Material extrusion tension control system, method based on immersion image and surface image analysis

Abstract

The application provides a material extrusion tension control system and method based on immersion image and surface image analysis, relates to the technical field of composite material pultrusion processes, and comprises the following steps: collecting the fiber bundle out-slot path and the formed profile discharge path in real time to obtain fiber bundle immersion state images and formed profile surface texture images, extracting immersion state characteristic parameters and formed surface characteristic parameters from the images respectively, generating a tension control signal based on qualified characteristic parameters, combining the immersion state characteristic parameters and the formed surface characteristic parameters, adjusting the fiber bundle out-slot path and the formed profile discharge path, sampling and monitoring the output formed products, and optimizing the tension control signal according to the sampling and monitoring results, so as to prevent defects such as immersion failure and surface wrinkles from the root, reduce the product rejection rate, and improve the production efficiency and quality.

Description

Technical Field

[0001] This invention relates to the field of composite material pultrusion technology, and in particular to a material extrusion molding tension control system and method based on impregnation image and surface image analysis. Background Technology

[0002] Pultruded composite materials, due to their outstanding advantages such as light weight, high strength, and strong corrosion resistance, are widely used in many industrial fields such as insulating strips and building profiles. Their production quality directly determines the reliability of the end products. Online real-time monitoring and closed-loop process control are core aspects of ensuring the quality of pultruded products. Especially under continuous production conditions, it is necessary to quickly identify potential defects and provide timely feedback and adjustments to avoid batch product scrapping.

[0003] However, existing pultrusion product defect detection and control technologies have significant limitations: for defects such as poor fiber impregnation (dry spots), destructive testing methods involving offline cutting and sampling are often used, which cannot provide real-time feedback on production status, resulting in batches of defective products being produced by the time the defect is discovered; for molding defects such as surface wrinkles, manual visual inspection is relied upon, which is not only inefficient but also susceptible to subjective human factors, leading to a high rate of missed detections and significant labor costs. More importantly, traditional detection technologies lack a real-time linkage and control mechanism with the production process, making it difficult to quickly trace the root cause and make targeted adjustments even when defects are detected, ultimately resulting in a persistently high product scrap rate.

[0004] There is an urgent need for a tension control system that can capture production process characteristics in real time, accurately identify potential defects, and dynamically link with process parameters to achieve proactive defect prevention and efficient control, thereby improving the production quality and efficiency of pultruded products.

[0005] Therefore, the present invention provides a material extrusion molding tension control system and method based on immersion image and surface image analysis. Summary of the Invention

[0006] This invention relates to a material extrusion molding tension control system and method based on impregnation image and surface image analysis. It acquires image data in real time through machine vision, accurately extracts feature parameters, and combines automatic tension control and closed-loop feedback mechanism to prevent defects such as poor impregnation and surface wrinkles from the root, thereby reducing product scrap rate and improving production efficiency and quality.

[0007] This invention provides a material extrusion molding tension control system based on immersion image and surface image analysis, comprising: The vision acquisition module is used to acquire the fiber bundle exit path and the molding profile exit path in real time, and obtain the corresponding fiber bundle impregnation state image and molding profile surface texture image. The feature extraction module is used to extract the surface texture of the fiber bundle impregnation state image and the surface texture image of the molded profile, respectively, to obtain the corresponding impregnation state feature parameters and molded surface feature parameters; The tension control module is used to generate a tension control signal based on qualified characteristic parameters combined with the immersion state characteristic parameters and the forming surface characteristic parameters, and to adjust the fiber bundle exit path and the forming profile exit path. The monitoring and feedback module is used to acquire the output molded products from the material discharge path of the molded profile, perform sampling monitoring on the output molded products, and optimize the tension control signal based on the sampling monitoring results.

[0008] In one implementable manner, the visual acquisition module includes: A first industrial camera is used to acquire first real-time video information of the fiber bundle exit path and generate a corresponding fiber bundle impregnation state image. The second industrial camera is used to collect real-time video information of the material discharge path of the molded profile and generate a corresponding surface texture image of the molded profile. The network access monitoring unit is used to connect the first industrial camera and the second industrial camera to a preset Ethernet, and to acquire the corresponding first image data delay feature and second image data transmission delay feature, respectively. When the delay feature of the first image data is less than 50ms or the delay feature of the second image data transmission is less than 50ms, the transmission of the first industrial camera or the second industrial camera is optimized.

[0009] In one implementable manner, the feature extraction module includes: The noise identification unit is used to perform noise identification on the fiber bundle impregnation state image and the molded profile surface texture image respectively, to obtain a first brightness defect and a first on-site interference in the fiber bundle impregnation state image, and to obtain a second brightness defect and a second on-site interference in the molded profile surface texture image. An image preprocessing unit is used to perform noise reduction processing on the fiber bundle impregnation state image based on the first brightness defect and the first field interference, and to perform noise reduction processing on the surface texture image of the molded profile based on the second brightness defect and the second field interference. The impregnation feature recognition unit is used to perform region separation on the noise-reduced fiber bundle impregnation state image, remove the background region to obtain the fiber bundle region of the noise-reduced fiber bundle impregnation state image, and calculate the gray mean and standard deviation of the fiber bundle region to obtain the impregnation state feature parameters. The forming surface feature recognition unit is used to perform edge detection on the surface texture image of the denoised forming profile. Based on the surface contour information and texture detail information obtained from the edge detection, it calculates several surface wrinkles of the forming product and the wrinkle length and wrinkle area ratio corresponding to each surface wrinkle, and obtains the forming surface feature parameters.

[0010] One feasible approach also includes: The defect identification unit is used to compare the number of surface wrinkles and the wrinkle length and wrinkle area ratio of each surface wrinkle using a preset threshold for the number of surface wrinkles, a threshold for the length of a single wrinkle, and the wrinkle area ratio, and generate a threshold comparison report. When the threshold comparison report contains comparison results that do not meet the standards, it is determined that the molded product has surface defects.

[0011] In one implementable manner, the tension control module includes: The parameter processing unit is used to normalize the immersion state characteristic parameters and the molding surface characteristic parameters to obtain several quantized scores within a preset value range, and to perform deviation analysis between each quantized score and the preset standard score to obtain several deviation parameters. The signal control unit is used to deduce the total required tension of the fiber bundle outlet path and the molding profile outlet path based on the deviation parameter, and to correct the original tension control signal using the total required tension to obtain the tension control signal. The process adjustment unit is used to perform multiple process adjustment simulations on the fiber bundle outlet path and the molded profile outlet path using the tension control signal with a single adjustment amount of 5V, until the total required tension is met.

[0012] One feasible approach also includes: The threshold setting unit is used to construct standard immersion state data and standard surface wrinkle data based on the image feature data corresponding to the qualified molded product, and at the same time obtain the first process adjustment amount corresponding to the fiber bundle exit path and the second process adjustment amount corresponding to the molded profile exit path. The tension setting unit is used to take 1% of the first adjustment amount as the first tension adjustment step threshold of the fiber bundle outlet path and 1% of the second adjustment amount as the second tension adjustment step threshold of the molded profile outlet path. The control execution unit is used to adjust the tension of the fiber bundle exit path based on the first tension adjustment step size threshold when the impregnation state characteristic parameters are inconsistent with the standard impregnation state data, and to adjust the tension of the molded profile exit path based on the second tension adjustment step size threshold when the molding surface characteristic parameters are inconsistent with the standard surface wrinkle data.

[0013] In one implementable manner, the supervision feedback module includes: The sampling and testing unit is used to obtain the output molded products of the molded profile discharge path, sample the output molded products at a specified ratio, obtain a number of sampled products, and perform product testing. Determine whether the current tension meets the acceptable standard based on the test results; An optimization execution unit is used to generate a tension optimization reference based on the information difference between the test results and the pass standard if the performance does not meet the requirements, and then feed it back to the tension control module to optimize the tension control signal.

[0014] In one implementable manner, the supervision feedback module includes: The real-time analysis unit is used to acquire several consecutive frames of images during the process adjustment, derive the quantitative feature value of the process adjustment, and stop the tension adjustment and maintain the current parameters when the quantitative feature value is consistent with the standard quantitative threshold. The continuous adjustment unit is used to continuously acquire several consecutive frames of images and perform threshold comparison when the quantized feature value is inconsistent with the standard threshold. The warning and alert unit is used to determine that the process adjustment has failed and issue an abnormal warning when the number of threshold comparisons exceeds the specified number.

[0015] This invention proposes a method for controlling the tension in material extrusion molding based on immersion image and surface image analysis, including: Step 1: Real-time acquisition of fiber bundle exit path and molded profile exit path to obtain corresponding fiber bundle impregnation state images and molded profile surface texture images; Step 2: Extract the surface texture of the fiber bundle impregnation state image and the molded profile surface texture image respectively to obtain the corresponding impregnation state feature parameters and molded surface feature parameters; Step 3: Based on the qualified characteristic parameters, combined with the impregnation state characteristic parameters and the forming surface characteristic parameters, a tension control signal is generated to adjust the fiber bundle exit path and the forming profile exit path. Step 4: Obtain the output molded product from the material discharge path of the molded profile, perform sampling monitoring on the output molded product, and optimize the tension control signal based on the sampling monitoring results.

[0016] The beneficial effects achievable by this invention are as follows: To improve the pass rate and production efficiency of pultruded products, reduce labor costs and raw material losses, and achieve strong compatibility, industrial cameras are first precisely deployed at key locations to achieve real-time, high-quality acquisition of fiber bundle impregnation status and surface texture of the formed profile. The image transmission latency is low and the anti-interference capability is strong, avoiding detection errors and control failures caused by data distortion. Then, professional image processing algorithms are used to perform noise reduction, segmentation, and edge detection on the acquired images, accurately extracting quantified impregnation status feature parameters and formed surface feature parameters. This converts visual information into digital signals that can be used for control, avoiding the subjectivity and missed detection rate of manual inspection, and significantly improving the accuracy and efficiency of defect identification. Furthermore, based on the pass feature parameters and the real-time extracted quantified... The system generates targeted tension control signals based on characteristic parameters, accurately selects control targets according to defect types, and avoids sudden tension changes through step size control. This achieves real-time, targeted, and stable process adjustments, preventing defects such as poor impregnation and surface wrinkles from the root. Finally, by sampling and monitoring the output molded products and tracking the process adjustment in real time, a closed-loop optimization mechanism is formed. This not only verifies the actual effect of tension control but also dynamically optimizes the control signals based on monitoring results, ensuring that the system continuously adapts to changes in production conditions and further improving the stability of product quality. At the same time, the abnormal early warning mechanism reduces the risk of batch scrapping. In this way, it can achieve defect prevention and stable optimization of production processes, solving problems such as delayed defect detection, high dependence on manual labor, and lack of targeted control.

[0017] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the material extrusion molding tension control system based on immersion image and surface image analysis in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the workflow of the material extrusion molding tension control method based on immersion image and surface image analysis in an embodiment of the present invention. Detailed Implementation

[0020] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0021] Example 1: This example provides a material extrusion molding tension control system based on immersion image and surface image analysis, such as... Figure 1 As shown, it includes: The vision acquisition module is used to acquire the fiber bundle exit path and the molding profile exit path in real time, and obtain the corresponding fiber bundle impregnation state image and molding profile surface texture image. The feature extraction module is used to extract the surface texture of the fiber bundle impregnation state image and the surface texture image of the molded profile, respectively, to obtain the corresponding impregnation state feature parameters and molded surface feature parameters; The tension control module is used to generate a tension control signal based on qualified characteristic parameters combined with the immersion state characteristic parameters and the forming surface characteristic parameters, and to adjust the fiber bundle exit path and the forming profile exit path. The monitoring and feedback module is used to acquire the output molded products from the material discharge path of the molded profile, perform sampling monitoring on the output molded products, and optimize the tension control signal based on the sampling monitoring results.

[0022] In this example, the fiber bundle exit path represents the entire path of the fiber bundle after it has been resin-impregnated in the impregnation tank, leaving the impregnation tank and being transferred to the mold.

[0023] In this example, the profile discharge path represents the entire path of the composite material profile after being extruded by the mold, from the mold outlet to the output and subsequent traction.

[0024] In this example, the fiber bundle impregnation status image represents a real-time image taken by an industrial camera deployed at the outlet of the impregnation tank, which reflects the cross-section of the fiber bundle and the fullness and uniformity of resin impregnation on the surface.

[0025] In this example, the surface texture image of the formed profile is captured by an industrial camera deployed at the mold exit, which can present real-time images of the surface contour, texture details, and whether there are defects such as wrinkles and scratches on the formed profile.

[0026] In this example, the impregnation state characteristic parameters represent quantitative indicators obtained by analyzing the impregnation state images of the fiber bundles, specifically including the mean gray value (reflecting the fullness of impregnation) and standard deviation (reflecting the uniformity of impregnation) of the fiber bundle region.

[0027] In this example, the surface feature parameters represent quantitative indicators obtained by analyzing the surface texture image of the formed profile, specifically including the number of surface wrinkles, the length of a single wrinkle, and the percentage of wrinkle area.

[0028] In this example, the tension control signal represents the instruction signal generated by the system for adjusting the tension, quantized as an analog signal or digital pulse signal of 0-5V, corresponding to a rated tension adjustment range of 0-5%.

[0029] In this example, process adjustment refers to targeted fine-tuning of the yarn frame tension (adjusting fiber bundle impregnation time) or traction machine tension (smoothing surface wrinkles) based on the tension control signal.

[0030] In this example, the output molded product refers to the final composite material profile (such as an insulating strip) that is output from the molded profile outlet path after the pultrusion process has been fully processed.

[0031] In this example, the sampling monitoring results indicate that the molded products are sampled according to a preset ratio, and offline testing is used to verify whether there are defects such as dry spots and wrinkles, thereby determining whether the current tension control effect meets the qualified standard.

[0032] In this example, optimizing the tension control signal means adjusting the parameters of the tension control signal (such as adjustment amplitude, step size, etc.) based on the difference between the sampling monitoring results and the qualification standard, so that the process adjustment is more in line with the actual production needs and the product quality stability is improved.

[0033] The working principle and beneficial effects of the above technical solution are as follows: To improve the pass rate and production efficiency of pultruded products, reduce labor costs and raw material losses, and ensure strong compatibility, industrial cameras are first precisely deployed at key locations to achieve real-time, high-quality acquisition of fiber bundle impregnation status and surface texture of the formed profile. This results in low image transmission latency and strong anti-interference capabilities, avoiding detection errors and control failures caused by data distortion. Then, professional image processing algorithms are used to reduce noise, segment, and detect edges in the acquired images, accurately extracting quantified impregnation status feature parameters and formed surface feature parameters. This converts visual information into digital signals that can be used for control, avoiding the subjectivity and missed detection rate of manual inspection, and significantly improving the accuracy and efficiency of defect identification. Furthermore, based on the pass / fail feature parameters and real-time extracted... Quantitative characteristic parameters generate targeted tension control signals, accurately select control targets based on defect types, and avoid sudden tension changes through step size control, achieving real-time, targeted, and stable process adjustments. This fundamentally prevents defects such as poor impregnation and surface wrinkles. Finally, through sampling monitoring of the output molded products and real-time tracking of the process adjustment process, a closed-loop optimization mechanism is formed. This not only verifies the actual effect of tension control but also dynamically optimizes the control signals based on monitoring results, ensuring that the system continuously adapts to changes in production conditions and further improves the stability of product quality. At the same time, the anomaly early warning mechanism reduces the risk of batch scrapping. In this way, it is possible to achieve defect prevention and stable optimization of the production process, solving problems such as delayed defect detection, high dependence on manual labor, and lack of targeted control.

[0034] Example 2: Based on Example 1, the material extrusion molding tension control system based on impregnation image and surface image analysis includes a vision acquisition module comprising: A first industrial camera is used to acquire first real-time video information of the fiber bundle exit path and generate a corresponding fiber bundle impregnation state image. The second industrial camera is used to collect real-time video information of the material discharge path of the molded profile and generate a corresponding surface texture image of the molded profile. The network access monitoring unit is used to connect the first industrial camera and the second industrial camera to a preset Ethernet, and to acquire the corresponding first image data delay feature and second image data transmission delay feature, respectively. When the delay feature of the first image data is less than 50ms or the delay feature of the second image data transmission is less than 50ms, the transmission of the first industrial camera or the second industrial camera is optimized.

[0035] In this example, the first real-time video information refers to real-time dynamic video data captured by a first industrial camera deployed at the outlet of the impregnation tank, which continuously reflects the cross-section of the fiber bundle and the surface impregnation process along the fiber bundle exit path.

[0036] In this example, the second real-time video information refers to real-time dynamic video data captured by a second industrial camera deployed at the mold exit, which continuously reflects the process of surface texture and defects on the profile along the material discharge path.

[0037] In this example, the preset Ethernet refers to a high-speed network environment that is pre-configured and deployed to enable data transmission between the first industrial camera, the second industrial camera, and the system main controller.

[0038] In this example, the first image data delay feature represents the time difference between the generation and receipt of the first real-time video information acquired by the first industrial camera and the transmission to the system's main controller.

[0039] In this example, the second image data delay feature represents the time difference between the generation and receipt of the second real-time video information acquired by the second industrial camera and the transmission to the system's main controller.

[0040] In this example, transmission optimization refers to targeted measures such as adjusting network parameters, optimizing transmission interfaces, and enhancing signal stability when the delay characteristics of the first or second image data are higher than 50ms, in order to reduce data transmission latency and ensure that transmission efficiency meets the standards.

[0041] The working principle and beneficial effects of the above technical solution are as follows: In order to achieve real-time, high-quality acquisition and efficient transmission of image data in key stages of the pultrusion process, two industrial cameras first focus on the fiber bundle exit path and the profile exit path respectively, accurately capturing the core information of impregnation status and surface texture, solving the problems of untimely and incomplete data acquisition. Then, the image data transmission delay is monitored in real time, and the transmission of cameras with insufficient delay is optimized to ensure that the image data transmission delay is controlled within 50ms, ensuring the timeliness of data transmission and avoiding untimely control response due to transmission lag, further improving the real-time performance and reliability of the entire control system. At the same time, the video information acquired by the cameras can be converted into clear images, which are suitable for the harsh working conditions of high temperature and dust in the production site, providing a high-quality data foundation for subsequent defect identification.

[0042] Example 3: Based on Example 1, the feature extraction module of the material extrusion molding tension control system based on impregnation image and surface image analysis includes: The noise identification unit is used to perform noise identification on the fiber bundle impregnation state image and the molded profile surface texture image respectively, to obtain a first brightness defect and a first on-site interference in the fiber bundle impregnation state image, and to obtain a second brightness defect and a second on-site interference in the molded profile surface texture image. An image preprocessing unit is used to perform noise reduction processing on the fiber bundle impregnation state image based on the first brightness defect and the first field interference, and to perform noise reduction processing on the surface texture image of the molded profile based on the second brightness defect and the second field interference. The impregnation feature recognition unit is used to perform region separation on the noise-reduced fiber bundle impregnation state image, remove the background region to obtain the fiber bundle region of the noise-reduced fiber bundle impregnation state image, and calculate the gray mean and standard deviation of the fiber bundle region to obtain the impregnation state feature parameters. The forming surface feature recognition unit is used to perform edge detection on the surface texture image of the denoised forming profile. Based on the surface contour information and texture detail information obtained from the edge detection, it calculates several surface wrinkles of the forming product and the wrinkle length and wrinkle area ratio corresponding to each surface wrinkle, and obtains the forming surface feature parameters.

[0043] In this example, the first brightness defect refers to an abnormal brightness problem in the fiber bundle impregnation state image caused by fluctuations in lighting conditions at the production site (too bright, too dark, uneven lighting).

[0044] In this example, the second brightness defect refers to an abnormal brightness problem in the surface texture image of the molded profile caused by changes in ambient lighting conditions.

[0045] In this example, the first field interference refers to non-target interference factors from the production site present in the fiber bundle impregnation status image, mainly including dust, resin droplets, etc.

[0046] In this example, the second field interference refers to non-target interference factors from the production site that exist in the surface texture image of the formed profile, mainly including dust, environmental reflection, etc.

[0047] In this example, the fiber bundle region represents the image region containing only the fiber bundle after region separation of the denoised fiber bundle impregnation state image and removal of irrelevant background regions.

[0048] In this example, the grayscale mean represents the average grayscale value of all pixels within the fiber bundle region, and is the core quantitative indicator reflecting the resin impregnation fullness of the fiber bundle.

[0049] In this example, the standard deviation represents the dispersion of gray values ​​of all pixels within the fiber bundle region, and is a core quantitative indicator reflecting the uniformity of resin impregnation in the fiber bundle.

[0050] In this example, the surface contour information represents the relevant data of the overall shape contour of the molded profile surface obtained after edge detection is performed on the surface texture image of the denoised molded profile.

[0051] In this example, texture detail information refers to the specific detail data such as minor bumps and textures on the surface of the profile captured after edge detection of the surface texture image of the denoised profile.

[0052] In this example, surface wrinkles refer to wrinkle-like molding defects on the surface of the molded profile caused by factors such as improper tension.

[0053] In this example, the wrinkle length represents the linear length of each surface wrinkle extending along the profile surface.

[0054] In this example, the wrinkle area percentage represents the ratio of the total area of ​​all surface wrinkles to the total area of ​​the inspection area on the surface of the formed profile.

[0055] The working principle and beneficial effects of the above technical solution are as follows: In order to achieve accurate processing throughout the entire process from image preprocessing to feature quantization, interference is first eliminated in a targeted manner to significantly improve image quality and avoid the influence of impurities on feature extraction. Then, through region separation and quantization calculation, the fiber bundle impregnation state is transformed into intuitive and comparable parameters, realizing accurate judgment of impregnation uniformity. Furthermore, edge detection technology is used to comprehensively capture and quantify the key information of the profile surface wrinkles, ensuring that no surface defects are missed. Finally, the visual image is transformed into standardized digital feature parameters, completely eliminating the subjectivity and limitations of manual inspection, significantly improving the accuracy and efficiency of defect identification, providing accurate and reliable data support for the tension control module, and ensuring the targeted and effective adjustment of subsequent processes.

[0056] Example 4: Based on Example 3, the material extrusion molding tension control system based on impregnation image and surface image analysis further includes: The defect identification unit is used to compare the number of surface wrinkles and the wrinkle length and wrinkle area ratio of each surface wrinkle using a preset threshold for the number of surface wrinkles, a threshold for the length of a single wrinkle, and the wrinkle area ratio, and generate a threshold comparison report. When the threshold comparison report contains comparison results that do not meet the standards, it is determined that the molded product has surface defects.

[0057] In this example, the preset surface wrinkle number threshold is a pre-set limit value based on the surface feature data of qualified molded products, used to determine whether the number of surface wrinkles of the molded product exceeds the standard.

[0058] In this example, the single wrinkle length threshold represents a pre-set limit value, based on the surface feature data of a qualified molded product, used to determine whether the length of a single surface wrinkle of the molded product exceeds the standard.

[0059] In this example, the wrinkle area ratio threshold represents a pre-set limit value, based on the surface feature data of qualified molded products, used to determine whether the ratio of the total area of ​​all surface wrinkles of the molded product to the total area of ​​the surface inspection area exceeds the standard.

[0060] The working principle and beneficial effects of the above technical solution are as follows: By establishing a standardized and quantitative threshold comparison mechanism, the problem of relying on subjective human judgment, inconsistent standards, and high rates of missed and false detection in traditional surface defect judgment is effectively solved. When the report shows a result that does not meet the standard, the surface defect is directly identified, providing a clear and specific control trigger signal for the subsequent tension control module. This ensures that the tension adjustment can be targeted at the defect problem, avoids blind control, and further improves the accuracy and efficiency of the entire control system in preventing and controlling surface defects, thus ensuring the consistency of the surface quality of the molded products.

[0061] Example 5: Based on Example 1, the tension control module of the material extrusion molding tension control system based on impregnation image and surface image analysis includes: The parameter processing unit is used to normalize the immersion state characteristic parameters and the molding surface characteristic parameters to obtain several quantized scores within a preset value range, and to perform deviation analysis between each quantized score and the preset standard score to obtain several deviation parameters. The signal control unit is used to deduce the total required tension of the fiber bundle outlet path and the molding profile outlet path based on the deviation parameter, and to correct the original tension control signal using the total required tension to obtain the tension control signal. The process adjustment unit is used to perform multiple process adjustment simulations on the fiber bundle outlet path and the molded profile outlet path using the tension control signal with a single adjustment amount of 5V, until the total required tension is met.

[0062] In this example, the quantized score represents the standardized value obtained after normalizing the immersion state characteristic parameters (grayscale mean, standard deviation) and the molding surface characteristic parameters (number of wrinkles, length, area ratio), which are within a unified numerical range.

[0063] In this example, the preset numerical range refers to the pre-defined numerical interval used to unify the quantification standard of feature parameters, specifically 0-100, to ensure that different types of feature parameters can be compared horizontally.

[0064] In this example, the deviation parameter represents the difference (or degree of difference index) between each quantified score and the corresponding preset standard score, which is used to quantify the gap between the current production status and the qualified standard.

[0065] In this example, the total required tension represents the target tension value calculated based on the deviation parameter, which can bring the fiber bundle impregnation state and the surface quality of the formed profile back to the qualified standard. It is the core objective of tension control.

[0066] In this example, the original tension control signal indicates that the initial tension adjustment command was not combined with the deviation analysis of the characteristic parameters of this production, but was only based on the preset initial tension adjustment command of the general production scenario, and was not adapted and optimized for the specific defects.

[0067] In this example, the single adjustment amount represents the signal value corresponding to the fixed tension adjustment amplitude used each time a process adjustment simulation is performed.

[0068] In this example, the simulation of multiple process adjustments represents a virtual adjustment process in which the tension of the fiber bundle exit path and the profile exit path is gradually adjusted and verified to achieve the total required tension before the actual physical tension adjustment is performed on the tension control signal and the single adjustment amount.

[0069] The working principle and beneficial effects of the above technical solution are as follows: To address the problems of lack of unified parameter standards, weak targeting of control signals, and easy product quality fluctuations caused by actual adjustments in traditional tension control, the solution firstly unifies the characteristic parameters of the immersion state and the characteristic parameters of the molding surface in different dimensions to a preset value range through normalization processing. This eliminates the comparison barriers caused by differences in parameter dimensions, forming an intuitive and easy-to-understand quantitative score. Through deviation analysis with the preset standard score, the gap between the current parameter and the qualified standard is accurately located, providing a clear target basis for subsequent tension control and avoiding blind control caused by parameter confusion. Then, the total required tension is derived based on the deviation parameters to ensure accurate tension adjustment. The target is directly linked to the defect improvement needs, making the control direction more targeted. By correcting the original tension control signal through the total required tension, the redundant parts of the original signal that do not match the actual needs are eliminated, making the final tension control signal more in line with the actual production, improving the accuracy and executability of the signal. Furthermore, multiple process adjustment simulations are performed with a fixed single adjustment amount to verify the adjustment effect before the actual physical adjustment is performed. This avoids the tension sudden change caused by a single large adjustment, reduces the risk of new product molding defects, and ensures that the total required tension is met through a gradual approximation method. This improves the stability and reliability of process adjustment and reduces the trial and error costs in actual production.

[0070] Example 6: Based on Example 5, the material extrusion molding tension control system based on impregnation image and surface image analysis further includes: The threshold setting unit is used to construct standard immersion state data and standard surface wrinkle data based on the image feature data corresponding to the qualified molded product, and at the same time obtain the first process adjustment amount corresponding to the fiber bundle exit path and the second process adjustment amount corresponding to the molded profile exit path. The tension setting unit is used to take 1% of the first adjustment amount as the first tension adjustment step threshold of the fiber bundle outlet path and 1% of the second adjustment amount as the second tension adjustment step threshold of the molded profile outlet path. The control execution unit is used to adjust the tension of the fiber bundle exit path based on the first tension adjustment step size threshold when the impregnation state characteristic parameters are inconsistent with the standard impregnation state data, and to adjust the tension of the molded profile exit path based on the second tension adjustment step size threshold when the molding surface characteristic parameters are inconsistent with the standard surface wrinkle data.

[0071] In this example, the first process adjustment amount represents the total range of tension adjustments that can be made (i.e., the adjustable range of the rated tension) in order to ensure that the fiber bundle impregnation state meets the qualified standard within the normal production range of the fiber bundle exit path.

[0072] In this example, the second process adjustment amount represents the total range of tension adjustments that can be made (i.e., the adjustable range of the rated tension) in order to ensure that the surface quality of the formed profile meets the qualified standard within the normal production range of the material discharge path of the formed profile.

[0073] In this example, the first tension adjustment step size threshold represents the upper limit of the single tension adjustment range set based on 1% of the first process adjustment amount of the fiber bundle exit path. It is used to control the amount of change of tension adjustment in this path each time and avoid excessive adjustment range.

[0074] In this example, the second tension adjustment step size threshold represents the upper limit of the single tension adjustment range set based on 1% of the second process adjustment amount of the profile outlet path. It is used to control the amount of change in tension adjustment for each step of the path and avoid excessive adjustment range.

[0075] In this example, the standard impregnation status data represents a pre-constructed quantitative standard data containing the grayscale mean threshold T1 and the standard deviation threshold T2, based on the fiber bundle impregnation image features of qualified molded products, used to determine whether the fiber bundle impregnation is uniform and full.

[0076] In this example, the standard surface wrinkle data represents a pre-constructed quantitative standard data based on the surface texture image features of qualified molded products. This data includes a threshold N for the number of surface wrinkles, a threshold L for the length of a single wrinkle, and a threshold S for the proportion of wrinkle area. It is used to determine whether there are defects on the surface of the molded profile.

[0077] The working principle and beneficial effects of the above technical solution are as follows: To further improve the stability, accuracy, and reliability of tension control and achieve defect prevention and product quality improvement, standard immersion state data and standard surface wrinkle data are first constructed using image feature data of qualified molded products. This provides an objective and unified reference benchmark for subsequent defect judgment, avoiding judgment deviations caused by the lack of clear standards. The process adjustment amounts corresponding to the two paths are obtained, providing basic data that fits the actual production conditions for subsequent step size threshold setting, ensuring that the step size setting does not deviate from actual production. Then, 1% of the process adjustment amount is used as the tension adjustment step size threshold, controlling the adjustment range within a small and safe range. This effectively avoids tension mutations caused by a single large adjustment, reduces the risk of new product defects due to improper adjustment, and ensures the stability of the production process and the quality of product molding. Finally, the control object and the corresponding step size threshold are accurately matched according to the defect type. When the immersion state is abnormal, the tension of the fiber bundle exit path is adjusted in a targeted manner; when the surface wrinkles are abnormal, the tension of the molded profile exit path is adjusted in a targeted manner. This ensures that the control action directly addresses the root cause of the problem, avoids blind control, and improves the effectiveness and efficiency of tension adjustment.

[0078] Example 7: Based on Example 1, the material extrusion molding tension control system based on impregnation image and surface image analysis, the supervision feedback module includes: The sampling and testing unit is used to obtain the output molded products of the molded profile discharge path, sample the output molded products at a specified ratio, obtain a number of sampled products, and perform product testing. Determine whether the current tension meets the acceptable standard based on the test results; An optimization execution unit is used to generate a tension optimization reference based on the information difference between the test results and the pass standard if the performance does not meet the requirements, and then feed it back to the tension control module to optimize the tension control signal.

[0079] In this example, the pass standard refers to the pre-set judgment criteria based on the quality requirements of pultruded products, which include core indicators such as product impregnation uniformity (no dry spots) and surface quality (no excessive wrinkles or scratches).

[0080] In this example, the specified ratio refers to the sampling ratio (such as 5%-10% of the production batch) that is pre-set based on the production batch, product importance, and testing efficiency requirements. This ensures that the sampled products can objectively and comprehensively reflect the quality status of the entire batch of finished products, guaranteeing the representativeness of the test results while avoiding resource waste caused by over-sampling.

[0081] The working principle and beneficial effects of the above technical solution are as follows: To ensure the stability and consistency of product quality and reduce the risk of batch scrap, the output molded products are first obtained and sampled for testing according to a specified ratio. This allows for rapid determination of whether the current tension meets the qualification standard without requiring full inspection of all products. While ensuring the effectiveness of the test, it significantly reduces testing costs and time. The sampling test results directly reflect the actual effect of tension control, providing a real and reliable basis for subsequent optimization. This avoids the limitations of relying solely on image data for control while ignoring the actual product quality. When the test results do not meet the qualification standard, the information difference between the test results and the qualification standard is accurately captured, generating a targeted tension optimization reference. This ensures that the optimization direction fed back to the tension control module is clear and meets actual needs. By feeding back the optimization reference in real time and adjusting the tension control signal, the tension control can continuously correct deviations, gradually approaching the optimal state, thus improving the adaptive capability and long-term stability of the entire control system.

[0082] Example 8: Based on Example 1, the material extrusion molding tension control system based on impregnation image and surface image analysis, the supervision feedback module includes: The real-time analysis unit is used to acquire several consecutive frames of images during the process adjustment, derive the quantitative feature value of the process adjustment, and stop the tension adjustment and maintain the current parameters when the quantitative feature value is consistent with the standard quantitative threshold. The continuous adjustment unit is used to continuously acquire several consecutive frames of images and perform threshold comparison when the quantized feature value is inconsistent with the standard threshold. The warning and alert unit is used to determine that the process adjustment has failed and issue an anomaly warning when the number of threshold comparisons exceeds a specified number. In this example, the quantized feature value represents the normalized value in the range of 0-100 obtained by processing the extracted immersion state feature parameters (mean gray level, standard deviation) and the forming surface feature parameters (number of wrinkles, length, area ratio) after processing the continuous images collected during the process adjustment.

[0083] In this example, the standard quantification threshold represents the range of quantification indicators preset based on the image feature data of qualified molded products to determine whether the production status meets the standards. These include the grayscale mean threshold T1 and standard deviation threshold T2 corresponding to impregnation uniformity, as well as the quantity threshold N, length threshold L, and area ratio threshold S corresponding to surface wrinkles. These serve as the benchmark for comparison of quantification feature values.

[0084] In this example, the specified number of comparisons refers to the pre-set maximum threshold number of comparisons used to determine whether the process adjustment is effective. It is set to 5 times. When the cumulative number of comparisons exceeds this value and still fails to meet the standard, it is determined that the automatic control cannot solve the current problem and triggers an abnormal warning.

[0085] The working principle and beneficial effects of the above technical solution are as follows: To further improve the dynamic response capability, control accuracy, and production risk prevention and control level of the entire control system, and to ensure the continuity and stability of pultrusion production, continuous images during the process adjustment are first acquired and quantitative characteristic values ​​are derived. This allows for real-time tracking of the impact of tension adjustment on product quality. When the quantitative characteristic value matches the standard quantitative threshold, the adjustment is stopped immediately, ensuring product quality meets standards while avoiding over-adjustment that could cause new process fluctuations, thus improving the accuracy and efficiency of control. Then, when the quantitative characteristic value does not meet the standard, continuous images are continuously acquired and threshold comparisons are performed to ensure that the system can dynamically follow changes in production status. By gradually adjusting to approach the standard threshold, optimization is avoided due to a single adjustment failing to meet the standard, ensuring the thoroughness of control. Finally, a specified number of times is set as the early warning trigger condition. When the number of threshold comparisons exceeds this range, the process adjustment is judged to have failed and an abnormality warning is initiated. This can promptly identify complex problems that cannot be solved by automatic control, avoiding wasted time and batch product defects caused by ineffective adjustments. At the same time, it provides operators with clear intervention signals, facilitating rapid troubleshooting of the root cause of the problem and reducing production losses.

[0086] Example 9: This example proposes a material extrusion molding tension control method based on impregnation image and surface image analysis, such as... Figure 2 As shown, it includes: Step 1: Real-time acquisition of fiber bundle exit path and molded profile exit path to obtain corresponding fiber bundle impregnation state images and molded profile surface texture images; Step 2: Extract the surface texture of the fiber bundle impregnation state image and the molded profile surface texture image respectively to obtain the corresponding impregnation state feature parameters and molded surface feature parameters; Step 3: Based on the qualified characteristic parameters, combined with the impregnation state characteristic parameters and the forming surface characteristic parameters, a tension control signal is generated to adjust the fiber bundle exit path and the forming profile exit path. Step 4: Obtain the output molded product from the material discharge path of the molded profile, perform sampling monitoring on the output molded product, and optimize the tension control signal based on the sampling monitoring results.

[0087] In this example, the fiber bundle exit path represents the entire path of the fiber bundle after it has been resin-impregnated in the impregnation tank, leaving the impregnation tank and being transferred to the mold.

[0088] In this example, the profile discharge path represents the entire path of the composite material profile after being extruded by the mold, from the mold outlet to the output and subsequent traction.

[0089] In this example, the fiber bundle impregnation status image represents a real-time image taken by an industrial camera deployed at the outlet of the impregnation tank, which reflects the cross-section of the fiber bundle and the fullness and uniformity of resin impregnation on the surface.

[0090] In this example, the surface texture image of the formed profile is captured by an industrial camera deployed at the mold exit, which can present real-time images of the surface contour, texture details, and whether there are defects such as wrinkles and scratches on the formed profile.

[0091] In this example, the impregnation state characteristic parameters represent quantitative indicators obtained by analyzing the impregnation state images of the fiber bundles, specifically including the mean gray value (reflecting the fullness of impregnation) and standard deviation (reflecting the uniformity of impregnation) of the fiber bundle region.

[0092] In this example, the surface feature parameters represent quantitative indicators obtained by analyzing the surface texture image of the formed profile, specifically including the number of surface wrinkles, the length of a single wrinkle, and the percentage of wrinkle area.

[0093] In this example, the tension control signal represents the instruction signal generated by the system for adjusting the tension, quantized as an analog signal or digital pulse signal of 0-5V, corresponding to a rated tension adjustment range of 0-5%.

[0094] In this example, process adjustment refers to targeted fine-tuning of the yarn frame tension (adjusting fiber bundle impregnation time) or traction machine tension (smoothing surface wrinkles) based on the tension control signal.

[0095] In this example, the output molded product refers to the final composite material profile (such as an insulating strip) that is output from the molded profile outlet path after the pultrusion process has been fully processed.

[0096] In this example, the sampling monitoring results indicate that the molded products are sampled according to a preset ratio, and offline testing is used to verify whether there are defects such as dry spots and wrinkles, thereby determining whether the current tension control effect meets the qualified standard.

[0097] In this example, optimizing the tension control signal means adjusting the parameters of the tension control signal (such as adjustment amplitude, step size, etc.) based on the difference between the sampling monitoring results and the qualification standard, so that the process adjustment is more in line with the actual production needs and the product quality stability is improved.

[0098] The working principle and beneficial effects of the above technical solution are as follows: To improve the pass rate and production efficiency of pultruded products, reduce labor costs and raw material losses, and ensure strong compatibility, industrial cameras are first precisely deployed at key locations to achieve real-time, high-quality acquisition of fiber bundle impregnation status and surface texture of the formed profile. This results in low image transmission latency and strong anti-interference capabilities, avoiding detection errors and control failures caused by data distortion. Then, professional image processing algorithms are used to reduce noise, segment, and detect edges in the acquired images, accurately extracting quantified impregnation status feature parameters and formed surface feature parameters. This converts visual information into digital signals that can be used for control, avoiding the subjectivity and missed detection rate of manual inspection, and significantly improving the accuracy and efficiency of defect identification. Furthermore, based on the pass / fail feature parameters and real-time extracted... Quantitative characteristic parameters generate targeted tension control signals, accurately select control targets based on defect types, and avoid sudden tension changes through step size control, achieving real-time, targeted, and stable process adjustments. This fundamentally prevents defects such as poor impregnation and surface wrinkles. Finally, through sampling monitoring of the output molded products and real-time tracking of the process adjustment process, a closed-loop optimization mechanism is formed. This not only verifies the actual effect of tension control but also dynamically optimizes the control signals based on monitoring results, ensuring that the system continuously adapts to changes in production conditions and further improves the stability of product quality. At the same time, the anomaly early warning mechanism reduces the risk of batch scrapping. In this way, it is possible to achieve defect prevention and stable optimization of the production process, solving problems such as delayed defect detection, high dependence on manual labor, and lack of targeted control.

[0099] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A material extrusion tension control system based on immersion image and surface image analysis, characterized in that, include: The vision acquisition module is used to acquire the fiber bundle exit path and the molding profile exit path in real time, and obtain the corresponding fiber bundle impregnation state image and molding profile surface texture image. The feature extraction module is used to extract the surface texture of the fiber bundle impregnation state image and the surface texture image of the molded profile, respectively, to obtain the corresponding impregnation state feature parameters and molded surface feature parameters; The tension control module is used to generate a tension control signal based on qualified characteristic parameters combined with the immersion state characteristic parameters and the forming surface characteristic parameters, and to adjust the fiber bundle exit path and the forming profile exit path. The monitoring and feedback module is used to acquire the output molded products from the material discharge path of the molded profile, perform sampling monitoring on the output molded products, and optimize the tension control signal based on the sampling monitoring results.

2. The material extrusion tension control system based on immersion image and surface image analysis of claim 1, wherein, The visual acquisition module includes: A first industrial camera is used to acquire first real-time video information of the fiber bundle exit path and generate a corresponding fiber bundle impregnation state image. The second industrial camera is used to collect real-time video information of the material discharge path of the molded profile and generate a corresponding surface texture image of the molded profile. The network access monitoring unit is used to connect the first industrial camera and the second industrial camera to a preset Ethernet, and to acquire the corresponding first image data delay feature and second image data transmission delay feature, respectively. When the delay characteristic of the first image data is less than 50ms or the delay characteristic of the second image data transmission is less than 50ms, the transmission of the first industrial camera or the second industrial camera is optimized.

3. The material extrusion tension control system based on immersion image and surface image analysis of claim 1, wherein, The feature extraction module includes: The noise identification unit is used to perform noise identification on the fiber bundle impregnation state image and the molded profile surface texture image respectively, to obtain a first brightness defect and a first on-site interference in the fiber bundle impregnation state image, and to obtain a second brightness defect and a second on-site interference in the molded profile surface texture image. An image preprocessing unit is used to perform noise reduction processing on the fiber bundle impregnation state image based on the first brightness defect and the first field interference, and to perform noise reduction processing on the surface texture image of the molded profile based on the second brightness defect and the second field interference. The impregnation feature recognition unit is used to perform region separation on the noise-reduced fiber bundle impregnation state image, remove the background region to obtain the fiber bundle region of the noise-reduced fiber bundle impregnation state image, and calculate the gray mean and standard deviation of the fiber bundle region to obtain the impregnation state feature parameters. The forming surface feature recognition unit is used to perform edge detection on the surface texture image of the denoised forming profile. Based on the surface contour information and texture detail information obtained from the edge detection, it calculates several surface wrinkles of the forming product and the wrinkle length and wrinkle area ratio corresponding to each surface wrinkle, and obtains the forming surface feature parameters.

4. The material extrusion tension control system based on immersion image and surface image analysis of claim 3, wherein, Also includes: The defect identification unit is used to compare the number of surface wrinkles and the wrinkle length and wrinkle area ratio of each surface wrinkle using a preset threshold for the number of surface wrinkles, a threshold for the length of a single wrinkle, and the wrinkle area ratio, and generate a threshold comparison report. When the threshold comparison report contains comparison results that do not meet the standards, it is determined that the molded product has surface defects.

5. The material extrusion molding tension control system based on immersion image and surface image analysis as described in claim 1, characterized in that, The tension control module includes: The parameter processing unit is used to normalize the immersion state characteristic parameters and the molding surface characteristic parameters to obtain several quantized scores within a preset value range, and to perform deviation analysis between each quantized score and the preset standard score to obtain several deviation parameters. The signal control unit is used to deduce the total required tension of the fiber bundle outlet path and the molding profile outlet path based on the deviation parameter, and to correct the original tension control signal using the total required tension to obtain the tension control signal. The process adjustment unit is used to perform multiple process adjustment simulations on the fiber bundle outlet path and the molded profile outlet path using the tension control signal with a single adjustment amount of 5V, until the total required tension is met.

6. The material extrusion molding tension control system based on immersion image and surface image analysis as described in claim 5, characterized in that, Also includes: The threshold setting unit is used to construct standard immersion state data and standard surface wrinkle data based on the image feature data corresponding to the qualified molded product, and at the same time obtain the first process adjustment amount corresponding to the fiber bundle exit path and the second process adjustment amount corresponding to the molded profile exit path. The tension setting unit is used to take 1% of the first process adjustment as the first tension adjustment step threshold of the fiber bundle outlet path and 1% of the second process adjustment as the second tension adjustment step threshold of the molded profile outlet path. The control execution unit is used to adjust the tension of the fiber bundle exit path based on the first tension adjustment step size threshold when the impregnation state characteristic parameters are inconsistent with the standard impregnation state data, and to adjust the tension of the molded profile exit path based on the second tension adjustment step size threshold when the molding surface characteristic parameters are inconsistent with the standard surface wrinkle data.

7. The material extrusion molding tension control system based on immersion image and surface image analysis as described in claim 1, characterized in that, The supervision and feedback module includes: The sampling and testing unit is used to obtain the output molded products of the molded profile discharge path, sample the output molded products at a specified ratio, obtain a number of sampled products, and perform product testing. Determine whether the current tension meets the acceptable standard based on the test results; An optimization execution unit is used to generate a tension optimization reference based on the information difference between the test results and the pass standard if the performance does not meet the requirements, and then feed it back to the tension control module to optimize the tension control signal.

8. The material extrusion molding tension control system based on immersion image and surface image analysis as described in claim 1, characterized in that, The supervision and feedback module includes: The real-time analysis unit is used to acquire several consecutive frames of images during the process adjustment, derive the quantitative feature value of the process adjustment, and stop the tension adjustment and maintain the current parameters when the quantitative feature value is consistent with the standard quantitative threshold. The continuous adjustment unit is used to continuously acquire several consecutive frames of images and perform threshold comparison when the quantized feature value is inconsistent with the standard threshold. The warning and alert unit is used to determine that the process adjustment has failed and issue an abnormal warning when the number of threshold comparisons exceeds the specified number.

9. A method for controlling the tension in material extrusion molding based on immersion image and surface image analysis, characterized in that, include: Step 1: Real-time acquisition of fiber bundle exit path and molded profile exit path to obtain corresponding fiber bundle impregnation state images and molded profile surface texture images; Step 2: Extract the surface texture of the fiber bundle impregnation state image and the molded profile surface texture image respectively to obtain the corresponding impregnation state feature parameters and molded surface feature parameters; Step 3: Based on the qualified characteristic parameters, combined with the impregnation state characteristic parameters and the forming surface characteristic parameters, a tension control signal is generated to adjust the fiber bundle exit path and the forming profile exit path. Step 4: Obtain the output molded product from the material discharge path of the molded profile, perform sampling monitoring on the output molded product, and optimize the tension control signal based on the sampling monitoring results.

Images