fourier transform infrared spectrometer is usually set at the key control points of the production line, such as mixing, conveying or packaging, to ensure that each batch of rubber additive mixture can be monitored in real time.

[0027] As an option, the FTIR spectrometer can be set to a wavelength range of 4000 cm -1 Up to 400cm -1 ; This range can cover the absorption peaks of common functional groups in rubber additives, such as C=O (carbonyl), CH (carbon-hydrogen bond) and NH (amino) and other characteristic peaks. These bands are closely related to the chemical structure of rubber additives.

[0028] The working principle of the Fourier transform infrared spectrometer is to divide the light beam of the infrared light source into two beams through the FTIR interferometer, one beam passes through the sample and the other beam passes through the reference channel; after the two beams interfere, the infrared absorption spectrum of the sample is obtained through Fourier transform; the Fourier transform infrared spectrometer can obtain the absorbance information of the sample at different wave numbers, and then determine the characteristic spectrum of each component in the sample.

[0029] Specifically, the infrared spectrum data of rubber additives are usually expressed as absorbance at different wave numbers. ,in: is the wave number, in cm -1 .

[0030] In this embodiment, the infrared spectrum data can be expressed as the following formula: ; in: For each component at wave number The absorbance at is a time variable, which is related to the scanning process of the FTIR interferometer; is the noise that may exist during the acquisition process.

[0031] This formula reflects the absorbance changes of each component in the rubber additive sample at different wave numbers and provides the necessary infrared spectrum data for subsequent component analysis.

[0032] In the production process of rubber additives, the Fourier transform infrared spectrometer continuously flows the sample through its sampling window to ensure that each part of the rubber additive mixture is excited and generates an absorption signal within the sampling window; through continuous sampling, the Fourier transform infrared spectrometer can record the infrared spectrum data of the rubber additive mixture in real time; these real-time infrared spectrum data provide component absorption information at each time point, laying the foundation for subsequent component identification and quantitative analysis.

[0033] Specifically, the flow of the rubber additive mixture in the production line can ensure that multiple samples are tested, and the collected infrared spectral data can cover the entire production process, thereby avoiding deviations caused by local sample unevenness or operational errors; the original infrared spectral data obtained from each sampling is Fourier transformed through the FTIR interferometer and converted into infrared spectral data in the same frequency domain, thereby obtaining complete spectral information containing the characteristic absorption peaks of each component.

[0034] The frequency of data acquisition should match the flow rate and mixing rate of the production line to ensure that no key characteristic spectrum is missed; in this real-time spectrum acquisition process, the Fourier transform infrared spectrometer will be able to accurately capture the changes in rubber additives and further enhance the reliability of the analysis model.

[0035] After the infrared spectral data collection is completed, in order to ensure the accuracy and reliability of subsequent component identification and quantitative analysis, the original collected infrared spectral data needs to be preprocessed; the purpose of preprocessing is to remove noise and baseline drift in the data, as well as eliminate spectral deviations caused by different sample thickness and concentration, to ensure the stability and availability of infrared spectral data, thereby providing high-quality input data for subsequent spectral analysis.

[0036] In general, infrared spectroscopy data is often affected by a variety of factors, such as equipment fluctuations, sample surface inhomogeneities, changes in the production environment, etc.; therefore, data preprocessing is a key step to ensure the accuracy of subsequent analysis results.

[0037] In order to remove high-frequency noise in the original collected infrared spectrum data, it is first necessary to perform smoothing filtering on the original collected infrared spectrum data; through smoothing filtering, signal fluctuations caused by equipment noise, environmental factors or other sudden interference can be reduced.

[0038] In this embodiment, the smoothing filter processing method is: applying a Savitzky-Golay filter to the collected infrared spectrum data to perform smoothing filter processing.

[0039] The smoothing filter process can effectively retain the main features of the spectrum while removing high-frequency noise; in this method, the smoothed infrared spectrum data is obtained by performing local polynomial fitting on short segments of the infrared spectrum data.

[0040] In this embodiment, the calculation formula for smoothing filtering is: ; in: is the smoothed infrared spectrum data point; is the original data point of infrared spectrum data; is the coefficient of the filter; is the window size, controlling the degree of smoothing.

[0041] By adjusting The value of can control the degree of smoothing, usually smaller Values ​​are used to remove slight noise, while larger Values ​​are used to remove more noticeable noise.

[0042] In the process of collecting infrared spectral data, there may be baseline drift due to equipment aging, changes in ambient temperature and humidity, or the inhomogeneity of the sample itself; baseline drift will affect the shape of the entire spectrum, and thus affect the accurate extraction of the characteristic spectrum; therefore, baseline correction is required in preprocessing.

[0043] In this embodiment, the baseline correction method is: the infrared spectrum data after smoothing and filtering is subjected to baseline correction by using a polynomial fitting method.

[0044] In baseline correction, the polynomial fitting method is used to estimate and correct the baseline; the polynomial function obtained by fitting can effectively remove the influence of baseline drift and ensure the stability of infrared spectral data.

[0045] In this embodiment, the polynomial fitting formula used for baseline correction is: ; in: is the wave number; is the fitting coefficient.

[0046] By performing baseline correction on infrared spectrum data, infrared spectrum data after baseline drift is removed It can be expressed as: ; The baseline correction step can effectively eliminate the low-frequency drift in the spectrum, making the characteristic peaks of the spectrum clearer, thereby improving the accuracy of subsequent analysis.

[0047] Since different batches of rubber additive samples may have differences in thickness and concentration, resulting in inconsistent intensity of infrared spectral data, normalization processing is required to eliminate the impact of these differences.

[0048] In this embodiment, the normalization process is performed by performing standard normal variable normalization process on the infrared spectrum data after baseline correction.

[0049] Through normalization processing, the infrared spectral data can be adjusted to have zero mean and unit standard deviation, thereby eliminating the variation caused by sample thickness and concentration, making the infrared spectral data between different samples more comparable.

[0050] In this embodiment, the normalization formula is: ; in: is the original collected infrared spectrum data; is the zero mean of infrared spectrum data; is the standard deviation of the infrared spectrum data.

[0051] After normalization, the infrared spectral data of all samples will have the same standardized scale, which is convenient for subsequent component analysis and quantitative calculation.

[0052] Generally, smoothing filtering is combined with normalization to ensure that different types of noise are removed; smoothing filtering is used to remove high-frequency noise, while baseline correction removes low-frequency drift, and normalization further eliminates variations due to sample thickness, concentration changes, and different experimental conditions.

[0053] Specifically, in the preprocessing process, the infrared spectral data is first smoothed and filtered to remove high-frequency noise, then baseline correction is performed to eliminate low-frequency drift, and finally normalization is performed to eliminate the impact of sample differences; in this way, the final infrared spectral data will have high stability and consistency, providing a reliable basis for subsequent component identification and quantitative analysis.

[0054] After the above preprocessing steps, the noise and unstable factors in the infrared spectral data can be removed to form a clean, noise-free and consistent multidimensional spectral matrix; the data in this matrix will be a refined version of the infrared spectral data and can be used for subsequent spectral decomposition, component identification and quantitative analysis.

[0055] The pre-processed infrared spectral data can provide more accurate and stable results for subsequent component identification and quantitative analysis; the effective implementation of the pre-processing step greatly improves the reliability of the entire analysis process and provides a solid technical foundation for the quality control of rubber additives.

[0056] S2. Based on the preprocessed infrared spectral data, the partial least squares regression method is used to construct a multi-component quantitative model for component identification and quantitative analysis.

[0057] In order to achieve quantitative analysis of each component in the rubber additive mixture, this embodiment uses partial least squares regression (PLS) to construct a multi-component component quantitative model; PLS is a multivariate linear regression method widely used in infrared spectral data analysis, especially suitable for situations where there is strong collinearity between infrared spectral data and component content; by modeling infrared spectral data and reference data of known concentration, PLS can extract potential variables, thereby achieving accurate prediction of the content of each additive component.

[0058] In this embodiment, the specific steps of step S2 are as follows: First, the preprocessed infrared spectral data have been free of noise and baseline drift, and have been normalized; the infrared spectral data obtained in this way are more stable and consistent, providing high-quality input data for subsequent modeling; based on these preprocessed infrared spectral data, a multi-component quantitative model was established.

[0059] Before performing PLS regression analysis, it is necessary to prepare the infrared spectrum data matrix and ingredient content matrix , the specific steps are as follows: First, obtain the preprocessed infrared spectrum data matrix ,in: is the sample size, is the dimension of infrared spectrum data.

[0060] The infrared spectrum data matrix It contains the preprocessed infrared spectrum data; each row represents a sample, and each column corresponds to a specific wave number point, that is, a different band of the infrared spectrum.

[0061] Then get the corresponding component content matrix ,in: is the number of components to be analyzed.

[0062] The ingredient content matrix The component content data of each sample is stored in ; each row corresponds to the multidimensional component content of a sample.

[0063] In order to ensure the accuracy and reliability of the model, it is necessary to obtain an accurate component content matrix through standardized sample preparation and benchmark testing methods. ; These data will serve as the basis for training data; Generally, the configuration of standard samples should ensure that all possible component concentration ranges are covered to ensure that the model has sufficient generalization ability.

[0064] The steps to obtain the ingredient content matrix include: 1. Preparation of standardized samples: The rubber additive mixture samples are standardized so that the content of each component is known.

[0065] Specifically, by accurately preparing rubber additive mixtures with different concentration ratios, the component concentrations of each sample are known and prepared before infrared spectroscopy data collection.

[0066] 2. Benchmark test method: The actual content value of each component is determined by the benchmark test method.

[0067] Specifically, chemical analysis methods (such as gas chromatography or liquid chromatography) are used to perform benchmark measurements on each standardized sample to obtain the actual content value of each component and ensure the accuracy of the actual content value of each component.

[0068] 3. Construct a component content matrix: Construct the actual content values ​​of each component obtained by measurement into a component content matrix.

[0069] Then, the infrared spectrum data matrix is ​​transformed into and ingredient content matrix Project to the latent variable space and get the projection matrix and , and calculate the covariance matrix to construct a linear relationship.

[0070] Prepare infrared spectrum data matrix and ingredient content matrix ,The next step is to establish a partial least squares regression model and optimize the model by minimizing the mean square error to obtain the quantitative analysis results of each auxiliary component.

[0071] Among them, the core idea of ​​the partial least squares regression model is to reduce the dimension of both input data and output data at the same time and extract the potential variables that can best explain the relationship between the two.

[0072] The mathematical expression of the partial least squares regression model is: ; ; in: and They are the projection matrices of the infrared spectrum data matrix and the component content matrix in the latent variable space; and is the loading matrix, describing the representation of the original collected infrared spectroscopy data in latent variables; and are the residual matrices, representing the part that cannot be explained by the latent variables.

[0073] In the partial least squares regression model, the selection of latent variables is optimized by minimizing the mean square error (MSE), and then the model is optimized, that is, by extracting the relevant information between the spectrum and the component to establish a mapping relationship between the two.

[0074] When training a partial least squares regression model, the goal of this embodiment is to find the optimal number of latent variables by minimizing the mean square error (MSE).

[0075] MSE is defined as: ; in: is the actual content data, is the content predicted by the partial least squares regression model; by minimizing the mean square error, the model will continuously adjust the number of latent variables and coefficients to improve the prediction accuracy.

[0076] In order to improve the stability and generalization ability of the model, this embodiment adopts a cross-validation method to evaluate the model. During the cross-validation process, the data set will be divided into multiple subsets, and some subsets will be used alternately as training sets and other subsets as validation sets. In this way, the optimal number of latent variables can be selected and it can be ensured that the model can adapt to different sample data.

[0077] After the partial least squares regression model is trained, the model can be used to predict the component content of new infrared spectral data; the prediction process projects the input infrared spectral data into the latent variable space and uses the loading matrix in the model The projected values ​​are mapped back to the component content space to obtain the predicted content value of each sample.

[0078] The calculation formula for the predicted content value is: ; in: is the projection matrix of the input infrared spectrum data in the latent variable space, is the loading matrix used to map the projection value back to the content space. It is the predicted component content value of each sample.

[0079] The established partial least squares regression model is used to quantitatively analyze the real-time infrared spectrum data and output the content data of each component of the rubber additive.

[0080] Through the partial least squares regression model analysis, the final result will be the component content information of each sample; for each sample, the partial least squares regression model can provide the predicted content of each component, thereby outputting the content data of each component of the rubber additive.

[0081] In some embodiments, the partial least squares regression model can be further optimized, for example, by using different preprocessing methods (such as different smoothing filtering techniques or other baseline correction methods) to improve the quality of the data, thereby improving the accuracy of the quantitative analysis; in addition, in the process of component content prediction, other advanced regression techniques (such as kernel principal component regression technology, Kriging regression technology, etc.) can be combined to improve the partial least squares regression model to improve the accuracy and stability of the prediction.

[0082] Through the above steps, the present invention adopts partial least squares regression method to identify components and quantitatively analyze infrared spectrum data; the method establishes a quantitative model of multiple components by extracting potential variables between infrared spectrum data and component content; after cross-validation and minimizing mean square error to optimize the model, the obtained model has good prediction ability and can accurately predict the content data of each component of the rubber additive.

[0083] S3. Generate a content report and quality control information of rubber additive ingredients based on the results of component identification and quantitative analysis.

[0084] After component identification and quantitative analysis of the infrared spectral data of the rubber additive mixture, the next step is to generate content reports and quality control information; this process is crucial for monitoring and optimizing the production process; the content report can record the content of each component in detail, while the quality control information is used to evaluate the volatility of the component content and determine whether it meets the set quality standards; through these content reports and quality control information, production personnel can identify problems in a timely manner and make adjustments to ensure the consistency and stability of product quality.

[0085] In this embodiment, the specific implementation of step S3 is as follows: Firstly, according to the content data of each component obtained by the partial least squares regression model, the content of each component was sorted and recorded to form the component content matrix of each component.

[0086] Among them: each row in the ingredient content matrix represents the content of each component at a time point or sample, and each column represents the content data of a specific ingredient at different sampling points; in general, the summarized data should include the quantitative analysis results corresponding to the infrared spectral data of all samples to ensure that all possible fluctuations and changes in the production process are covered.

[0087] In order to evaluate the content volatility and stability of each component, the content data of each component were statistically analyzed, and the average content and standard deviation of each component were calculated to evaluate its stability.

[0088] Specifically, the stability and consistency of the content of each component are evaluated by calculating the average content and standard deviation of each component; the specific calculation formula is as follows: Average content The calculation formula is: ; in: For the The components in The content at each time point.

[0089] Standard Deviation The calculation formula is: ; in: For the The average content of each component, For the The components in The content at each time point.

[0090] By calculating the average content and standard deviation of each component, we can fully understand the content changes of each component and evaluate its volatility; this statistical result provides an important basis for subsequent quality control and production adjustments.

[0091] In order to ensure that the product quality meets the predetermined standards, the content data of each component is compared with the preset quality control range to determine whether it is within the specified quality control range.

[0092] The preset quality control range is usually based on the maximum and minimum content range of each component. Indicates that the content of each component must be within this range; if the content of a certain component exceeds this range, it means that it does not meet the quality requirements.

[0093] The content data determination rules for each component are: ; If the content data of a certain component If it exceeds the preset range, the data is marked as not meeting the quality standards and a corresponding alarm message is generated; this judgment process ensures that each component in the production process is within the qualified content range, thereby ensuring the quality of the product.

[0094] In order to more intuitively observe the changing trend of the content of each component, a content fluctuation chart is usually generated based on the statistical results; the fluctuation chart shows the content changes of each component at different time points or sampling points, and also marks the average content and standard deviation range of each component; through the fluctuation chart, operators can quickly identify abnormal fluctuations in content and take timely measures to make adjustments to ensure stable operation of the production process.

[0095] For example, in a fluctuation chart, if the content of a certain component exceeds the standard deviation range, it will be visualized as an obvious abnormal point, prompting production personnel that the content of this component needs to be adjusted.

[0096] By summarizing and statistically analyzing the content data and the results of quality judgment, a content report and quality control information of rubber additive components are finally generated. The report includes the following items: The average content of each component, standard deviation and whether it meets the preset quality control range.

[0097] In addition, quality control information is generated based on the above reports and includes suggestions and alarm information for the production process, reminding operators to adjust or optimize ingredients that do not meet quality requirements.

[0098] The generated rubber additive content report and quality control information will be stored in the database for quality management personnel to review and subsequent analysis; the stored historical data can not only be used for quality assessment of the current production batch, but also for long-term trend analysis to help optimize the production process and improve the stability of product quality.

[0099] In some embodiments, the storage and analysis of historical data can also be used to continuously improve control parameters in the production process, further reduce fluctuations in product quality, and ensure that each batch of rubber additives can achieve the best composition ratio and stability.

[0100] Through the above steps, this embodiment successfully generates a complete content report and quality control information of the rubber additive component; through statistical analysis of the component content data, quality judgment, fluctuation chart generation and report output, the present invention provides a quality monitoring and feedback mechanism for the production process, effectively ensuring the stability of product quality.

[0101] In general, the present invention constructs a multi-component quantitative model by combining Fourier transform infrared spectroscopy technology and partial least squares regression method. First, the infrared spectrum data of the rubber additive mixture is preprocessed and the component quantitative analysis is performed to accurately identify the content of each component; then, based on the analysis results, a detailed content report and quality control information are generated, including the average content of each component, the standard deviation and the judgment result of whether it meets the preset quality control range; finally, through the statistical analysis of the fluctuation of the component content, the stability of each component in the production process is ensured, accurate data support is provided for quality control, the production process is optimized and the stability of product quality is improved.

[0102] Example 2: Please refer to the attached Figure 2Based on the above embodiment 1, the specific steps of the method include: before the step of constructing a multi-group component quantitative model for component identification and quantitative analysis using partial least squares regression method based on the pre-processed infrared spectrum data in S2, the method also includes: J. Use the orthogonal matching pursuit algorithm to perform sparse decomposition of infrared spectral data and extract the characteristic spectra of each component in rubber additives.

[0103] In this embodiment, the orthogonal matching pursuit (OMP) algorithm is used to perform sparse decomposition on the infrared spectrum data of the rubber additive, which can effectively extract the characteristic spectrum of each component in the mixed spectrum; since the infrared spectrum data of the rubber additive is usually a superposition of multiple components, directly using the original collected infrared spectrum data for analysis is easily interfered by the overlap of characteristic spectra, resulting in inaccurate analysis results; therefore, the orthogonal matching pursuit algorithm (OMP) is used for sparse decomposition, which can decompose the complex infrared spectrum data into the characteristic spectrum of each component, providing more accurate spectral information for subsequent component identification and quantitative analysis.

[0104] In this embodiment, the preprocessed infrared spectrum data is first expressed as a vector form, and the infrared spectrum data matrix is ​​assumed to be ,in: is the time point or sample size, is the dimension of infrared spectrum data, usually the absorbance value corresponding to the wave number; by vectorizing the infrared spectrum data, it is convenient for subsequent matrix calculation and decomposition operations.

[0105] Next, construct a sparse dictionary matrix ,in: is the number of dictionary basis vectors; each basis vector ,in: Represents the characteristic spectra of known components; these basis vectors are combined into a dictionary matrix , which can represent the characteristic space of each component in the rubber additive; in this way, the infrared spectral data can be represented as a sparse linear combination, that is, the contribution degree of each basis vector in the infrared spectral data.

[0106] In the process of sparse decomposition, the residual is first initialized , and the sparse coefficient vector Set to zero.

[0107] Then, in each iteration, from the dictionary matrix Select one of the following that matches the current residual The most relevant basis vector Specifically, the basis vector that best explains the current residual is selected through inner product calculation, and its selection rule is: ; in: Residual With basis vectors The inner product of The index of the currently selected basis vector.

[0108] After selecting the basis vector, update the sparse coefficient corresponding to the basis vector , sparse coefficient The formula is: ; in: Indicates Basis vectors The corresponding sparse coefficient; after updating The value of represents the contribution of the basis vector to the infrared spectrum data.

[0109] Subsequently, the new residual is calculated using the updated sparse coefficients ; New residual The formula is: ; in: is the new residual, which is the original residual minus the part of the infrared spectrum data explained by the current basis vector.

[0110] This process is iterated until the residual norm is less than the preset threshold, or reaches the maximum number of iterations to avoid infinite loops; the final sparse coefficient vector It contains the contribution value of each basis vector in the infrared spectrum data, and can effectively decompose the characteristic spectrum of each component.

[0111] Finally, by transforming the sparse coefficient vector With dictionary matrix Combine and reconstruct the infrared spectrum data sparsely.

[0112] Infrared spectrum data after sparse reconstruction for: ; in: The sparsely reconstructed infrared spectral data can independently reflect the characteristic spectra of each component in the rubber additive. The sparse decomposition of the orthogonal matching pursuit algorithm can effectively reduce the overlap between the spectral peaks of multiple components, thereby ensuring the accuracy and reliability of subsequent component identification and quantitative analysis.

[0113] In general, in this embodiment, the orthogonal matching pursuit algorithm is used to perform sparse decomposition of infrared spectral data and extract the characteristic spectra of each component in the rubber additive, which provides a more accurate basis for the subsequent quantitative analysis of components and can effectively improve the accuracy of the analysis process.

[0114] Example 3: Please refer to the attached Figure 3 Based on the above embodiment 1, the specific steps of the method include: before the step of constructing a multi-group component quantitative model for component identification and quantitative analysis using partial least squares regression method based on the pre-processed infrared spectrum data in S2, the method also includes: Y. De-noise and drift correction are performed on infrared spectral data through dynamic data fusion and environmental compensation to reduce the impact of environmental changes on analysis results.

[0115] In this embodiment, the infrared spectrum data is denoised and drift corrected through dynamic data fusion and environmental compensation, which can reduce the influence of interference caused by external environmental factors (such as temperature, humidity, etc.) on the analysis results; in the actual production environment, the infrared spectrum data of rubber additives is easily affected by environmental changes, and these environmental changes may cause the drift of infrared spectrum data or increase noise, thereby affecting the accuracy of component analysis; in order to ensure the stability and reliability of infrared spectrum data, the present invention combines dynamic data fusion and environmental compensation technology in the process of infrared spectrum data processing, which can effectively remove environmental noise and drift, and ensure the accuracy of subsequent analysis results.

[0116] In this embodiment, first, during the spectrum acquisition process, environmental sensors are installed to monitor parameters such as temperature and humidity in the production environment in real time; these environmental data provide a reference basis for subsequent compensation and correction; in the actual operation process, by real-time monitoring of these environmental variables, real-time feedback information can be provided for drift correction of infrared spectrum data, thereby avoiding the adverse effects of environmental factors on the analysis process.

[0117] In the specific implementation of denoising and drift correction, this embodiment uses a Kalman filter algorithm to perform preliminary dynamic compensation and denoising; Kalman filtering is a recursive algorithm that is often used to estimate the state of a dynamic system and is particularly suitable for processing noisy signals.

[0118] The state transfer equation of the Kalman filter is: ; The measurement equation of Kalman filter is: ; in: is the current system status, representing the current infrared spectrum data; is the state transfer matrix, which describes the change of system state; It is the control input, which indicates the changes in the external environment; is the process noise, which is assumed to be Gaussian distributed; is the measured value, i.e. the infrared spectrum data actually collected; is the observation matrix, which maps the state to the measurement value; For the measurement noise, a Gaussian distribution is also assumed.

[0119] Through Kalman filtering, the state of infrared spectral data can be estimated, and the noise interference caused by changes in the external environment can be effectively reduced.

[0120] However, the noise and environmental influences in infrared spectral data often have more complex nonlinear characteristics, and Kalman filtering is mainly applicable to linear systems; therefore, this embodiment also combines variational autoencoder (VAE) to further denoise the infrared spectral data; VAE is a generative model that generates noise-free infrared spectral data by optimizing the objective function.

[0121] The formula for optimizing the objective function is: ; in: is the input infrared spectrum data; is the approximate posterior distribution, for the reconstructed distribution; is the Kullback-Leibler divergence, which measures the gap between the approximate posterior and the true posterior.

[0122] By optimizing the above objectives, VAE can generate infrared spectral data with reduced noise, which is particularly suitable for processing complex nonlinear noise; after VAE denoising, the signal-to-noise ratio of infrared spectral data is significantly improved while maintaining the characteristic information of each component.

[0123] On the basis of Kalman filtering and VAE denoising, this embodiment further utilizes dynamic data fusion technology to perform drift correction on infrared spectral data; dynamic data fusion is based on the linear weighted average method, and dynamically fuses the measurement data from environmental monitoring with the compensated infrared spectral data to further correct the drift in the infrared spectral data.

[0124] The specific dynamic data fusion formula is: ; in: is the infrared spectrum data after Kalman filtering and VAE denoising; Representation and wave number Relevant environmental compensation data; and The dynamic data fusion weight coefficient is set according to the experience value and data in the actual production environment to ensure that the infrared spectrum data after dynamic data fusion can accurately reflect the characteristic information of the rubber additive components.

[0125] In this way, the infrared spectrum data after dynamic data fusion can more accurately reflect the true characteristics of rubber additive components and reduce the interference caused by environmental factors.

[0126] The infrared spectral data after dynamic data fusion and environmental compensation processing has higher reliability and stability; by removing environmental noise and drift, the accuracy of subsequent component identification and quantitative analysis results is ensured; this method provides a solid data foundation for spectral analysis and can effectively improve the accuracy of component identification and content analysis.

[0127] Example 4: Please refer to the attached Figure 4 Based on the above-mentioned embodiments 1-3, in the specific steps of this embodiment 4, the method further includes the following steps before the step of constructing a multi-component quantitative model for component identification and quantitative analysis based on the pre-processed infrared spectrum data by using partial least squares regression method in S2: Y. De-noise and drift correction of infrared spectrum data through dynamic data fusion and environmental compensation to reduce the impact of environmental changes on analysis results; J. Use the orthogonal matching pursuit algorithm to perform sparse decomposition of infrared spectral data and extract the characteristic spectra of each component in rubber additives.

[0128] Specifically, the quality of infrared spectral data is first improved through environmental compensation and denoising technology to eliminate environmental noise and drift; then the orthogonal matching pursuit algorithm is used to sparsely decompose the optimized infrared spectral data to extract accurate characteristic spectra of each component, thereby ensuring the accuracy and reliability of component identification and quantitative analysis results; thereby improving the data's anti-interference ability and enhancing the accuracy of component analysis.

[0129] Example 5: Please refer to the attached Figure 5 Based on the above embodiments 1-4, the present invention also provides a rubber additive component analysis and quality control system based on infrared spectroscopy, which is applied to the above rubber additive component analysis and quality control method based on infrared spectroscopy, and the system comprises: A Fourier transform infrared spectrometer 10, used for collecting infrared spectrum data of the rubber additive mixture in real time; The data preprocessing module 20 is used to preprocess the infrared spectrum data, and the preprocessing includes smoothing filtering, baseline correction and normalization processing; The component identification and quantitative analysis module 30 is used to construct a multi-component quantitative model for component identification and quantitative analysis of the infrared spectrum data after preprocessing by using partial least squares regression method; The sparse decomposition module 40 is used to perform sparse decomposition on the infrared spectrum data by using an orthogonal matching pursuit algorithm to extract characteristic spectra of each component in the rubber additive; The environmental compensation module 50 is used to perform denoising and drift correction on the infrared spectrum data through dynamic data fusion and environmental compensation to reduce the impact of environmental changes on the analysis results; The result generation module 60 is used to generate a content report and quality control information of the rubber additive component according to the results of the component identification and quantitative analysis.

[0130] In this embodiment, the infrared spectroscopy-based rubber additive component analysis and quality control system is used to execute the above-mentioned infrared spectroscopy-based rubber additive component analysis and quality control method, and its principles and technical effects are similar and will not be repeated here.

[0131] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The method for analyzing and controlling the composition of rubber additives based on infrared spectroscopy is characterized by: The following steps are involved: Fourier transform infrared spectrometer is used to collect infrared spectrum data of rubber additive mixture in real time, and the infrared spectrum data is preprocessed; Based on the preprocessed infrared spectral data, partial least squares regression method was used to construct a multi-component quantitative model for component identification and quantitative analysis. Based on the results of component identification and quantitative analysis, a content report and quality control information of rubber additive components are generated.

2. The method for analyzing and controlling the composition of rubber additives based on infrared spectroscopy according to claim 1, characterized in that: The steps for preprocessing infrared spectral data include: The collected infrared spectrum data is smoothed by using Savitzky-Golay filter; The infrared spectrum data after smoothing and filtering were subjected to baseline correction using polynomial fitting method. The baseline-corrected infrared spectral data were normalized using standard normal distribution.

3. The method for analyzing and controlling the composition of rubber additives based on infrared spectroscopy according to claim 2, characterized in that: Based on the preprocessed infrared spectral data, the steps of constructing a multi-component quantitative model for component identification and quantitative analysis using partial least squares regression method include: Get the preprocessed infrared spectrum data matrix ,in: is the sample size, is the dimension of infrared spectrum data; Get the corresponding component content matrix ,in: is the number of components to be analyzed; The infrared spectrum data matrix is ​​transformed into and ingredient content matrix Project to the latent variable space and get the projection matrix and , and calculate the covariance matrix to construct a linear relationship; A partial least squares regression model was established and optimized by minimizing the mean square error to obtain the quantitative analysis results of each additive component. The established partial least squares regression model is used to quantitatively analyze the real-time infrared spectrum data and output the content data of each component of the rubber additive.

4. The method for analyzing and controlling the composition of rubber additives based on infrared spectroscopy according to claim 3, characterized in that: The steps to obtain the ingredient content matrix include: The rubber additive mixture samples are standardized and prepared so that the content of each component is known; Determine the actual content value of each component by benchmark testing method; The actual content values ​​of each component obtained by measurement are constructed into a component content matrix.

5. The method for analyzing and controlling the composition of rubber additives based on infrared spectroscopy according to claim 4, characterized in that: Based on the results of component identification and quantitative analysis, the steps to generate content reports and quality control information for rubber additive components include: According to the content data of each component obtained by the partial least squares regression model, the content of each component is sorted and recorded to form a component content matrix of each component; Conduct statistical analysis on the content data of each component, calculate the average content and standard deviation of each component, and evaluate its stability; Compare the content data of each component with the preset quality control range to determine whether it is within the specified quality control range; Generate content reports and quality control information for rubber additive components, including the average content of each component, standard deviation, and whether it meets the preset quality control range.

6. The method for analyzing and controlling the composition of rubber additives based on infrared spectroscopy according to claim 1, characterized in that: Based on the preprocessed infrared spectral data, the steps of using partial least squares regression method to construct a multi-component component quantitative model for component identification and quantitative analysis also include: The orthogonal matching pursuit algorithm is used to perform sparse decomposition of infrared spectrum data and extract the characteristic spectra of each component in rubber additives.

7. The method for analyzing and controlling the composition of rubber additives based on infrared spectroscopy according to claim 1, characterized in that: Based on the preprocessed infrared spectral data, the steps of using partial least squares regression method to construct a multi-component component quantitative model for component identification and quantitative analysis also include: Dynamic data fusion and environmental compensation are used to denoise and drift correct infrared spectrum data to reduce the impact of environmental changes on analysis results.

8. The infrared spectroscopy-based rubber chemical component analysis and quality control system is applied to the infrared spectroscopy-based rubber chemical component analysis and quality control method according to any one of claims 1 to 7, characterized in that: The system comprises: Fourier transform infrared spectrometer, used to collect infrared spectrum data of rubber additive mixture in real time; A data preprocessing module is used to preprocess infrared spectrum data, including smoothing filtering, baseline correction and normalization processing; The component identification and quantitative analysis module is used to construct a multi-component quantitative model for component identification and quantitative analysis of the pre-processed infrared spectrum data using partial least squares regression method; The result generation module is used to generate content reports and quality control information of rubber additive components based on the results of component identification and quantitative analysis.

9. The infrared spectrum-based rubber additive component analysis and quality control system according to claim 8, characterized in that: The system further comprises: The sparse decomposition module is used to perform sparse decomposition on infrared spectrum data using the orthogonal matching pursuit algorithm and extract the characteristic spectra of each component in the rubber additive.

10. The infrared spectrum-based rubber additive component analysis and quality control system according to claim 8, characterized in that: The system further comprises: The environmental compensation module is used to denoise and drift correct infrared spectrum data through dynamic data fusion and environmental compensation to reduce the impact of environmental changes on analysis results.

Images