A method for evaluating and adjusting the dispersion state of a melt extrusion material

Abstract

The application relates to a melt extrusion material dispersion state evaluation and adjustment method in the information technology field, which comprises the following steps: separating color space deviation data and particle distribution quantitative indexes through the surface image information to form surface quality evaluation parameters; fusing the initial quantitative results and the surface quality evaluation parameters to generate comprehensive evaluation values reflecting distribution uniformity through a preset signal cross-domain mapping method; judging whether dispersion state abnormalities exist according to the comprehensive evaluation values and preset process threshold values, and identifying abnormal sources in combination with signal time series and data change mode; determining an abnormal mode identification path by analyzing the correlation between color coordinate mutations and dispersion state stability; and generating adjustment instructions for the running state of the extrusion equipment according to the abnormal mode identification path and process parameter fluctuation historical data and sending the adjustment instructions to a production line control system to complete parameter closed-loop adjustment.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for evaluating and adjusting the dispersion state of melt extruded materials. Background Technology

[0002] Real-time monitoring of the uniformity of pigment and UV-resistant additive distribution in polymer melt extrusion production lines faces severe technical challenges: the intense flow of high-temperature melt severely interferes with the acquisition of UV-Vis absorbance signals, resulting in low accuracy in extracting the intensity of absorption peaks at specific wavelengths and making it impossible to reliably quantify the initial dispersion state of the material; at the same time, when high-resolution images are captured in real time on the surface of the extruded strip, the changes in color saturation and surface texture distribution are difficult to separate accurately due to equipment vibration and ambient light fluctuations, leading to increased deviations in the extraction of color space deviation data and particle distribution indicators. These heterogeneous signals exhibit cross-domain mismatch issues, making it difficult to effectively integrate spectral absorbance characteristics with image texture indicators. This hinders the generation of comprehensive evaluation values ​​that reflect distribution uniformity. In particular, when dispersion fluctuation trends are coupled with color signals, the system struggles to analyze the causal chain of "dispersion deterioration leading to a decrease in color saturation" or the independent pattern of "color coordinate mutations accompanied by dispersion stabilization," resulting in unclear diagnosis of the root causes of anomalies. Furthermore, fluctuations in process parameters, such as the coupling of temperature and shear fields, exacerbate signal noise, hindering the time-series correlation analysis of historical data. This leads to a lag in the generation of closed-loop adjustment commands for production line parameters, ultimately amplifying the risk of uneven color and surface defects in production batches, severely restricting product quality consistency and process optimization efficiency. Summary of the Invention

[0003] This invention provides a method for evaluating and adjusting the dispersion state of melt extrusion materials, mainly including:

[0004] Online spectral data and surface image information during the melt extrusion process are acquired. Absorbance feature values ​​are extracted from the online spectral data, and absorption peak intensities are determined. A material distribution characteristic map is generated based on the absorption peak intensities to obtain an initial quantification result of the dispersion state. Color space deviation data and particle distribution quantification indicators are separated from the surface image information to form surface quality assessment parameters. The initial quantification result and the surface quality assessment parameters are fused, and a comprehensive evaluation value reflecting distribution uniformity is generated using a preset signal cross-domain mapping method. Based on the comprehensive evaluation value and a preset process threshold, it is determined whether there are any abnormal dispersion states. The source of the abnormality is identified by combining signal time series and data change patterns. The abnormality pattern identification path is determined by analyzing the correlation between color coordinate abrupt changes and dispersion state stability. Based on the abnormality pattern identification path and historical process parameter fluctuation data, adjustment instructions for the extrusion equipment's operating status are generated and sent to the production line control system to complete closed-loop parameter adjustment. Furthermore, the acquisition of online spectral data and surface image information during the melt extrusion process, and the extraction of absorbance feature values ​​and determination of absorption peak intensities from the online spectral data, includes: real-time acquisition of the online spectral data during the melt extrusion process; extraction of absorbance feature values ​​for a specific wavelength range; analysis of the absorbance feature values ​​in the ultraviolet and visible light bands respectively, and determination of corresponding absorption peak intensities; mapping the distribution characteristics of pigment components and UV-resistant additives using the absorption peak intensities to generate a distribution characteristic map; comparison of the distribution characteristic map with a preset threshold to determine whether the material dispersion meets the standard, and obtaining an initial quantification result of the dispersion state; storing the initial quantification result as the basis for subsequent analysis, and performing correlation processing with the real-time acquired surface image information to ensure data synchronization and consistency; and through the correlation processing, initially assessing the distribution state of the material during the extrusion process, providing input basis for subsequent comprehensive evaluation. Furthermore, the step of separating color space deviation data and particle distribution quantification index from the surface image information to form surface quality assessment parameters includes: using a high-resolution image acquisition device to capture real-time images of the extruded polymer strip surface to obtain image information containing color saturation changes and surface texture distribution; separating color space deviation data representing color from the image information; monitoring the color space deviation data through a preset polymer quality standard, adjusting the deviation threshold, and obtaining color deviation analysis results; extracting particle distribution quantification index from the color deviation analysis results, and using a preset threshold to determine particle distribution uniformity; combining the particle distribution quantification index and the color space deviation data to fuse the characteristics representing color deviation and particle distribution to form the surface quality assessment parameters; and using the surface quality assessment parameters as an important input for subsequent comprehensive evaluation to ensure the comprehensiveness and accuracy of the evaluation results.Furthermore, the step of fusing the initial quantification results with the surface quality assessment parameters and generating a comprehensive assessment value reflecting the uniformity of distribution through a preset signal cross-domain mapping method includes: extracting a quantification index of surface texture distribution from the initial quantification results; fusing the quantification index with the spectral absorbance characteristics obtained from the material sample, calculating the correlation between the two through weighted averaging, and determining the dispersion fluctuation trend; for the dispersion fluctuation trend, using a preset signal cross-domain mapping method, processing the trend data through a domain transformation function to generate an intermediate value of pigment distribution uniformity; obtaining a deviation parameter of additive distribution assessment from the intermediate value, correcting it by multiplying the deviation parameter by the intermediate value, and obtaining the comprehensive assessment value; using the comprehensive assessment value to compare the uniformity of batch materials to determine the degree of difference between different batches; and using the comprehensive assessment value to provide a quantitative basis for subsequent anomaly judgment. Furthermore, the step of determining whether there is an abnormal dispersion state based on the comprehensive evaluation value and the preset process threshold, and identifying the source of the abnormality by combining the signal time series and data change patterns, includes: acquiring the signal time series and data change patterns collected by sensors during the production process; extracting time-domain features from the signal time series to obtain a quantitative index of the material dispersion state; calculating the comprehensive evaluation value using a weighted summation method for the quantitative index, where the weights are preset based on historical data; if the comprehensive evaluation value is lower than the preset process threshold, it is determined that there is an abnormal material dispersion state; acquiring the signal time series segment at the time of the abnormality, analyzing the dispersion fluctuation trend, and determining whether the abnormality is caused by the dispersion fluctuation trend; and combining the analysis of physical field coupling relationships to identify whether the abnormality is caused by the physical field coupling relationship, providing a basis for subsequent adjustments. Furthermore, the step of determining the abnormal pattern recognition path by analyzing the correlation between color coordinate mutations and dispersion stability includes: collecting color space deviation data of the current batch of extruded products to obtain saturation change data; comparing the coordinate values ​​of the standard color sample to determine the initial degree of difference between the saturation change data and the standard color sample; analyzing color coordinate mutations based on the initial degree of difference and saturation changes, and obtaining a dispersion stability index by calculating the displacement distance of the coordinate points; for the dispersion stability index, integrating the real-time calibration attributes of the extruded product batch, and extracting relevant attribute values ​​from a preset calibration database; determining the abnormal pattern recognition path based on the attribute values; and combining color change recognition, obtaining mutation correlation calculation results by matching the path with the change data, providing support for subsequent abnormal pattern analysis.Furthermore, the step of generating adjustment instructions for the extrusion equipment's operating status based on the anomaly pattern identification path and historical data of process parameter fluctuations, and sending these instructions to the production line control system to complete closed-loop parameter adjustment, includes: acquiring historical data of process parameter fluctuations; performing anomaly pattern identification on the historical data by comparing the fluctuation amplitude with a preset fluctuation range to obtain anomaly pattern results; based on the anomaly pattern results, analyzing the coupling influence path by tracing the interaction chain between parameters to determine the correlation between the path and the time series; using the path and correlation, generating adjustment instructions for the extrusion equipment's operating status by mapping the current parameter deviation of the equipment; sending the adjustment instructions through the production line control system to complete closed-loop parameter adjustment; obtaining real-time anomaly warnings from the closed-loop adjustment; and repeating path analysis if the warning exceeds a preset threshold to obtain an optimized equipment status. Furthermore, the acquisition of online spectral data and surface image information during the melt extrusion process, and the extraction of absorbance feature values ​​and determination of absorption peak intensities from the online spectral data, includes: acquiring spectral signals during the melt extrusion process in real time using an online spectrometer; extracting absorbance feature values ​​from signals within a specific wavelength range; dividing the absorbance feature values ​​into bands and analyzing the absorption peak positions and intensities in the ultraviolet and visible light ranges, respectively; generating distribution characteristic maps for pigment components and UV-resistant additives based on the absorption peak intensities; comparing the distribution characteristic maps with preset thresholds to determine whether the material dispersion state meets process requirements; converting the determination results into initial quantization values ​​as the basis for subsequent fusion analysis; and ensuring the synergy between multi-source data through the correlation between the initial quantization values ​​and surface image information. Furthermore, the step of separating color space deviation data and particle distribution quantification index from the surface image information to form surface quality assessment parameters includes: continuously photographing the surface of the extruded polymer strip using a high-resolution image acquisition device to obtain image data containing color saturation changes and surface texture distribution; separating color space deviation data representing color from the image data; monitoring the color space deviation data using a preset quality standard, dynamically adjusting the deviation threshold to obtain color deviation analysis results; extracting particle distribution quantification index from the color deviation analysis results, and using a preset threshold to determine particle distribution uniformity; combining the particle distribution quantification index and color space deviation data to generate the surface quality assessment parameters; and correlating the surface quality assessment parameters with spectral data features to provide multi-dimensional input for comprehensive evaluation.Furthermore, the process of fusing the initial quantization results with the surface quality assessment parameters to generate a comprehensive evaluation value reflecting the uniformity of distribution through a preset signal cross-domain mapping method includes: extracting a quantification index of surface texture distribution from the initial quantization results; combining the quantification index with spectral absorbance characteristics, determining the correlation between the two through weighted calculation to obtain the dispersion fluctuation trend; processing the trend data using a preset signal cross-domain mapping method and a domain transformation function to generate an intermediate evaluation value of pigment distribution uniformity; extracting a deviation parameter of additive distribution from the intermediate evaluation value, correcting the intermediate evaluation value using the deviation parameter to obtain the comprehensive evaluation value; comparing the comprehensive evaluation value with historical batch data to determine uniformity differences; and providing quantitative support for anomaly identification through the comprehensive evaluation value.

[0005] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0006] This invention discloses a method for the comprehensive evaluation and anomaly control of material dispersion and color uniformity during melt extrusion. It constructs a comprehensive evaluation system reflecting the uniformity of pigment and additive distribution by integrating spectral absorbance characteristics, surface texture distribution quantification indicators, and color space deviation data. First, this invention extracts absorption peak intensity, color saturation, and particle distribution information through online spectral and high-resolution image acquisition, and generates a comprehensive evaluation value using a signal cross-domain mapping method. When the evaluation value is lower than the process threshold, the invention analyzes the dispersion fluctuation trend and the coupling relationship with the physical field to identify abnormal patterns. Further, through time-series correlation analysis of color coordinate mutations and historical process parameters, the cause of the anomaly is determined, and equipment operation adjustment instructions are generated and sent to the production line to complete closed-loop regulation. This invention achieves real-time monitoring and dynamic optimization of material dispersion and color consistency, effectively improving the stability of extruded product quality and production efficiency. Attached Figure Description

[0007] Figure 1 This is a flowchart illustrating a method for evaluating and adjusting the dispersion state of melt extruded materials according to the present invention. Detailed Embodiments

[0008] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0009] like Figure 1 This embodiment of a method for evaluating and adjusting the dispersion state of melt extrusion materials may specifically include:

[0010] Step S101: Obtain online spectral absorbance characteristic data within a specific wavelength range during melt extrusion. By extracting absorption peak intensity information in the ultraviolet and visible light bands, conduct preliminary analysis on the distribution characteristics of pigment components and UV-resistant additives to obtain initial quantitative results of material dispersion state.

[0011] Online spectral data is acquired from the melt extrusion process, and absorbance characteristic values ​​are extracted for specific wavelength ranges. These absorbance characteristic values ​​are used to determine absorption peak intensities in the ultraviolet and visible light bands. The absorption peak intensities are then used to generate distribution characteristic maps for the pigment component distribution and the UV-resistant additive distribution. These distribution characteristic maps are generated by mapping absorption peak intensities to component location coordinates. Based on these distribution characteristic maps, a preset threshold is applied to determine the material dispersion, yielding an initial quantification value.

[0012] In one embodiment, the material during the melt extrusion process is monitored in real time using an online spectrometer to obtain absorbance characteristic data within a specific wavelength range.

[0013] Specifically, this particular wavelength range is typically set between 200 nm and 800 nm to cover the ultraviolet and visible light bands, ensuring that the typical absorption characteristics of pigment components and UV-resistant additives are captured.

[0014] For example, on a plastic extrusion production line, a spectrometer is installed at the extruder outlet to continuously collect transmission or reflectance spectral data of the melt sample, thereby forming a time-series absorbance curve. This data reflects the concentration distribution of components within the material, providing a basis for subsequent analysis. Furthermore, the characteristics of key components are quantified by extracting absorption peak intensity information in the ultraviolet and visible light bands.

[0015] It should be noted that the absorption peak intensity refers to the absorbance value at the peak of the spectral curve. Peak detection algorithms are usually used to identify the peak position, such as methods based on derivatives or curve fitting.

[0016] In one possible implementation, for pigment components, such as titanium dioxide, the absorption peak may appear around 400 nm. The intensity value is obtained by calculating the difference between the peak height and the baseline. Similarly, for UV-resistant additives, such as benzotriazole compounds, the UV absorption peak is extracted around 300 nm. This extraction process helps reveal the local concentration differences of components in the melt, avoiding the time delay problem of traditional offline testing. Preliminary analysis of the distribution characteristics of pigment components and UV-resistant additives is one of the innovative aspects of this method.

[0017] Specifically, distribution characteristic analysis involves calculating statistical indicators of absorption peak intensity, such as mean, variance, and spatial correlation, to assess the uniformity of the material.

[0018] For example, during melt extrusion, a large variance in the absorption peak intensity of the pigment component indicates uneven distribution, which may lead to inconsistent product color. By comparing peak intensity data at different time points, the dispersion state can be preliminarily quantified, such as using a dispersion index formula, where the index value is the ratio of the peak intensity standard deviation to the mean. This analysis process is particularly useful in plastic film production scenarios, enabling real-time adjustment of extrusion parameters to optimize material mixing. Preferably...

[0019] In one embodiment, the extracted absorption peak intensity information is input into a distribution model for simulation analysis.

[0020] For example, for the distribution of UV-resistant additives, the model assumes laminar flow of the material and calculates the diffusion coefficient through peak intensity gradients to obtain initial quantification results, such as a dispersion uniformity score between 0 and 1. This embodiment is applicable to the extrusion of cable sheath materials, ensuring uniform distribution of additives to improve weather resistance. In another embodiment, the analysis of pigment components can be extended to multi-band extraction, such as simultaneously monitoring multiple peaks in the visible light region to handle composite pigment systems. In this way, more comprehensive dispersion state quantification results are obtained. For example, in melt monitoring before injection molding, the analysis results show that when the dispersion index is below 0.1, it indicates that the material has reached a uniform state, which can achieve the technical effect of improving product quality.

[0021] Understandably, the results of the above preliminary analysis can be used as feedback signals and integrated into the extrusion control system to further optimize process parameters, such as screw speed or temperature settings, thereby demonstrating the versatility of the technology within the same plastics processing field.

[0022] Step S102: A high-resolution image acquisition device is used to capture real-time images of the surface of the extruded polymer strip, and image information containing color saturation changes and surface texture distribution is extracted. From this, color space deviation data for characterizing color and quantitative indicators for characterizing particle distribution are separated.

[0023] A high-resolution image acquisition device is used to capture real-time images of the surface of the extruded polymer strip. Image information including color saturation changes and surface texture distribution is extracted from the images. Color space deviation data, used to characterize color, is separated from the image information. This deviation data is monitored using a preset polymer quality standard, and the deviation threshold is adjusted to obtain color deviation analysis results. From the color deviation analysis results, particle distribution indicators are quantified. A preset threshold is used to determine the uniformity of particle distribution, thus determining the particle distribution quantification index. Combining the particle distribution quantification index and the color space deviation data, the color deviation data and particle distribution characterization are fused to form polymer strip surface quality assessment parameters, separating the color space deviation data and the particle distribution quantification index.

[0024] In one embodiment, a high-resolution image acquisition device is used to capture real-time images of the surface of the extruded polymer strip.

[0025] Specifically, the device can be an industrial-grade CCD camera, mounted near the extruder exit, capturing images at a rate of at least 30 frames per second to ensure the dynamic surface changes of the polymer strip during extrusion are captured. This real-time imaging helps to detect surface defects such as uneven color or abnormal texture in a timely manner. By adjusting the camera's focal length and lighting conditions, it can be adapted to different extrusion speeds and strip widths of different polymer materials, thereby achieving efficient monitoring in the polymer manufacturing field. Furthermore, image information containing changes in color saturation and surface texture distribution can be extracted from the captured images.

[0026] In one possible implementation, the image is first preprocessed, such as with noise reduction and contrast enhancement, and then image segmentation algorithms are used to separate the striped surface regions. Color saturation variations can be extracted using HSV color space conversion, where the saturation component represents the variation in color vibrancy; surface texture distribution is calculated using the gray-level co-occurrence matrix to determine texture features such as contrast and correlation. This extracted information provides the foundational data for subsequent separation, and when applied in polymer extrusion production lines, it can cover various scenarios involving materials such as polyethylene or polypropylene.

[0027] For example, the process of separating color space deviation data used to characterize color involves detailed calculations. Color space deviation data refers to the quantification of the difference between actual image pixels and reference colors in a standard color model.

[0028] Specifically, the image is first converted to the Lab color space, where L represents luminance and a and b represent chromaticity coordinates. Then, the Euclidean distance to a preset standard color is calculated for each pixel, quantifying the deviation, for example, by comparing the difference between the a and b values. This deviation data can reflect color variations in the polymer strip due to uneven extrusion temperatures. In polymer manufacturing, this separation method is suitable for real-time quality control, such as monitoring color consistency on continuous extrusion lines to ensure batch-to-batch uniformity. This process improves detection accuracy, avoids human error, and remains effective at different extrusion speeds.

[0029] It should be noted that the quantitative index characterizing particle distribution is obtained by further separation from texture information.

[0030] Specifically, particle distribution refers to the aggregated or dispersed state of particles that may exist on the polymer surface, and quantitative indicators may include particle density and uniformity coefficient.

[0031] For example, particle boundaries can be identified using edge detection algorithms such as Canny, and the number of particles per unit area can be calculated as a density indicator; uniformity is quantified by statistically analyzing the variance of particle spacing. This separation of indicators helps identify surface particle problems caused by uneven mixing of raw materials during the extrusion process.

[0032] In one embodiment, for rubber polymer strips, this index can be used to optimize extrusion parameters to achieve a quantitative assessment of surface quality.

[0033] Preferably, in another implementation scenario in the field of polymer extrusion, the above steps are combined to realize an integrated system. The image acquisition device is connected to the processing software to output deviation data and quantitative indicators in real time.

[0034] For example, after photographing PVC strips, the information is extracted and the data is separated for feedback control of the extruder temperature. This approach demonstrates the versatility of the technical solution, enabling it to adapt to batch or continuous production modes within the same field.

[0035] Understandably, these separated data can be used to further analyze surface quality.

[0036] For example, in one implementation, color space deviation data is combined with particle distribution indicators to form a comprehensive quality score used to determine strip quality. This analysis process brings technical benefits in polymer manufacturing, such as reducing scrap rates and improving production efficiency. In another embodiment, for high-gloss polymer strips, the image acquisition resolution is adjusted to 4K level to capture subtle saturation variations, and then threshold filtering is introduced when separating deviation data to eliminate noise. This variation enhances the flexibility of the solution and is suitable for precision extrusion applications.

[0037] Step S103: Based on the initial quantification results of the material dispersion state and the quantification index of the surface texture distribution, the spectral absorbance characteristics and dispersion fluctuation trend are integrated, and a comprehensive evaluation value reflecting the uniformity of pigment and additive distribution is generated through a preset signal cross-domain mapping method.

[0038] An initial quantification result of the material dispersion state is obtained, and a quantification index of surface texture distribution is extracted from the initial quantification result. This quantification index is then fused with the spectral absorbance characteristics obtained from the material sample, and the correlation between the two is calculated using a weighted average to determine the dispersion fluctuation trend. For this dispersion fluctuation trend, a preset signal cross-domain mapping method is used, and the trend data is processed through a domain transformation function to generate an intermediate value for pigment uniformity. A deviation parameter for evaluating the additive distribution is obtained from this intermediate value, and a comprehensive evaluation value reflecting the distribution uniformity is obtained by correcting the product of the deviation parameter and the intermediate value. This comprehensive evaluation value is used for batch material comparison to determine uniformity differences.

[0039] In one implementation, initial quantitative results of the material dispersion state and quantitative indicators of surface texture distribution are first obtained. These results can be obtained by analyzing the pigment mixture sample using an optical scanning device.

[0040] For example, high-resolution microscopy is used to capture images of the material surface, and then image processing algorithms are applied to calculate statistical values ​​of the dispersion state, such as particle density distribution. Quantification of surface texture distribution involves texture analysis techniques, calculating parameters such as the gray-level co-occurrence matrix to quantify texture roughness and uniformity. This initial quantification aids subsequent fusion, providing fundamental data on material distribution. Further, spectral absorbance characteristics are fused with dispersion fluctuation trends. Spectral absorbance characteristics refer to the absorption values ​​of pigments and additives at different wavelengths measured using a spectrophotometer, forming a feature vector to reflect the chemical composition distribution of the material. Dispersion fluctuation trends are analyzed through time-series analysis of dispersion changes during the material mixing process.

[0041] For example, monitoring the particle size variation rate over the mixing time generates a fluctuation curve. This fusion process emphasizes the integration of multimodal data to ensure the comprehensiveness of the assessment.

[0042] Preferably, a preset signal cross-domain mapping method is used to generate the comprehensive evaluation value. The signal cross-domain mapping method is a technique that maps features from different domains to a unified space.

[0043] For example, spectral features can be mapped from the frequency domain to the spatial domain and combined with dispersed state results.

[0044] Specifically, this method first standardizes all input features, such as normalizing quantification metrics, and then uses a linear transformation matrix to project the feature vectors onto a common evaluation dimension. This mapping allows for the quantification of the uniformity of pigment and additive distribution.

[0045] For example, the variance of the mapped vector can be used as a homogeneity index. The core of this method lies in the accuracy of cross-domain transformation, which can handle heterogeneity among features and improve the reliability of the evaluation.

[0046] One possible implementation considers applications in paint production scenarios.

[0047] For example, in the mixing process of water-based coatings, the initial quantification results are obtained from an online scanning system, capturing the distribution of pigment particles in the matrix. Surface texture distribution is acquired using a laser scanner, and texture entropy is calculated as an indicator. During fusion, spectral absorbance characteristics are obtained from a UV-Vis spectrometer, and the dispersion fluctuation trend is analyzed based on sensor data from the stirring equipment to determine the fluctuation amplitude. In this scenario, the signal cross-domain mapping method uses a preset mapping function to convert these features into a uniformity score between 0 and 1, with a higher score indicating a more uniform distribution. This implementation demonstrates the adaptability of the technology in actual production.

[0048] It should be noted that the prerequisite for the cross-domain signal mapping method includes selecting a suitable mapping function.

[0049] For example, principal component analysis can be used as a basis to reduce the dimensionality of high-dimensional features before mapping, thereby reducing computational complexity. In the evaluation of pigment and additive uniformity, this method can effectively capture distribution biases.

[0050] For example, when additives aggregate, the mapping values ​​will show high fluctuations, thus guiding the adjustment of mixing parameters.

[0051] Specifically, in another embodiment, for the evaluation of oil-based pigments, the initial quantification results are obtained through electron microscopy image processing to quantify the particle spacing distribution. The surface texture distribution index is calculated as the average surface roughness. When fusing spectral absorbance features, infrared spectral data is considered, and the dispersion fluctuation trend analysis is used to determine the dispersion curve under varying mixing temperatures. A preset cross-domain signal mapping method employs nonlinear mapping, such as kernel function transformation, to fuse features and generate a comprehensive value. This approach expands the application scope of the technology, making it suitable for different pigment types.

[0052] For example, in a laboratory setting for pigment dispersion assessment, the generated comprehensive evaluation value can be used for quality control. By comparing the evaluation values ​​of different batches, the sources of uneven distribution can be identified, such as insufficient stirring speed. This objective assessment helps optimize the production process and achieve more stable material uniformity. Furthermore, the versatility of this technical solution is reflected in the adjustable mapping parameters.

[0053] For example, in one implementation, the mapping weights are adjusted according to the material viscosity to enhance the sensitivity to additive distribution, thereby covering a variety of mixing conditions in the coatings industry.

[0054] In one embodiment, the entire process begins with data acquisition and ends with fusion and mapping, ensuring logical coherence. The effect is to provide quantitative data to support the optimization of uniform distribution of pigments and additives, without relying on subjective judgment.

[0055] Step S104: If the comprehensive evaluation value is lower than the preset process threshold, it is determined that there is an abnormal material dispersion state in the production process. By combining the signal time series and data change pattern, it is identified whether the abnormality is caused by the dispersion fluctuation trend or by other physical field coupling relationships.

[0056] The process involves acquiring the time series signals and data change patterns collected by sensors during production. By extracting time-domain features from the signal time series, a quantitative index of material dispersion is obtained. For this quantitative index, a weighted summation method is used to calculate a comprehensive evaluation value, where weights are preset based on historical data. If the comprehensive evaluation value is lower than a preset process threshold, an abnormal material dispersion state is identified during production, and a signal time series segment at the time of the abnormality is acquired. The dispersion fluctuation trend is analyzed using this signal time series segment to determine whether the abnormality is caused by this dispersion fluctuation trend. Finally, based on the dispersion fluctuation trend and physical field coupling relationship analysis, it is identified whether the abnormality is caused by the physical field coupling relationship.

[0057] In one implementation, when the comprehensive evaluation value is lower than a preset process threshold, the system determines that there is an abnormal material dispersion state in the production process.

[0058] Specifically, the comprehensive evaluation value is an indicator calculated by integrating multiple production parameters, such as the weighted average of material particle size distribution, stirring speed, and temperature changes. The process threshold is set based on historical production data and experience; for example, it is set to 0.8 in chemical stirring processes to ensure material uniformity. If the evaluation value is lower than this threshold, it indicates that the dispersion process may deviate from normal conditions, thus triggering the anomaly detection mechanism. This detection helps to promptly identify production deviations and avoid product quality problems. Furthermore, the causes of anomalies are identified by combining signal time series and data change patterns. The signal time series refers to the continuous data stream collected from sensors, such as the time-series signal of material flow during stirring recorded by a vibration sensor. The data change pattern involves the statistical characteristics of these signals, such as mean, variance, or trend lines.

[0059] In one possible implementation, the system first extracts the fluctuation amplitude of the time series. If the amplitude exceeds the normal range, it is determined that the anomaly may be caused by a dispersion fluctuation trend. Here, the dispersion fluctuation trend refers to the change in the uneven distribution of material particles during stirring over time, such as the increase in local density caused by particle aggregation. By calculating the autocorrelation function of the series, the persistence of this trend can be quantified, thereby distinguishing between short-term fluctuations and long-term anomalies.

[0060] It should be noted that the process of identifying dispersion fluctuation trends includes segmented analysis of the time series.

[0061] Specifically, the signal sequence is divided into multiple time windows, for example, each lasting 10 seconds, and then a dispersion index, such as the standard deviation of particle diameter, is calculated within each window. If the standard deviation shows an upward trend across consecutive windows, the anomaly is confirmed to originate from dispersion fluctuations. This method is commonly used in material mixing production, such as in pharmaceutical processes to monitor powder dispersion and ensure drug homogeneity. Through this analysis, the system can accurately pinpoint anomalies caused by fluctuations, providing data support for adjusting mixing parameters.

[0062] In one embodiment, if the anomaly is not entirely caused by the dispersion fluctuation trend, the possibility of other physical field coupling relationships is considered. Physical field coupling relationships refer to the mutual influence of multiple physical factors, such as the interaction between the temperature field and the fluid field, which causes changes in material viscosity at high temperatures and thus affects dispersion.

[0063] Preferably, the system identifies this coupling through multivariate correlation analysis, such as calculating the cross-correlation coefficient between temperature and vibration signals. If the coefficient is higher than a preset value, such as 0.7, the anomaly is determined to originate from the coupling relationship. This identification enhances the accuracy of diagnosis and avoids misjudgment based on a single factor in actual production.

[0064] For example, in the production scenario of a chemical reactor, a time series signal might show a sudden rise in temperature accompanied by increased vibration. In this case, data change pattern analysis can reveal a coupling effect. The specific process includes acquiring real-time data, constructing a vector model where each vector element represents a physical field parameter, and then extracting the dominant pattern through principal component analysis. If the dominant pattern shows that temperature dominates the vibration change, then the coupling is confirmed to be causing the anomaly. This method is applicable to various scenarios in similar fields, such as the coating dispersion process, ensuring the comprehensiveness of anomaly identification.

[0065] Understandably, the output of the aforementioned identification process is used to guide production adjustments. For example, if the anomaly is caused by dispersion fluctuations, the stirring time is increased; if it is caused by coupling relationships, temperature control is optimized. This technical solution enables refined management of anomalies in industrial production, improving process stability. Furthermore, in another implementation, the system can integrate machine learning models to assist in identification.

[0066] For example, a support vector machine is used to classify signal time series, with input features including spectral characteristics of data variation patterns. If the model output shows a high-probability fluctuation trend class, then the dispersion problem is addressed first. This extension enhances the flexibility of the approach while remaining applicable to areas of material dispersion.

[0067] In one embodiment, wavelet transform is used to decompose the signal time series for extracting data change patterns.

[0068] Specifically, wavelet transform decomposes a sequence into components of different frequencies. Low-frequency components are used to capture long-term trends, while high-frequency components identify sudden fluctuations. By comparing these components with historical patterns, the causes of anomalies can be determined more accurately. For example, in the dispersion of plastic particles, high-frequency noise is detected as a result of coupling caused by equipment failure.

[0069] Preferably, the entire process is implemented through a real-time monitoring platform, with a signal acquisition frequency set to 10 times per second to ensure timely capture of data change patterns. This setup is common in production environments and provides a reliable basis for anomaly identification.

[0070] Step S105: Calculate the degree of difference between the current batch of extruded products and the standard color sample based on color space deviation and color saturation changes. By analyzing the correlation between color coordinate mutations and dispersion stability, determine the specific path for abnormal pattern recognition.

[0071] By collecting color space deviation data of the current batch of extruded products, saturation change data is obtained and compared with the coordinate values ​​of a standard color sample to determine the initial degree of difference between the saturation change and the standard color sample. Based on the initial degree of difference and saturation change, color coordinate abrupt changes are analyzed, and the dispersion state stability index corresponding to the color coordinate abrupt change is obtained by calculating the displacement distance of the coordinate points. For the dispersion state stability index, real-time calibration attributes of the extruded product batch are integrated, and relevant attribute values ​​are extracted from a preset calibration database to determine the abnormal pattern recognition path. The abnormal pattern recognition path is obtained, and combined with color change recognition, the abrupt change correlation calculation result is obtained by matching the path with the change data.

[0072] In one implementation, for color quality control of extruded products, color data of the current batch of products is first collected.

[0073] Specifically, a spectrophotometer is used to scan the surface of the extruded plastic pipe to obtain its L*a*b* coordinate values in the CIE Lab color space, where L* represents brightness, a* represents the red-green axis, and b* represents the yellow-blue axis. The selection of this color space helps to quantify the deviation because it is close to human eye perception. By comparing these coordinates with the corresponding values of the standard color sample, the color space deviation is calculated. For example, the ΔE value is calculated using the Euclidean distance formula, that is, √[(ΔL*)² + (Δa*)² + (Δb*)²], so as to evaluate the overall color difference. Further, the change in color saturation is analyzed. On the extrusion production line, the saturation can be represented by the S value in the HSV model through color coordinate conversion. The higher the S value, the purer the color. For the current batch of products, monitor the change of saturation over time or position. For example, sample at different segments of the pipe and calculate the standard deviation of the S value to quantify the degree of change. This change is related to the uniformity of raw material mixing. If the saturation fluctuation exceeds the threshold, it may indicate unstable temperature of the extruder. Based on the above data, calculate the difference degree between the extruded products of the current batch and the standard color sample. The specific process includes averaging the color coordinates of multiple sampling points to form a representative value of the batch, and then comparing it with the standard color sample.

[0074] In a possible implementation, for a batch of extruded cable sheaths, the ΔE values of 10 sampling points are collected. If the average ΔE is less than 2, it is considered qualified; otherwise, it is marked as a deviation batch. This calculation ensures an objective quantification of the difference degree and supports production adjustment. When analyzing the correlation between the mutation of color coordinates and the stability of the dispersion state, the dynamic characteristics of the coordinate sequence need to be examined. The stability of the dispersion state refers to the coefficient of variation of color values among sampling points. For example, calculate the ratio of the standard deviation to the mean of the a* and b* axes. The mutation is identified by detecting the jump points in the coordinate sequence. For example, use a sliding window to compare the ΔE of adjacent points. If it exceeds the set threshold, it is regarded as a mutation. The correlation analysis can be carried out by calculating the correlation coefficient. For example, the Pearson correlation coefficient is used to evaluate the correlation degree between the mutation frequency and the stability index. In the scenario of extruded products, this correlation reveals the influence of process parameters such as screw speed on color uniformity. If the correlation coefficient is higher than 0.8, it indicates that the mutation directly leads to a decrease in stability, thus guiding the troubleshooting of abnormalities.

[0075] Preferably, the specific path for abnormal pattern recognition is achieved by constructing a decision tree model.

[0076] Specifically, first extract features from historical data, including the deviation ΔE, the standard deviation of saturation, the mutation frequency, and the stability correlation coefficient. Then, train the decision tree, where the root node branches based on the ΔE threshold, and the child nodes consider the correlation index. For example.

[0077] In one embodiment, for extruded sheet products, if ΔE > 3 and the correlation coefficient > 0.7, the path points to the "raw material contamination" mode; if the stability is low but the mutations are few, it points to the "temperature fluctuation" mode. This path provides clear anomaly diagnosis and supports rapid response.

[0078] It should be noted that, in another implementation, this can be extended to scenarios with different extrusion speeds.

[0079] For example, for high-speed extruded cables, a real-time monitoring module can be added to dynamically adjust path thresholds to adapt to production variations. This flexibility ensures the universal application of the technical solution within the same field.

[0080] For example, in actual extruded pipe production, after identifying deviations caused by saturation changes using the above methods, the dye addition ratio can be adjusted to improve color consistency. This helps reduce scrap rates and improve product quality.

[0081] It is understood that the above embodiments cover the core features of the claims, provide specific implementations in various scenarios, and ensure the practicality of the solution.

[0082] Step S106: Based on the results of abnormal pattern recognition and historical data of process parameter fluctuations, analyze the correlation between the coupling influence path and the time series, generate adjustment instructions for the operating status of the extrusion equipment, and send them to the production line control system to complete the closed-loop adjustment of parameters.

[0083] Historical data on process parameter fluctuations are acquired. Anomaly patterns are identified by comparing the fluctuation amplitude with a preset fluctuation range, yielding anomaly pattern results. Based on these results, the coupling influence path is analyzed by tracing the interaction chain between parameters, determining the correlation between the path and the time series. Using this path and correlation, adjustment commands are generated for the extrusion equipment's operating status by mapping the current parameter deviation. These adjustment commands are sent through the production line control system to complete closed-loop parameter adjustment. Real-time anomaly warnings are obtained from the closed-loop adjustment. If the warning exceeds a preset threshold, the path analysis is repeated to obtain an optimized equipment state.

[0084] In one implementation, the method first acquires the results of abnormal pattern recognition and historical data on process parameter fluctuations.

[0085] Specifically, abnormal pattern recognition is obtained by processing real-time monitoring data of the extrusion equipment.

[0086] For example, when the extruder's temperature sensor detects fluctuations exceeding a threshold, the system flags it as a temperature anomaly. These results, combined with historical data such as screw speed, melt pressure, and cooling water flow rate records from the past week, form the basis for analysis. This ensures the integrity of the data input and provides reliable support for subsequent analysis. Furthermore, analyzing coupled influence paths involves examining the interaction paths between different process parameters. A coupled influence path refers to how a change in one parameter affects other parameters through a chain reaction.

[0087] For example, increasing the screw speed may lead to an increase in melt temperature, which in turn affects the thickness uniformity of the extruded product.

[0088] In one possible implementation, the analysis employs a causal graph model, treating parameters as nodes and paths as edges. Edge weights are calculated using historical data to map the chain of influence. This path analysis helps reveal hidden equipment problems, such as anomalous amplification caused by the coupling of rotational speed and pressure.

[0089] It should be noted that time series correlation analysis is based on statistical processing of historical data.

[0090] Specifically, time series correlation examines the strength of the association between parameters at different points in time.

[0091] For example, the autocorrelation function is used to calculate the autocorrelation coefficient of the melt pressure sequence to determine whether the fluctuations are periodic; simultaneously, the cross-correlation function is used to assess the correlation between the temperature and rotational speed sequences, such as the effect of a 1-hour temperature lag on rotational speed. This analysis process includes data smoothing and correlation coefficient calculation to ensure the identification of the relationship between long-term trends and short-term fluctuations, thereby providing data support for adjustments.

[0092] For example, during the operation of the extrusion equipment, adjustment instructions for the operating status are generated based on the aforementioned analysis results.

[0093] For example, if the coupling path shows that temperature fluctuations affect product quality through the pressure path, the system will generate instructions such as reducing the engine speed by 10% and increasing the cooling water flow. This instruction generation uses a rule engine, triggering the adjustment logic when the correlation threshold is set above 0.8, ensuring the instruction is targeted.

[0094] Preferably, the adjustment command is sent to the production line control system to complete the closed-loop adjustment of parameters.

[0095] In one embodiment, instructions are transmitted to the PLC controller via industrial Ethernet, and the controller performs adjustments in real time and provides feedback on actual parameter changes, forming a closed loop.

[0096] For example, on a plastic pipe extrusion line, when a thickness deviation is detected, the system sends an instruction to adjust the traction speed, monitors the feedback, and verifies the effect to achieve stable operation.

[0097] It can be understood that this method can be applied to different extrusion scenarios, such as film extrusion or profile extrusion. In film extrusion, the analysis focuses on the coupling path between the air gap height and the wind ring wind speed, and generates an instruction to optimize the wind speed to reduce film bubble instability; while in profile extrusion, the time series correlation between the die temperature and the extrusion speed is emphasized, and the instruction adjusts the temperature to maintain shape consistency. These scenarios demonstrate the generality of the method. Further, in another embodiment, the coupling influence path analysis can be extended to multi-device linkage.

[0098] For example, in a continuous production line, analyze the path between the parameters of the upstream feeder and the pressure of the extruder, calculate the correlation to generate a joint instruction, such as synchronously reducing the feeding speed to avoid overload. This extension enhances the adaptability of the system.

[0099] Specifically, the process of time series correlation analysis includes data segmentation and trend extraction.

[0100] For example, divide the historical data into day and night segments, calculate the correlation coefficient of the parameters within each segment. If it is found that the influence of humidity at night on the temperature sequence is stronger, then the instruction preferentially adjusts the humidity control at night. This segmented analysis improves the accuracy.

[0101] In one embodiment, the effect of the entire adjustment process is reflected in the improvement of equipment stability.

[0102] For example, through closed-loop adjustment, the abnormal occurrence rate of the extruder can be reduced to ensure production continuity. This method realizes parameter optimization through objective steps and supports the efficient operation of the production line.

[0103] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for evaluating and adjusting the dispersion state of melt extruded materials, characterized in that, include: Online spectral data and surface image information are acquired during the melt extrusion process. Absorbance feature values ​​are extracted from the online spectral data and the absorption peak intensity is determined. A material distribution characteristic map is generated based on the absorption peak intensity to obtain the initial quantification result of the dispersion state. Color space deviation data and particle distribution quantization index are separated from the surface image information to form surface quality assessment parameters; the initial quantization results and the surface quality assessment parameters are fused together, and a comprehensive assessment value reflecting the distribution uniformity is generated by a preset signal cross-domain mapping method; Based on the comprehensive evaluation value and the preset process threshold, it is determined whether there is an abnormal dispersion state. The source of the abnormality is identified by combining the signal time series and data change patterns. The abnormality pattern identification path is determined by analyzing the correlation between color coordinate mutation and dispersion state stability. Based on the abnormality pattern identification path and historical data of process parameter fluctuations, adjustment instructions for the operating status of the extrusion equipment are generated and sent to the production line control system to complete the closed-loop adjustment of parameters.

2. The method for evaluating and adjusting the dispersion state of melt extruded materials as described in claim 1, characterized in that, The acquisition of online spectral data and surface image information during the melt extrusion process, and the extraction of absorbance feature values ​​and determination of absorption peak intensities from the online spectral data, includes: real-time acquisition of the online spectral data during the melt extrusion process, and extraction of absorbance feature values ​​for specific wavelength ranges; analysis of the absorbance feature values ​​in the ultraviolet and visible light bands respectively, and determination of corresponding absorption peak intensities; mapping the distribution characteristics of pigment components and UV-resistant additives using the absorption peak intensities to generate a distribution characteristic map; comparing the distribution characteristic map with a preset threshold to determine whether the material dispersion meets the standard, and obtaining an initial quantification result of the dispersion state; storing the initial quantification result as the basis for subsequent analysis, and performing correlation processing with the real-time acquired surface image information to ensure data synchronization and consistency; and through the correlation processing, initially assessing the distribution state of the material during the extrusion process, providing input basis for subsequent comprehensive evaluation.

3. The method for evaluating and adjusting the dispersion state of melt extruded materials as described in claim 1, characterized in that, The step of separating color space deviation data and particle distribution quantification index from the surface image information to form surface quality assessment parameters includes: using a high-resolution image acquisition device to capture real-time images of the extruded polymer strip surface, obtaining image information containing color saturation changes and surface texture distribution; separating color space deviation data representing color from the image information; monitoring the color space deviation data using a preset polymer quality standard, adjusting the deviation threshold, and obtaining color deviation analysis results; extracting particle distribution quantification index from the color deviation analysis results, and using a preset threshold to determine particle distribution uniformity; combining the particle distribution quantification index and the color space deviation data, fusing the characteristics representing color deviation and particle distribution to form the surface quality assessment parameters; and using the surface quality assessment parameters as an important input for subsequent comprehensive evaluation to ensure the comprehensiveness and accuracy of the evaluation results.

4. The method for evaluating and adjusting the dispersion state of melt extruded materials as described in claim 1, characterized in that, The process of integrating the initial quantification results with the surface quality assessment parameters to generate a comprehensive evaluation value reflecting the uniformity of distribution through a preset signal cross-domain mapping method includes: extracting a quantification index of surface texture distribution from the initial quantification results; integrating the quantification index with spectral absorbance characteristics obtained from material samples, calculating the correlation between the two through weighted averaging, and determining the dispersion fluctuation trend; for the dispersion fluctuation trend, using a preset signal cross-domain mapping method, processing the trend data through a domain transformation function to generate an intermediate value of pigment distribution uniformity; obtaining a deviation parameter for the additive distribution assessment from the intermediate value, correcting it by multiplying the deviation parameter by the intermediate value, and obtaining the comprehensive evaluation value; using the comprehensive evaluation value to compare the uniformity of batch materials to determine the degree of difference between different batches; and providing a quantitative basis for subsequent anomaly judgment through the comprehensive evaluation value.

5. The method for evaluating and adjusting the dispersion state of melt extruded materials as described in claim 1, characterized in that, The step of determining whether an abnormal dispersion state exists based on the comprehensive evaluation value and a preset process threshold, and identifying the source of the abnormality by combining signal time series and data change patterns, includes: acquiring signal time series and data change patterns collected by sensors during the production process; extracting time-domain features from the signal time series to obtain a quantitative index of the material dispersion state; calculating the comprehensive evaluation value using a weighted summation method for the quantitative index, where the weights are preset based on historical data; if the comprehensive evaluation value is lower than the preset process threshold, it is determined that an abnormal material dispersion state exists; acquiring a signal time series segment at the time of the abnormality, analyzing the dispersion fluctuation trend, and determining whether the abnormality is caused by the dispersion fluctuation trend; and combining physical field coupling relationship analysis to identify whether the abnormality is caused by the physical field coupling relationship, providing a basis for subsequent adjustments.

6. The method for evaluating and adjusting the dispersion state of melt extruded materials as described in claim 1, characterized in that, The method for determining the abnormal pattern recognition path by analyzing the correlation between color coordinate mutations and dispersion stability includes: collecting color space deviation data of the current batch of extruded products to obtain saturation change data; comparing the coordinate values ​​of a standard color sample to determine the initial degree of difference between the saturation change data and the standard color sample; analyzing color coordinate mutations based on the initial degree of difference and saturation changes, and obtaining a dispersion stability index by calculating the displacement distance of the coordinate points; for the dispersion stability index, integrating the real-time calibration attributes of the extruded product batch, and extracting relevant attribute values ​​from a preset calibration database; determining the abnormal pattern recognition path based on the attribute values; and combining color change recognition, obtaining mutation correlation calculation results by matching the path with the change data, providing support for subsequent abnormal pattern analysis.

7. The method for evaluating and adjusting the dispersion state of melt extruded materials as described in claim 1, characterized in that, The step of generating adjustment instructions for the extrusion equipment's operating status based on the anomaly pattern identification path and historical data of process parameter fluctuations, and sending these instructions to the production line control system to complete closed-loop parameter adjustment, includes: acquiring historical data of process parameter fluctuations; identifying anomaly patterns by comparing the fluctuation amplitude with a preset fluctuation range to obtain anomaly pattern results; analyzing the coupling influence path by tracing the interaction chain between parameters based on the anomaly pattern results to determine the correlation between the path and the time series; using the path and correlation, generating adjustment instructions for the extrusion equipment's operating status by mapping the current parameter deviations of the equipment; sending the adjustment instructions through the production line control system to complete closed-loop parameter adjustment; obtaining real-time anomaly warnings from the closed-loop adjustment; and repeating path analysis to obtain an optimized equipment status if the warning exceeds a preset threshold.

8. The method for evaluating and adjusting the dispersion state of melt extruded materials as described in claim 1, characterized in that, The acquisition of online spectral data and surface image information during the melt extrusion process, and the extraction of absorbance feature values ​​and determination of absorption peak intensities from the online spectral data, includes: acquiring spectral signals during the melt extrusion process in real time using an online spectrometer, and extracting absorbance feature values ​​from signals within a specific wavelength range; dividing the absorbance feature values ​​into bands, and analyzing the absorption peak positions and intensities in the ultraviolet and visible light ranges respectively; generating distribution characteristic maps for pigment components and UV-resistant additives based on the absorption peak intensities; comparing the distribution characteristic maps with preset thresholds to determine whether the material dispersion state meets process requirements; converting the determination results into initial quantization values ​​as the basis for subsequent fusion analysis; and ensuring the synergy between multi-source data through the correlation between the initial quantization values ​​and surface image information.

Images