A ribbon positioning device for a shoe rolling process and a control method thereof

Abstract

This invention relates to the field of shoe edge binding technology, and discloses a ribbon positioning device and its control method for shoe edge binding. The method includes: determining an error benchmark value for the ribbon positioning device based on historical average sewing errors and historical average alignment errors; determining the tension range of the ribbon positioning device based on the error benchmark value and a tension control model; judging the tension stability of the ribbon positioning device based on the relationship between tension measurements and the tension range; when the tension stability is deemed unqualified, determining a historical decision or a model decision based on the traversal results of the tension measurements in a historical database; when determining a historical decision, determining a historical reuse decision or a historical clustering decision based on the data similarity of the tension measurements in the historical database; when determining a model decision, determining a tension calibration value based on the tension positioning model; and straightening the ribbon according to the tension calibration value. This invention improves the automation level of the edge binding process.

Description

Technical Field

[0001] This invention relates to the field of shoe edge binding technology, and more specifically, to a ribbon positioning device and control method for shoe edge binding. Background Technology

[0002] Footwear edge binding is a crucial process in shoe manufacturing, used to optimize the edge structure and appearance of shoe components. Ribbon binding prevents issues like burrs, loose threads, and cracks, improving durability and lifespan, while also enriching the overall look and meeting the decorative and artistic needs of different shoe styles. Currently, the footwear industry is upgrading towards large-scale, automated, and standardized production. Traditional footwear edge binding typically relies on manual hand-held ribbon application and positioning, followed by sewing. This process depends entirely on the operator's experience to control the ribbon's placement, tension, and edge alignment. However, the precision and tension of the ribbon pulled manually depend entirely on the operator's hand movements and strength, making it difficult to guarantee product consistency. Furthermore, manual ribbon application results in longer production times per pair, hindering the achievement of edge binding standards and limiting the automation level of the process.

[0003] Therefore, it is necessary to design a ribbon positioning device and its control method for shoe edge binding process to solve the problems existing in the current technology. Summary of the Invention

[0004] In view of this, the present invention proposes a ribbon positioning device and control method for shoe edge binding process, aiming to solve the above-mentioned problems.

[0005] In one aspect, the present invention proposes a control method for a ribbon positioning device in a shoe edge-binding process, comprising: The historical average sewing error and historical average alignment error of the ribbon positioning device are obtained. Based on the historical average sewing error and historical average alignment error, the error reference value of the ribbon positioning device is determined. Based on the error reference value and the tension control model, the tension range of the ribbon positioning device is determined. The ribbon is threaded into the ribbon positioning device, and the tension measurement value of the ribbon positioning device is obtained. The tension stability of the ribbon positioning device is determined based on the relationship between the tension measurement value and the tension range. When the tension stability is determined to be unqualified, a historical decision or a model decision is determined based on the traversal results of the tension measurement value in the historical database. When the historical decision is determined, a historical reuse decision or a historical clustering decision is determined based on the data similarity of the tension measurement value in the historical database. When the historical reuse decision is determined, a tension calibration value is determined based on the historical database. When the historical clustering decision is determined, a tension calibration value is determined based on a clustering algorithm. When the model decision is determined, a tension calibration value is determined based on a tension positioning model. The ribbon is straightened according to the tension calibration value, and the sewing needle is aligned with the ribbon positioning device to complete the edge rolling process.

[0006] Furthermore, in determining the tension range of the ribbon positioning device, the process includes: acquiring a historical sewing dataset, dividing the historical sewing dataset into a training set and a test set, pre-acquiring a machine learning model, iteratively training the machine learning model based on the training set, validating the iteratively trained machine learning model based on the test set, determining the tension control model, and substituting the error benchmark value into the tension control model to determine the tension range of the ribbon positioning device.

[0007] Furthermore, when determining the tension range of the ribbon positioning device, the method further includes: if the accuracy of the machine learning model after the current iteration is less than the accuracy of the machine learning model after the previous iteration, then cosine annealing is used to adjust the learning rate of the machine learning model after the current iteration, and iterative training continues until the tension control model is determined; if the accuracy of the machine learning model after the current iteration is greater than or equal to the accuracy of the machine learning model after the previous iteration, then iterative training is stopped, and the machine learning model after the current iteration is determined as the tension control model.

[0008] Furthermore, when determining the tension stability of the ribbon positioning device based on the relationship between the tension measurement value and the tension range, the method includes: if the tension measurement value is within the tension range, the tension stability is determined to be qualified, and the hemming process is completed according to the tension measurement value; if the tension measurement value is not within the tension range, the tension stability is determined to be unqualified.

[0009] Furthermore, when determining historical decisions or model decisions based on the traversal results of the tension measurement values ​​in the historical database, the following steps are taken: the historical database includes several historical tension measurement values ​​and several historical tension calibration values, and each historical tension measurement value corresponds to a historical tension calibration value. When there are historical tension measurement values ​​in the historical database with a data similarity greater than a data similarity threshold, the historical decision is determined to be executed. When there are no historical tension measurement values ​​in the historical database with a data similarity greater than a data similarity threshold, the model decision is determined to be executed.

[0010] Furthermore, when determining a historical reuse decision or a historical clustering decision based on the data similarity of the tension measurement value in the historical database, the process includes: if a historical tension measurement value with a data similarity of 1 exists in the historical database, then the historical reuse decision is executed; if a historical tension measurement value with a data similarity of 1 does not exist in the historical database, then the historical clustering decision is executed.

[0011] Furthermore, when determining the tension calibration value based on the historical database, the method includes: counting the number of historical tension measurement values ​​with a data similarity of 1 in the historical database; if the number of historical data is 1, then the historical tension calibration value corresponding to the historical tension measurement value is determined as the tension calibration value; if the number of historical data is not 1, then the average of the historical tension calibration values ​​corresponding to several historical tension measurement values ​​with a data similarity of 1 is determined as the tension calibration value.

[0012] Furthermore, when determining the tension calibration value based on the clustering algorithm, the process includes: using the tension measurement value and the historical database as a historical clustering dataset, determining the expected number of clusters and initializing the parameters of the Gaussian distribution, determining the probability that each data point in the historical clustering dataset belongs to each Gaussian distribution, determining the historical dataset corresponding to the tension measurement value based on the probability, and determining the mean of the historical tension calibration values ​​in the historical dataset as the tension calibration value.

[0013] Furthermore, when determining the tension calibration value based on the tension positioning model, the process includes: constructing a model dataset based on the historical database, and determining a tension positioning model based on the model dataset, with the tension measurement value as the input and the tension calibration value as the output.

[0014] Compared with existing technologies, the advantages of this invention are as follows: By obtaining the historical average sewing error and historical average alignment error of the ribbon positioning device to determine the error benchmark value, and then combining it with the tension control model to determine the tension range, the risks of traditional manual edge binding relying on the operator's feel to control the tightness are avoided. This provides a quantitative basis for tension stability judgment and eliminates the arbitrariness and uncertainty of subjective judgment. By comparing the tension measurement value with the tension range and traversing the historical database to determine historical or model-based decisions, the reliability of the edge binding process is ensured by relying on historical data, while model-based decisions cover special working conditions without matching data, balancing efficiency and adaptability. When determining historical decisions, historical reuse decisions or historical clustering decisions are determined based on data similarity. Reuse matches past data experience, clustering optimizes scenario adaptability, and model-based decisions predict tension calibration values ​​through the tension positioning model, ensuring the automation level of tension control and avoiding risks such as ribbon wrinkles, stretching deformation, and positioning offset caused by abnormal tension. This improves the accuracy of edge binding and the consistency of batch products, ensuring the stability of shoe edge binding processes and product quality.

[0015] On the other hand, this application also provides a ribbon positioning device for shoe edge binding process, and a control method for applying the above-mentioned ribbon positioning device for shoe edge binding process, including: The positioning body is provided with a ribbon positioning hole for positioning the ribbon, and the positioning body is provided with a slot, a plurality of first positioning holes and a plurality of second positioning holes.

[0016] It is understandable that the ribbon positioning device and control method of the above-mentioned shoe edge rolling process have the same beneficial effects, and will not be described in detail here. Attached Figure Description

[0017] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating a control method for a ribbon positioning device in a shoe edge-binding process, provided in an embodiment of the present invention; Figure 2 A schematic diagram of the structure of a ribbon positioning device for shoe edge binding process provided in an embodiment of the present invention. Figure 1 ; Figure 3 A schematic diagram of the structure of a ribbon positioning device for shoe edge binding process provided in an embodiment of the present invention. Figure 2 ; Figure 4A schematic diagram of the structure of a ribbon positioning device for shoe edge binding process provided in an embodiment of the present invention. Figure 3 .

[0018] Among them, 1. positioning body; 2. ribbon positioning hole; 3. card slot; 4. first positioning hole; 5. second positioning hole. Detailed Implementation

[0019] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] In some embodiments of this application, see Figure 1 As shown, a control method for a ribbon positioning device in a shoe edge binding process includes: S100: Obtain the historical average sewing error and historical average alignment error of the ribbon positioning device, determine the error reference value of the ribbon positioning device based on the historical average sewing error and historical average alignment error, and determine the tension range of the ribbon positioning device based on the error reference value and the tension control model. S200: Thread the ribbon into the ribbon positioning device and obtain the tension measurement value of the ribbon positioning device. Determine the tension stability of the ribbon positioning device based on the relationship between the tension measurement value and the tension range. When the tension stability is determined to be unqualified, determine the historical decision or model decision based on the traversal results of the tension measurement value in the historical database. S300: When determining historical decisions, determine historical reuse decisions or historical clustering decisions based on the data similarity of tension measurement values ​​in the historical database. When determining historical reuse decisions, determine tension calibration values ​​based on the historical database. When determining historical clustering decisions, determine tension calibration values ​​based on clustering algorithms. When determining model decisions, determine tension calibration values ​​based on tension positioning models. S400: Straighten the ribbon according to the tension calibration value and align the sewing needle with the ribbon positioning device to complete the edge rolling process.

[0021] Specifically, the historical average sewing error and historical average alignment error of the ribbon positioning device are obtained. The historical average sewing error is the average of all deviations from the standard process position during multiple past shoe edge sewing operations using the ribbon positioning device. The historical average alignment error is the average of the alignment deviations between the ribbon edge and the shoe edge during multiple past operations using the ribbon positioning device. The historical average sewing error and historical average alignment error are normalized and weighted. The corresponding weights can be determined using methods such as hierarchical analysis and expert scoring. The error benchmark value is determined based on the weighted calculation results. This avoids the risk of product error fluctuations and poor consistency caused by relying entirely on operator feel and experience for edge sewing without a quantified error reference standard. Determining the error benchmark value based on the historical production data of the ribbon positioning device reflects the actual process error level. Based on the error benchmark value and tension... The force control model determines the tension range of the ribbon positioning device. The tension range is the range of tension values ​​that ensures the ribbon does not wrinkle, bunch up, or break during conveying, positioning, and straightening, while maintaining the required sewing and alignment accuracy. By combining historical errors with the model to determine the tension range, it can accurately match the ribbon tension requirements of shoe edge-binding processes. This solves the problems of ribbon deformation, skewed sewing, and alignment deviations caused by the lack of standard tension and inconsistent tension in traditional manual ribbon pulling. At the same time, it provides a unified and quantitative standard for subsequent tension stability assessment. The ribbon is threaded into the ribbon positioning device to complete the ribbon positioning and conveying. Simultaneously, the tension measurement value of the ribbon positioning device is acquired in real time. The tension measurement value is the actual tension monitored during the positioning and conveying process. Based on the correspondence between the tension measurement value and the tension range, the tension stability of the ribbon positioning device is judged, avoiding forced sewing under unqualified tension conditions that would result in defective products. When tension stability is deemed unacceptable, the system determines whether to use historical decision or model decision based on the results of traversing the historical database of tension measurements. A dual decision-making mode is employed based on the presence of tension measurements in the historical database: on the one hand, it can directly access a large amount of past tension calibration data to adapt to regular production scenarios; on the other hand, the model decision can handle special scenarios such as new shoe styles or new ribbon materials without matching historical data, thus covering all production situations. When historical decision is chosen, the system further determines whether to use historical reuse decision or historical clustering decision based on the data similarity of tension measurements in the historical database. Data similarity is an indicator of the degree of matching between the current tension measurement and historical data. Historical reuse decision can directly access completely matching historical data to maximize efficiency, while historical clustering decision can optimize approximate historical data to adapt to scenarios without completely matching data but with similar data, thereby ensuring the accuracy of sewing operations.The tension calibration value is determined by a corresponding decision-making method. The tension calibration value ensures that the ribbon is always flat, appropriately tight, and free of wrinkles and deviations. The sewing needle is aligned with the ribbon positioning device to complete the edge-binding process. Data-driven and automated control replaces the manual operation of pulling the ribbon, positioning, and adjusting. This not only ensures the consistency of sewing accuracy and alignment accuracy of the edge-binding process for each pair of shoes, but also shortens the production time of a single pair of shoes, thereby improving the automation level of the shoe edge-binding process.

[0022] In some embodiments of this application, determining the tension range of the ribbon positioning device includes: acquiring a historical sewing dataset and dividing the historical sewing dataset into a training set and a test set; acquiring a machine learning model in advance; iteratively training the machine learning model based on the training set; validating the iteratively trained machine learning model based on the test set; determining a tension control model; and substituting the error benchmark value into the tension control model to determine the tension range of the ribbon positioning device.

[0023] In some embodiments of this application, when determining the tension range of the ribbon positioning device, the method further includes: if the accuracy of the machine learning model after the current iteration training is less than the accuracy of the machine learning model after the previous iteration training, then cosine annealing is used to adjust the learning rate of the machine learning model after the current iteration training, and iterative training continues until the tension control model is determined; if the accuracy of the machine learning model after the current iteration training is greater than or equal to the accuracy of the machine learning model after the previous iteration training, then iterative training is stopped, and the machine learning model after the current iteration training is determined as the tension control model.

[0024] Specifically, the historical sewing dataset includes data such as the historical average sewing error, historical average alignment error, tension corresponding to different shoe material materials, tension corresponding to different ribbon widths and materials, sewing speed, and fluctuations under the influence of temperature and humidity in the production environment. The historical sewing dataset is divided into a training set and a test set according to a preset ratio, typically 4:1, to ensure the model's generalization level. The training set is used to train the machine learning model, and the test set is used to evaluate the performance of the trained model. A pre-designed machine learning model is selected as the initial model. It should be noted that the machine learning model can be implemented using existing technologies in the field, and it is not the focus of the improvement claimed in this application. The focus of this application's improvement lies in its constraint processing logic in the edge-rolling process. The machine learning model takes the error benchmark value as input, and through training, obtains a tension control model and outputs a tension range. Those skilled in the art can, based on the input-output relationship, parameter configuration rules, and calling sequence disclosed in this application, combine existing disclosed technologies or conventional engineering methods to complete the adaptation, replacement, or equivalent implementation without affecting the implementation of the technical solution of this application. When iteratively training a machine learning model using data from the training set, the model attempts to learn patterns and relationships in the data during each iteration to improve its predictive ability. After each iteration, the model is validated using data from the test set to measure its performance. If the learning rate after the current iteration is lower than the accuracy of the previous iteration, it indicates a decline in model performance. In this case, cosine annealing is used to adjust the learning rate, avoiding both excessively large learning rates that cause model oscillations and convergence, and excessively small learning rates that lead to inefficient training. Continued iterative training helps the model more stably approach the global optimum. If the accuracy of the machine learning model after the current iteration is greater than or equal to the previous one, it indicates that the model's performance has improved or remained stable. At this point, iterative training can be stopped. Dynamic optimization training enables the tension control model to accurately output the tension range of the ribbon positioning device, thereby improving the automation level of the shoe edge-binding process.

[0025] In some embodiments of this application, when determining the tension stability of the ribbon positioning device based on the relationship between the tension measurement value and the tension range, the following steps are taken: if the tension measurement value is within the tension range, the tension stability is determined to be qualified, and the hemming process is completed according to the tension measurement value; if the tension measurement value is not within the tension range, the tension stability is determined to be unqualified.

[0026] Specifically, ribbon tension is a key process parameter that determines the quality of the piping and sewing, the ribbon's forming state, and product consistency. Excessive tension will directly cause the ribbon to be overstretched and deformed, or even break. Insufficient tension will cause problems such as wrinkles, bunching, and positioning misalignment, leading to defects such as skewed sewing, excessive alignment errors, and irregular edges. When the tension measurement value is within the tension range, it indicates that the actual stress state of the ribbon fully meets the qualified requirements of the shoe piping process. The ribbon can maintain a flat, smooth, and appropriately tight state without deformation, misalignment, or wrinkles that affect the sewing quality. In this case, the tension stability is deemed qualified, and the subsequent piping process can be completed according to the current tension measurement value to improve piping efficiency. When the tension measurement value is outside the tension range, it indicates that the actual tension of the ribbon is in an abnormal state of being too high or too low, which cannot meet the piping and sewing process standards. If sewing continues, it will produce defective products, and the tension stability is deemed unqualified. By comparing the tension measurement value with the tension range, the blind reliance on the operator's hand feel can be avoided, thereby eliminating the uncertainty brought about by human operation.

[0027] In some embodiments of this application, when determining historical decisions or model decisions based on the traversal results of tension measurement values ​​in a historical database, the following steps are taken: the historical database includes several historical tension measurement values ​​and several historical tension calibration values, and each historical tension measurement value corresponds to a historical tension calibration value. When there are historical tension measurement values ​​in the historical database with a data similarity greater than a data similarity threshold, it is determined to execute a historical decision. When there are no historical tension measurement values ​​in the historical database with a data similarity greater than a data similarity threshold, it is determined to execute a model decision.

[0028] Specifically, data similarity reflects the degree of similarity between historical tension measurements. Data similarity can be determined using methods such as cosine similarity and Euclidean distance. Based on past production data aligned with the production patterns of shoe edge binding, the system searches the historical database for historical data similar to the current tension measurement. A data similarity threshold is used to determine the degree of matching between current and historical conditions. When historical data with high similarity is found, historical decisions are executed, and tension calibration values ​​are determined using the historical database, thus ensuring the reliability and consistency of the edge binding operation. For cases where current operating conditions do not perfectly match historical data, model decisions are executed. The model's predictive capabilities generate tension calibration values ​​adapted to the specific edge binding situation, ensuring that the ribbon positioning device can adapt to various production conditions and consistently provide a standard tension state for the edge binding process, thereby guaranteeing edge binding quality and production continuity.

[0029] In some embodiments of this application, when determining a historical reuse decision or a historical clustering decision based on the data similarity of tension measurement values ​​in a historical database, the following steps are taken: if a historical tension measurement value with a data similarity of 1 exists in the historical database, then a historical reuse decision is determined to be executed; if a historical tension measurement value with a data similarity of 1 does not exist in the historical database, then a historical clustering decision is determined to be executed.

[0030] Specifically, regarding the matching of tension measurements with data in the historical database, the historical reuse decision and historical clustering decision are further clarified. When a historical tension measurement with a similarity of 1 exists in the historical database, it indicates that the current tension measurement is completely consistent with the production process parameters of a past instance. In this case, the historical reuse decision is executed, directly using the parameters corresponding to that historical tension measurement to avoid duplicate calculations and ensure the consistency of sewing quality. When no historical tension measurement with a similarity of 1 exists in the historical database, it indicates that the current tension measurement does not have completely matching past production data, but contains some similar historical data. If the historical clustering decision is executed, the historical data closest to the current tension measurement value will be found through cluster analysis. The historical clustering decision addresses historical conditions and data-driven automated accumulation, reducing reliance on human experience and intuition, and lowering the uncertainty and operational risks brought about by human judgment. By comprehensively utilizing a large amount of historical data, rich reference information is provided for determining the tension calibration value. It can learn and optimize from historical experience, thereby continuously improving the accuracy and efficiency of ribbon positioning, ensuring the accuracy of ribbon positioning and sewing quality, ensuring the consistency of the piping of each pair of shoes in mass production, and improving production efficiency and product qualification rate.

[0031] In some embodiments of this application, when determining the tension calibration value based on a historical database, the method includes: counting the number of historical tension measurement values ​​with a data similarity of 1 in the historical database; if the number of historical data is 1, then the historical tension calibration value corresponding to the historical tension measurement value is determined as the tension calibration value; if the number of historical data is not 1, then the average of the historical tension calibration values ​​corresponding to several historical tension measurement values ​​with a data similarity of 1 is determined as the tension calibration value.

[0032] Specifically, the number of historical tension measurements in the historical database that have a similarity of 1 to the current tension measurement is counted. A similarity of 1 means that the current tension condition of the ribbon is completely consistent with the condition in the past hemming production. If the number of historical data is 1, it means that there is only one historical tension measurement in the historical database that completely matches the current tension measurement. In this case, the historical tension calibration value corresponding to the historical tension measurement is directly determined as the tension calibration value, which minimizes the response time and ensures the smoothness of the hemming process. At the same time, directly using the parameters from the past production can eliminate calibration deviations and ensure that the ribbon tension returns to the process range. If the number of historical data obtained from the statistics is not 1, that is, if there are multiple historical tension measurement values ​​in the historical database that perfectly match the current tension measurement value, then the average value of the historical tension calibration values ​​corresponding to several historical tension measurement values ​​with a similarity of 1 is calculated. This offsets the errors caused by minor fluctuations in working conditions and environmental interference that may exist in a single historical data, avoids the randomness defects of a single historical data, and improves the accuracy and universality of the tension calibration value. It not only makes full use of the data in historical production, but also further reduces the error through averaging optimization, ensuring that the tension calibration value is completely adapted to the current production working conditions. This ensures that the ribbon always maintains a suitable tension, is flat and wrinkle-free, and meets the quality requirements of the hemming and sewing process, thus guaranteeing the processing accuracy of the hemming and the quality consistency of the batch products.

[0033] In some embodiments of this application, when determining the tension calibration value based on the clustering algorithm, the process includes: using the tension measurement value and the historical database as a historical clustering dataset, determining the expected number of clusters and initializing the parameters of the Gaussian distribution, determining the probability that each data in the historical clustering dataset belongs to each Gaussian distribution, determining the historical dataset corresponding to the tension measurement value based on the probability, and determining the mean of the historical tension calibration values ​​in the historical dataset as the tension calibration value.

[0034] Specifically, tension measurements are combined with historical databases to form a historical clustering dataset, establishing a correlation between current working conditions and similar historical working conditions, ensuring that the clustering results align with the actual production patterns of shoe edge rolling processes. The desired number of clusters is determined and the parameters of the Gaussian distribution are initialized. The desired number of clusters can be adjusted according to the amount of data in the historical clustering dataset. Usually, the desired number of clusters is set to 2. Based on cluster analysis, all historical data that belong to the same Gaussian distribution and have the highest matching degree with the current tension measurement value are selected to form a historical dataset corresponding to the tension measurement value. The mean of all historical tension calibration values ​​in the historical dataset is calculated and the calculated mean is determined as the tension calibration value. This not only quickly finds the most suitable similar historical data in scenarios where there is no completely matching historical data, but also shortens the response time. By analyzing historical data through clustering algorithms, the historical dataset that is closest to the current conditions is found, which improves the accuracy of the tension calibration value determination, ensures the stability and efficiency of the hemming process, reduces the reliance on human experience and judgment, reduces human error in sewing, avoids ribbon wrinkles, stretching deformation and positioning offset, and ensures the processing accuracy of hemming sewing is consistent with the mass-produced products.

[0035] In some embodiments of this application, when determining the tension calibration value based on the tension positioning model, the process includes: constructing a model dataset based on a historical database, and determining a tension positioning model based on the model dataset, where the input is the tension measurement value and the output is the tension calibration value.

[0036] Specifically, the model dataset contains historical tension measurements, corresponding historical tension calibration values, and related process parameters under all past production conditions for shoe edge binding. This provides comprehensive and reliable data support for model training. The establishment and training process of the tension positioning model is consistent with that of the tension control model, and will not be repeated here. By determining the tension calibration value through the tension positioning model, the reliance on manual operation experience is eliminated, the blindness and uncertainty of human judgment are removed, and the processing accuracy of edge binding and sewing, the quality consistency of batch products, and production efficiency are improved.

[0037] In summary, the beneficial effects of this invention are as follows: By obtaining the historical average sewing error and historical average alignment error of the ribbon positioning device to determine the error benchmark value, and then combining it with the tension control model to determine the tension range, the risks of traditional manual edge binding relying on the operator's feel to control the tightness are avoided. This provides a quantitative basis for tension stability judgment and eliminates the arbitrariness and uncertainty of subjective judgment. By comparing the tension measurement value with the tension range and traversing the historical database to determine historical or model-based decisions, the reliability of the edge binding process is ensured by relying on historical data, while model-based decisions cover special working conditions without matching data, balancing efficiency and adaptability. When determining historical decisions, historical reuse decisions or historical clustering decisions are determined based on data similarity. Reuse matches past data experience, clustering optimizes scenario adaptability, and model-based decisions predict tension calibration values ​​through the tension positioning model, ensuring the automation level of tension control and avoiding risks such as ribbon wrinkles, stretching deformation, and positioning offset caused by abnormal tension. This improves the accuracy of edge binding and the consistency of batch products, ensuring the stability of shoe edge binding processes and product quality.

[0038] Based on another preferred embodiment described above, see [link to preferred embodiment]. Figure 2-4 As shown, this embodiment provides a ribbon positioning device for shoe edge binding process, and a control method for applying the above-mentioned ribbon positioning device for shoe edge binding process, including: Positioning body 1, the positioning body 1 is provided with ribbon positioning hole 2, the ribbon positioning hole 2 is used to position the ribbon, the positioning body 1 is provided with slot 3, a number of first positioning holes 4 and a number of second positioning holes 5.

[0039] Specifically, the overall shape of the positioning body 1 is a curved surface that adapts to the contour of the shoe's piping. The curved shape of the positioning body 1 can fit closely with the contour of the shoe, avoiding poor fit caused by suspension or offset during the ribbon transportation process. The positioning body 1 is provided with ribbon positioning holes 2, which are channels for ribbon transportation and positioning. The cross-sectional shape of the hole matches the specifications (width and thickness) of the ribbon, which can restrict the position of the ribbon inserted into it in all directions, constrain the lateral and longitudinal displacement of the ribbon, prevent the ribbon from being misaligned due to tension fluctuations or mechanical vibration during the ribbon transportation process, and ensure that the edge of the ribbon is always aligned with the edge of the shoe's piping, avoiding risks such as misaligned seams, exposed soles, and overflowing edges. In addition, the inner wall of the ribbon positioning hole 2 is smoothed to reduce the frictional resistance during ribbon transportation. The first positioning hole 4 and the second positioning hole 5 are preferably two in number. The slot 3, the two first positioning holes 4 and the two second positioning holes 5 can be matched with the external structure (axe cover) to ensure the stability of the ribbon positioning device and the sewing needle, avoiding problems such as skewed sewing, skipped stitches and missing stitches, and providing a guarantee for the sewing process.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention.

Claims

1. A control method for a ribbon positioning device in a shoe edge-binding process, characterized in that, include: The historical average sewing error and historical average alignment error of the ribbon positioning device are obtained. Based on the historical average sewing error and historical average alignment error, the error reference value of the ribbon positioning device is determined. Based on the error reference value and the tension control model, the tension range of the ribbon positioning device is determined. The ribbon is threaded into the ribbon positioning device, and the tension measurement value of the ribbon positioning device is obtained. The tension stability of the ribbon positioning device is determined based on the relationship between the tension measurement value and the tension range. When the tension stability is determined to be unqualified, a historical decision or a model decision is determined based on the traversal results of the tension measurement value in the historical database. When the historical decision is determined, a historical reuse decision or a historical clustering decision is determined based on the data similarity of the tension measurement value in the historical database. When the historical reuse decision is determined, a tension calibration value is determined based on the historical database. When the historical clustering decision is determined, a tension calibration value is determined based on a clustering algorithm. When the model decision is determined, a tension calibration value is determined based on a tension positioning model. The ribbon is straightened according to the tension calibration value, and the sewing needle is aligned with the ribbon positioning device to complete the edge rolling process.

2. The control method for the ribbon positioning device in the shoe edge-binding process according to claim 1, characterized in that, Determining the tension range of the ribbon positioning device includes: acquiring a historical sewing dataset and dividing the historical sewing dataset into a training set and a test set; pre-acquiring a machine learning model; iteratively training the machine learning model based on the training set; validating the iteratively trained machine learning model based on the test set; determining the tension control model; and substituting the error benchmark value into the tension control model to determine the tension range of the ribbon positioning device.

3. The control method for the ribbon positioning device in the shoe edge-binding process according to claim 2, characterized in that, When determining the tension range of the ribbon positioning device, the method further includes: if the accuracy of the machine learning model after the current iteration is less than the accuracy of the machine learning model after the previous iteration, then cosine annealing is used to adjust the learning rate of the machine learning model after the current iteration, and iterative training continues until the tension control model is determined; if the accuracy of the machine learning model after the current iteration is greater than or equal to the accuracy of the machine learning model after the previous iteration, then iterative training is stopped, and the machine learning model after the current iteration is determined as the tension control model.

4. The control method for the ribbon positioning device in the shoe edge-binding process according to claim 3, characterized in that, When determining the tension stability of the ribbon positioning device based on the relationship between the tension measurement value and the tension range, the method includes: if the tension measurement value is within the tension range, the tension stability is determined to be qualified, and the hemming process is completed according to the tension measurement value; if the tension measurement value is not within the tension range, the tension stability is determined to be unqualified.

5. The control method for the ribbon positioning device in the shoe edge-binding process according to claim 4, characterized in that, When determining a historical decision or a model decision based on the traversal results of the tension measurement value in the historical database, the following steps are taken: the historical database includes several historical tension measurements and several historical tension calibration values, and each historical tension measurement value corresponds to a historical tension calibration value. When there is a historical tension measurement value in the historical database with a data similarity greater than a data similarity threshold, the historical decision is determined to be executed. When there is no historical tension measurement value in the historical database with a data similarity greater than a data similarity threshold, the model decision is determined to be executed.

6. The control method for the ribbon positioning device in the shoe edge-binding process according to claim 5, characterized in that, When determining a historical reuse decision or a historical clustering decision based on the data similarity of the tension measurement value in the historical database, the following steps are taken: if a historical tension measurement value with a data similarity of 1 exists in the historical database, then the historical reuse decision is determined to be executed; if a historical tension measurement value with a data similarity of 1 does not exist in the historical database, then the historical clustering decision is determined to be executed.

7. The control method for the ribbon positioning device in the shoe edge-binding process according to claim 6, characterized in that, When determining the tension calibration value based on the historical database, the process includes: counting the number of historical tension measurement values ​​with a data similarity of 1 in the historical database; if the number of historical data is 1, then the historical tension calibration value corresponding to the historical tension measurement value is determined as the tension calibration value; if the number of historical data is not 1, then the average of the historical tension calibration values ​​corresponding to several historical tension measurement values ​​with a data similarity of 1 is determined as the tension calibration value.

8. The control method for the ribbon positioning device in the shoe edge-binding process according to claim 7, characterized in that, When determining the tension calibration value based on the clustering algorithm, the process includes: using the tension measurement value and the historical database as a historical clustering dataset, determining the expected number of clusters and initializing the parameters of the Gaussian distribution, determining the probability that each data point in the historical clustering dataset belongs to each Gaussian distribution, determining the historical dataset corresponding to the tension measurement value based on the probability, and determining the mean of the historical tension calibration values ​​in the historical dataset as the tension calibration value.

9. The control method for the ribbon positioning device in the shoe edge-binding process according to claim 8, characterized in that, When determining the tension calibration value based on the tension positioning model, the process includes: constructing a model dataset based on the historical database, and determining a tension positioning model based on the model dataset, with the tension measurement value as the input and the tension calibration value as the output.

10. A ribbon positioning device for a shoe edge binding process, used in a control method for applying the ribbon positioning device for a shoe edge binding process as described in any one of claims 1-9, characterized in that, include: The positioning body is provided with a ribbon positioning hole for positioning the ribbon, and the positioning body is provided with a slot, a plurality of first positioning holes and a plurality of second positioning holes.

Images