A Multi-Parameter Coordination-Based Control Method and System for Polyester Fabric Weaving

Abstract

This invention provides a method and system for controlling polyester fabric weaving based on multi-parameter collaboration, relating to the field of textile control technology. The method includes: discretizing the combination space of multiple process parameters in the polyester fabric weaving process into multiple parameter arms; establishing multiple preset perturbation constraints for the multiple parameter arms; iteratively applying perturbations to the multiple parameter arms during the weaving process based on the multiple preset perturbation constraints; during the perturbation process, calculating the reward trend of the parameter arms by collecting real-time weaving quality index data of the multiple parameter arms when the perturbation is applied; and performing sequential adjustment of the parameter arms. This solves the technical problem of existing technologies relying on fixed optimal parameter control, ignoring the time-varying characteristics of the weaving process, leading to fabric quality fluctuations and production instability. By individually perturbing multiple process parameters, the method achieves continuous optimization of weaving parameters, improving the quality and production stability of polyester fabric.

Description

Technical Field

[0001] This invention relates to the field of textile control technology, specifically to a method and system for controlling polyester fabric weaving based on multi-parameter coordination. Background Technology

[0002] In existing polyester fabric weaving production, process control methods typically rely on fixed optimal parameters. The optimal combination of process parameters is determined through offline optimization experiments, historical production data analysis, and process specification requirements. For example, single-factor or multi-factor experimental designs are used to find the parameter combination that achieves the best fabric quality indicators. Simultaneously, historical production data statistics are combined to analyze the optimal parameters under different batches of raw materials and environmental conditions, forming recommended process values. These offline-obtained fixed recommended process values ​​are then locked into the production process, with the expectation that a single setup can achieve stable and high-quality fabric.

[0003] However, the weaving process is inherently time-varying. Affected by batch differences in raw materials, fluctuations in environmental temperature and humidity, and equipment wear, fabric quality fluctuates with slight changes in process conditions. Fixed-parameter methods cannot cope with sudden changes in production, such as fluctuations in yarn tension or minor adjustments to shed speed, leading to defects such as stripes and blemishes on the fabric surface, reducing product yield. Furthermore, because fixed parameters cannot dynamically adapt to the real-time state of the loom, production stability is difficult to guarantee, and optimization potential is limited to pre-set static process points. Therefore, existing methods cannot achieve continuous optimization and real-time quality control of the weaving process, affecting polyester fabric quality and production stability.

[0004] In summary, existing technologies rely on fixed optimal parameter control, ignoring the time-varying characteristics of the weaving process, which leads to technical problems such as fabric quality fluctuations and production instability. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for controlling polyester fabric weaving based on multi-parameter coordination, in order to solve the technical problems of existing technologies that rely on fixed optimal parameter control, ignore the time-varying characteristics of the weaving process, and cause fluctuations in fabric quality and unstable production.

[0006] In view of the above problems, this application provides a method and system for controlling polyester fabric weaving based on multi-parameter coordination.

[0007] The first aspect of this application provides a method for controlling polyester fabric weaving based on multi-parameter collaboration. The method includes: discretizing the combination space of multiple process parameters in the polyester fabric weaving process into multiple parameter arms; establishing multiple preset perturbation constraints for the multiple parameter arms; iteratively applying perturbations to the multiple parameter arms in the weaving process based on the multiple preset perturbation constraints; during the perturbation process, calculating the reward trend of the parameter arms by collecting real-time weaving quality index data of the multiple parameter arms when the perturbation is applied, and performing sequential adjustment of the parameter arms.

[0008] Optionally, during the steady-state operation of the loom, a single step disturbance is applied to the first process parameter corresponding to the first parameter arm with a preset conservative amplitude, and the first response sequence of the weaving quality index is continuously collected; based on the first response sequence, the transition time of the weaving quality from the moment the disturbance is applied to the moment of recovery to steady state is calculated, and the maximum deviation of the weaving quality index during the transition time is extracted; the maximum deviation is compared with a preset visible quality fluctuation threshold, and if the maximum deviation is less than the visible quality fluctuation threshold, the single step disturbance amplitude is used as the preset disturbance constraint of the first parameter arm, and the multiple preset disturbance constraints are established by traversing the multiple parameter arms.

[0009] Optionally, if the maximum deviation is greater than or equal to the visible quality fluctuation threshold, the single step disturbance is reduced by a preset ratio, and the step disturbance is reapplied until the step disturbance amplitude with the maximum deviation is less than the visible quality fluctuation threshold is obtained, which serves as the preset disturbance constraint for the first parameter arm.

[0010] Optionally, during the offline calibration process of the same fabric variety, multiple normal fabric samples and multiple fabric defect samples containing known defects are collected in advance; the uniformity quantification values ​​of the normal fabric samples and the fabric defect samples are extracted, and the visible quality fluctuation threshold is generated under the normal-defect separation constraint.

[0011] Optionally, a weaving quality sensitivity analysis is performed on the plurality of parameter arms to establish an initial selection weight distribution; a first parameter arm is selected from the plurality of parameter arms based on the initial selection weight; a perturbation is applied to the first parameter arm according to the corresponding preset perturbation constraint, and real-time weaving quality index data is collected during the perturbation process to analyze and explore the reward trend; if the explored reward trend meets the preset optimization constraint, the current perturbation result is maintained and the perturbation is continued to be applied to the first parameter arm until the explored reward trend does not meet the preset optimization constraint, the last perturbation is canceled and adjusted, and a second parameter arm is selected from the plurality of parameter arms based on the initial selection weight, and the same perturbation process as the first parameter arm is performed.

[0012] Optionally, the reward trend can be determined by calculating the changing trend of the weaving quality index during the disturbance process based on real-time weaving quality index data.

[0013] Optionally, after completing one round of adjustment of the plurality of parameter arms, the initial selection weight distribution is adjusted as follows: the adjustment record of the plurality of parameter arms is obtained; the degree of quality optimization of the plurality of parameter arms is compared based on the adjustment record to obtain the quality optimization ratio; a quality selection weight distribution is generated based on the quality optimization ratio, and an updated selection weight distribution is generated by weight fusion with the initial selection weight distribution; the next round of sequential adjustment of the parameter arms is performed according to the updated selection weight distribution.

[0014] Optionally, if the quality optimization degree of any parameter arm is lower than the preset threshold in multiple consecutive sequential adjustments or if no disturbance optimization is performed in multiple consecutive sequential adjustments, the corresponding parameter arm will be shielded, and the shielding will be lifted after one loom start-up and shutdown.

[0015] Optionally, when the quality selection weight distribution and the initial selection weight distribution are weighted and fused, the fusion coefficient of the quality selection weight distribution is greater than the fusion coefficient of the initial selection weight distribution.

[0016] A second aspect of this application provides a polyester fabric weaving control system based on multi-parameter collaboration. The system includes: a parameter discretization module for discretizing the combination space of multiple process parameters in the polyester fabric weaving process into multiple parameter arms; a disturbance constraint establishment module for establishing multiple preset disturbance constraints on the multiple parameter arms; and a parameter arm adjustment module for iteratively applying disturbances to the multiple parameter arms during the weaving process based on the multiple preset disturbance constraints. During the disturbance process, the system calculates the reward trend of the parameter arms by collecting real-time weaving quality index data of the multiple parameter arms when the disturbance is applied, and performs sequential adjustment of the parameter arms.

[0017] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0018] The method provided in this application discretizes the combination space of multiple process parameters in the polyester fabric weaving process into multiple parameter arms; establishes multiple preset perturbation constraints for the multiple parameter arms; iteratively applies perturbations to the multiple parameter arms during the weaving process based on the multiple preset perturbation constraints; during the perturbation process, the reward trend of the parameter arms is calculated by collecting real-time weaving quality index data of the multiple parameter arms when the perturbation is applied, and the sequential adjustment of the parameter arms is performed. This achieves the technical effect of continuously optimizing weaving parameters and improving the quality and production stability of polyester fabric by individually perturbing multiple process parameters.

[0019] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0021] Figure 1 A flowchart illustrating the polyester fabric weaving control method based on multi-parameter collaboration provided in this application.

[0022] Figure 2 A schematic diagram of the structure of the polyester fabric weaving control system based on multi-parameter coordination provided in this application.

[0023] Explanation of reference numerals in the attached figures: Parameter discretization module 11, disturbance constraint establishment module 12, parameter arm adjustment module 13. Detailed Implementation

[0024] This application provides a method and system for controlling polyester fabric weaving based on multi-parameter coordination. It addresses the technical problem of existing technologies relying on fixed optimal parameters for control, neglecting the time-varying characteristics of the weaving process, leading to fluctuations in fabric quality and production instability. The method achieves the technical effect of continuously optimizing weaving parameters by individually perturbing multiple process parameters, thereby improving polyester fabric quality and production stability.

[0025] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0026] Example 1, as Figure 1As shown, this application provides a polyester fabric weaving control method based on multi-parameter coordination, the method comprising: The combination space of multiple process parameters in the polyester fabric weaving process is discretized into multiple parameter arms.

[0027] Specifically, based on the characteristics of polyester fabric weaving and the loom control system, several process parameters are determined during the weaving process, such as weft density, warp density, yarn tension, loom shed speed, and wet spray volume. Specifically, weft density and warp density are measured in real-time by data collected from weft density and warp density sensors via the loom control system PLC or industrial control software, with the unit being threads / cm. Yarn tension is obtained through a tension sensor, with the unit being Newtons. Shed speed is provided by the loom speed feedback system, with the unit being meters / minute. Wet spray volume is obtained through a humidity control device or liquid metering device, with the unit being milliliters / meter.

[0028] Then, the combination of multiple process parameters is spatially discretized, treating each process parameter as an independent parameter arm, i.e., each parameter arm corresponds to one process parameter, used for subsequent application of single step perturbation and sequential optimization. By treating each process parameter as a separate parameter arm, independent perturbation and sensitivity analysis can be performed on each process parameter, thereby achieving multi-parameter collaborative optimization of the entire weaving process while maintaining fabric quality safety.

[0029] Multiple preset perturbation constraints are established for the multiple parameter arms.

[0030] Furthermore, multiple preset disturbance constraints are established for the multiple parameter arms, including: during the steady-state operation of the loom, applying a single step disturbance to the first process parameter corresponding to the first parameter arm with a preset conservative amplitude, and continuously collecting the first response sequence of the weaving quality index; based on the first response sequence, calculating the transition time of the weaving quality from the moment the disturbance is applied to the moment of recovery to steady state, and extracting the maximum deviation of the weaving quality index within the transition time; comparing the maximum deviation with a preset visible quality fluctuation threshold, and if the maximum deviation is less than the visible quality fluctuation threshold, then using the single step disturbance amplitude as the preset disturbance constraint of the first parameter arm, and traversing the multiple parameter arms to establish the multiple preset disturbance constraints.

[0031] Furthermore, the setting of the visible quality fluctuation threshold includes: collecting multiple normal fabric samples and multiple fabric defect samples containing known defects during the offline calibration process of the same weaving variety; extracting the uniformity quantification values ​​of the normal fabric samples and the fabric defect samples, and generating the visible quality fluctuation threshold under the normal-defect separation constraint.

[0032] Specifically, the loom is placed in a steady-state operation phase, which refers to a phase where weaving quality indicators and process parameters tend to stabilize. This phase involves continuously collecting quality indicators such as evenness, fabric smoothness, or defect rate during each weaving cycle using the loom's real-time sensors or an online fabric inspection system, recording these data to form a time series. For example, the loom fabric surface is divided into several regions at fixed sampling intervals, such as once per meter. The yarn spacing in each region is measured, and the local evenness value at each sampling point is calculated. This local evenness value is quantified by measuring changes in local fabric density or grayscale distribution using an online fabric inspection system or optical sensors. By calculating the average density or brightness value for each region, the mean and standard deviation of all regions are calculated. A smaller standard deviation indicates a more uniform fabric surface, i.e., higher evenness. The evenness index is defined as the ratio of the standard deviation to the mean. For example, if the mean density is 30 yarns / cm and the standard deviation is 0.6 yarns / cm, then the evenness = 0.6 / 30 ≈ 0.02, thus obtaining the evenness index at each sampling time. For each weaving quality indicator, the difference or standard deviation between the maximum and minimum values ​​within a certain time window, such as 5 minutes, is calculated as the fluctuation range. This is then compared with a historical normal range threshold, which can be obtained through offline calibration: multiple sample data points of the same weaving variety within the allowable range of the process and with normal fabric quality are statistically analyzed, and the 95th percentile of the evenness fluctuation is calculated as the upper limit. For example, if historical calibration shows that the evenness fluctuation is between 0.00 and 0.04, 0.04 is selected as the threshold. If the evenness fluctuation is less than or equal to 0.04 during a continuous 5-minute data collection period, the loom is considered to have reached a steady state.

[0033] During the steady-state operation of the loom, a single step disturbance is applied to the first process parameter corresponding to the first parameter arm with a preset conservative amplitude. The first parameter arm refers to one of several parameter arms obtained through discretization. For example, the first parameter arm may correspond to the warp tension parameter, and the corresponding first process parameter, such as yarn tension, is extracted. The preset conservative amplitude refers to the adjustment range of the process parameter limited to ensure that no visible defects occur in the fabric quality when the disturbance is applied; it is a small change within a safe range. For example, the first process parameter may be +3% of the baseline value. This adjustment range is determined by combining historical production data and process experience, and can be verified through offline testing to determine whether the fabric quality fluctuation is within an acceptable range. A single step disturbance refers to applying a parameter offset of a fixed amplitude at once, rather than a gradual change, and is used to assess the sensitivity of process parameters to weaving quality. During the application of a step disturbance, the weaving quality indicators are continuously collected by real-time sensors on the loom, such as weft density sensors, warp density sensors, tension sensors, and fabric optical detection devices, until the fabric quality converges back to a steady state. These data are stored in chronological order to form a first response sequence. This first response sequence can be represented as a continuous change in time-quality indicators. For example, evenness is sampled once per second after the disturbance, forming a curve of evenness changing over time. This curve is used to calculate the maximum deviation and transition time, thereby assessing whether the disturbance amplitude is within a safe range. For instance, after applying a +0.5 weft density step disturbance of 31 weft ends / cm, the evenness change sequence sampled once per second is as follows: 0.020, 0.021, 0.022, 0.022, 0.021, 0.020. The maximum deviation caused by the disturbance is 0.002. This sequence is recorded as the first response sequence to determine whether a preset visible quality fluctuation threshold is met.

[0034] Based on the first response sequence, the transition time of the weaving quality from the moment of disturbance application to the recovery to steady state is calculated. Simultaneously, the maximum change in the weaving quality index relative to the steady-state weaving quality index during the transition time is calculated; that is, the maximum absolute value of the difference between the collected weaving quality index value and the steady-state weaving quality index value before the disturbance. The maximum deviation is compared with a preset visible quality fluctuation threshold, which is obtained through offline calibration: multiple normal fabric samples and fabric defect samples containing known defects of the same weaving variety are pre-collected. Normal fabric samples refer to fabric segments without obvious defects in appearance within the allowable range of the process; each segment can be 10-50 meters long. Images or weaving density data are continuously acquired using optical scanning or online fabric inspection devices. Fabric defect samples introduce small, controllable defects during the weaving process or through manual marking, such as stripes, horizontal bars, or localized abnormal yarn tension. Each fabric segment is also optically scanned or sensor-collected, recording the fabric location and defect type.

[0035] After data acquisition, the fabric image or sensor data is segmented into several small grid units, such as one unit per square centimeter. The average weave density or grayscale value of each unit is calculated, and then the mean and standard deviation of the entire fabric surface are calculated. The uniformity quantification values, i.e., the ratio of the standard deviation to the mean, are extracted for normal fabric samples and fabric defect samples. Through calculation, a set of uniformity index sequences is obtained for both normal fabric samples and fabric defect samples.

[0036] A visible quality fluctuation threshold is generated under the constraint of separating normal and defective fabrics. This constraint means that the set visible quality fluctuation threshold should meet the following requirements: the deviation in uniformity caused by disturbance should not exceed the visible quality fluctuation threshold, thus preventing misjudgment as a fabric defect, while also covering the fluctuation range of normal fabric surfaces as much as possible. The uniformity distribution of normal fabric samples can be compared with that of defective fabric samples, and a safe dividing point between the two can be selected as the threshold. For example, if the upper limit of uniformity for normal fabric surfaces is 0.025 and the lower limit for uniformity for defective fabric surfaces is 0.030, setting the visible quality fluctuation threshold to 0.028 ensures that a deviation less than 0.028 caused by disturbance will not be identified as a fabric defect.

[0037] If the deviation is less than the visible quality fluctuation threshold obtained from offline calibration, meaning the disturbance will not cause fabric defects that can be identified by the naked eye or detection system, then the applied step amplitude is determined as the preset disturbance constraint for that parameter arm. This means that the process parameters of that parameter arm can be safely adjusted within this range during subsequent optimization and sequential disturbance processes. Then, the same operation is repeated for other parameter arms in sequence, traversing the entire parameter arm set, and determining the preset disturbance constraint for each parameter arm one by one, providing a reliable safety boundary for multi-parameter collaborative optimization. The preset disturbance constraint is the amplitude that does not cause visible fabric quality fluctuations.

[0038] By determining the safe disturbance range for each parameter arm, it is ensured that the disturbance will not cause significant fabric defects during the multi-parameter collaborative optimization process, thereby achieving safe, controllable, and continuous online optimization and improving the effectiveness and safety of polyester fabric weaving control.

[0039] Furthermore, if the maximum deviation is greater than or equal to the visible quality fluctuation threshold, the single step disturbance is reduced by a preset ratio, and the step disturbance is reapplied until the step disturbance amplitude with the maximum deviation is less than the visible quality fluctuation threshold is obtained, which serves as the preset disturbance constraint for the first parameter arm.

[0040] Specifically, when a single step disturbance is applied to a parameter arm during the steady-state operation of the loom, if the maximum deviation calculated from the first response sequence formed by real-time acquisition of weaving quality indicators is greater than or equal to the visible quality fluctuation threshold, it indicates that the current step amplitude is too large and may cause visible defects in the fabric or quality fluctuations exceeding the acceptable range. The single step disturbance is then scaled down proportionally by multiplying it by a coefficient less than 1, such as 0.8 or 0.7, reducing the single step disturbance to 80% or 70% of its original amplitude. Then, the scaled-down single step disturbance is applied again, and the response sequence is repeatedly acquired to calculate the maximum deviation.

[0041] After each reduction in amplitude, a step perturbation is reapplied, and the transition time and maximum deviation are recalculated until the maximum deviation caused by the perturbation is less than the visible quality fluctuation threshold. This step perturbation is then used as a preset perturbation constraint for the first parameter arm, allowing for safe fine-tuning within this amplitude range during subsequent online perturbations and sequential optimization without causing visible fabric defects. The entire process is performed independently for each parameter arm, forming preset perturbation constraints for multiple parameter arms, providing a safety boundary for multi-parameter collaborative optimization.

[0042] When the current step amplitude is determined to be too large, the single step disturbance is reduced by a preset ratio and the calculation is repeated to ensure that the disturbance amplitude of each parameter arm is continuously and safely adjusted within a safe range. This avoids excessive disturbances during online optimization that could lead to abnormal fabric quality, thereby ensuring that multi-parameter collaborative optimization is both safe and effective, improving the effectiveness and safety of polyester fabric weaving control, and simultaneously improving the production quality and reliability of polyester fabric.

[0043] Based on the multiple preset perturbation constraints, the multiple parameter arms are perturbed iteratively applied to the weaving process. During the perturbation process, the reward trend of the parameter arms is calculated by collecting real-time weaving quality index data of the multiple parameter arms when the perturbation is applied, and the sequential adjustment of the parameter arms is executed.

[0044] Furthermore, based on the multiple preset perturbation constraints, perturbations are iteratively applied to the multiple parameter arms during the weaving process. During the perturbation process, real-time weaving quality index data of the multiple parameter arms at the time of perturbation are collected to calculate the reward trend of the parameter arms and perform sequential adjustment of the parameter arms, including: performing weaving quality sensitivity analysis on the multiple parameter arms and establishing an initial selection weight distribution; selecting a first parameter arm from the multiple parameter arms based on the initial selection weight; applying perturbation to the first parameter arm according to the corresponding preset perturbation constraints, and collecting real-time weaving quality index data during the perturbation process to analyze and explore the reward trend; if the explored reward trend meets the preset optimization constraints, maintaining the current perturbation result and continuing to apply perturbation to the first parameter arm until the explored reward trend does not meet the preset optimization constraints, canceling the last perturbation adjustment, selecting a second parameter arm from the multiple parameter arms based on the initial selection weight, and performing the same perturbation process as the first parameter arm.

[0045] Furthermore, the reward trend is determined by calculating the changing trend of the weaving quality index during the disturbance process based on real-time weaving quality index data.

[0046] Specifically, during the weaving process, each parameter arm, i.e., each process parameter such as weft density, warp density, yarn tension, shed speed, and spray amount, is perturbed as an independent unit. Perturbations are not applied simultaneously between different parameter arms to avoid interactive effects. First, a weaving quality sensitivity analysis is performed on multiple parameter arms. By comparing the change in uniformity index before and after a single step perturbation within the visible quality fluctuation threshold, the sensitivity of each parameter arm to weaving quality is quantified. That is, weaving quality sensitivity = uniformity after perturbation - uniformity before perturbation. After obtaining the sensitivity data for each parameter arm, the sensitivity values ​​are converted into initial selection weights. A normalization method is used, such as dividing the sensitivity amplitude of each parameter arm by the sum of the sensitivity amplitudes of all parameter arms, so that the weight sums to 1, establishing an initial selection weight distribution. A larger weight indicates a more significant impact of the parameter arm on weaving quality, and it will be preferentially selected in subsequent sequential perturbations. For example, the current weft density is 31 threads / cm, with a safe disturbance range of ±0.5 threads / cm; the current yarn tension is 10 Newtons, with a safe disturbance range of ±0.5 Newtons; and the current shed speed is 120 m / min, with a safe disturbance range of ±5 m / min. First, a step perturbation of +0.2 threads / cm is applied to the weft density, and the uniformity is collected in real time as it decreases from 0.020 to 0.018. The weft density weaving quality sensitivity is 0.002. A step perturbation of +0.2 Newtons is applied to the yarn tension, and the uniformity decreases from 0.020 to 0.019. The yarn tension weaving quality sensitivity is 0.001. A perturbation of +2 m / min is applied to the shed speed, and the uniformity decreases from 0.020 to 0.0195. The shed speed weaving quality sensitivity is 0.0005. After normalization, the initial selection weights are obtained: weft density weight is 0.57, yarn tension weight is 0.29, and shed speed weight is 0.14, forming the initial selection weight distribution.

[0047] Based on the initial selection weight distribution, the first parameter arm with the largest weight is selected sequentially from multiple parameter arms, and step perturbation and reward trend optimization are performed. For the first parameter arm, a single step perturbation is applied according to its corresponding preset perturbation constraint, and weaving quality index data is continuously collected through the loom's real-time sensors or online fabric inspection system. The reward trend is explored based on the collected weaving quality index data. The explored reward trend is determined based on the changing trend of the weaving quality index during the perturbation process. That is, the improvement of the weaving quality index after the perturbation is calculated minus the deterioration as the reward value. For example, the change of the evenness index sampled per second relative to the steady-state value before the perturbation is calculated to form the curve of the exploration reward value over time, which is the exploration reward trend. If there are multiple weaving quality indicators, such as evenness, fabric smoothness, and defect rate, the comprehensive value of the improvement of each indicator after the perturbation minus the deterioration is calculated as the reward value, forming the curve of the total exploration reward value over time, which is the exploration reward trend.

[0048] If the exploration reward trend satisfies the preset optimization constraints (i.e., the total reward value is positive or exceeds the set reward threshold), the current perturbation parameter value is maintained, the current step perturbation result is used as the new steady-state parameter, and the next step perturbation is applied to that parameter arm to further explore the quality improvement space. If, during continuous perturbation, the exploration reward trend does not satisfy the preset optimization constraints (e.g., the total reward value is negative or any quality index exceeds the allowable range), the last perturbation adjustment is canceled, and the perturbation parameter is restored to the previous optimal value to avoid fabric quality deterioration.

[0049] The reward threshold can be preset based on historical production data and the allowable optimization range of the process. For example, during offline calibration, multiple step perturbations are applied to a parameter arm, the maximum improvement in weaving quality indicators is recorded, and 60% to 80% of this improvement is taken as the threshold to ensure that the perturbation has practical optimization significance without causing fabric defects. Any quality indicator exceeding the allowable range refers to the weaving quality indicator collected in real time after the perturbation, such as evenness, fabric flatness, or defect rate, exceeding the preset process tolerance value. For example, evenness should not exceed 0.04, defect rate should not exceed 0.001, and flatness should not exceed 0.5mm. Fabric flatness can be obtained in real time through an online fabric inspection system. By installing a high-resolution laser displacement sensor or optical 3D scanning device at the loom exit, the fabric height information is continuously scanned along the width and length directions of the fabric surface, forming a time series data of fabric height. Statistical analysis is performed on this time series data of fabric height, such as calculating the standard deviation of the fabric height, quantifying the degree of fabric undulation, and using the standard deviation value as a flatness index. The smaller the value, the flatter the fabric surface.

[0050] Defect rates are quantified using high-speed visual inspection equipment for fabric surfaces. CCD or CMOS cameras can be installed at the loom exit or subsequent tension rollers. Fabric images are acquired using natural or structured light illumination. Image processing algorithms are used to identify fabric defect types. The high-speed camera captures fabric images, and a weighted average method is used to synthesize grayscale values ​​from the RGB channels according to human visual perception weights. Specifically, the fabric image is grayscaled using 0.2989×R + 0.5870×G + 0.1140×B, converting color or color texture information into a single-channel grayscale image. The grayscale image is then binarized, and a threshold is set to distinguish normal areas from potential defect areas. For example, if the average grayscale value of the fabric is 180, and fine stripes or blemishes are around 150, a threshold of 165 is set. Pixels with values ​​less than 165 are marked as defect areas, while those greater than or equal to 165 are marked as normal fabric. Edge detection algorithms, such as the Canny operator, are used to extract the contours of potential defects on the fabric surface. Combined with morphological filtering, dilation, erosion, opening, and closing operations are performed on the extracted contours to remove noise and fill gaps, enhancing defect features. Contour analysis determines the location and area of ​​each defect, and the ratio of the defect area to the total detected area is calculated to obtain the defect rate, enabling automatic identification and quantification of fabric defects. For example, if the total pixel area of ​​the fabric surface is 10,000 pixels, and a defective area occupying 100 pixels is detected, the defect rate is 1%.

[0051] After the first parameter arm is optimized, a second parameter arm is selected from among the multiple parameter arms based on the initial selection weights. The same step perturbation, real-time data acquisition and reward trend analysis process as the first parameter arm is performed. Sequential optimization is then performed on each parameter arm in turn, thereby achieving independent, continuous and dynamic optimization of multiple parameters.

[0052] For example, with a current steady-state density of 31 fibers / cm, a safe disturbance range of ±0.5 fibers / cm, an initial step disturbance of +0.2 fibers / cm, and a collected uniformity decreasing from 0.020 to 0.018, a defect rate changing from 0.0005 to 0.0006, the total reward value is 0.002. 0.0001 = 0.0019, a positive value, exceeding the reward threshold of 0.001, satisfying the optimization constraint. The perturbation is maintained and the next +0.1 strands / cm perturbation continues. Real-time uniformity is 0.021, defect rate is 0.0012, and total reward value = If the value is 0.0012, the optimization constraint is not met. Therefore, the +0.1 root / cm perturbation is cancelled, and the weft density parameter is restored to the optimal value of 31.2 root / cm. Then, the second parameter arm is selected according to the initial selection weight and the same sequential perturbation optimization is performed.

[0053] Sensitivity analysis and reward trend optimization were used to achieve single-parameter safe perturbation and sequential adjustment. This ensured that each polyester fabric weaving process parameter was optimized one by one, continuously and safely without causing visible fabric defects. The parameters in the weaving process dynamically drifted near the optimal working condition, achieving multi-parameter collaborative optimization and continuous quality improvement. At the same time, it reduced the risk of mutual interference that may be caused by simultaneous perturbation of multiple parameters, thereby improving the quality and production stability of polyester fabric.

[0054] Furthermore, after completing one round of adjustment of the multiple parameter arms, the initial selection weight distribution is adjusted as follows: the adjustment record of the multiple parameter arms is obtained; the degree of quality optimization of the multiple parameter arms is compared based on the adjustment record to obtain the quality optimization ratio; a quality selection weight distribution is generated based on the quality optimization ratio, and an updated selection weight distribution is generated by weight fusion with the initial selection weight distribution; the next round of sequential adjustment of the parameter arms is performed according to the updated selection weight distribution.

[0055] Furthermore, when the quality selection weight distribution and the initial selection weight distribution are weighted and fused, the fusion coefficient of the quality selection weight distribution is greater than the fusion coefficient of the initial selection weight distribution.

[0056] Specifically, after completing one round of sequential adjustment for multiple parameter arms, the adjustment records for each parameter arm are obtained. These records include the final parameter value applied during the perturbation process, the corresponding real-time weaving quality index data, and the exploration reward value calculated during the perturbation. Based on these adjustment records, the degree of quality optimization for each parameter arm is compared: the improvement in weaving quality index before and after the perturbation is calculated for each parameter. If multiple weaving quality indices exist, a weighted average method is used to calculate the overall improvement. The improvement of each parameter arm is normalized to the overall improvement of all parameter arms in this round to obtain the quality optimization ratio of that parameter arm in one round of optimization. That is, the quality optimization ratio = improvement of that parameter arm / overall improvement of all parameter arms.

[0057] The quality optimization ratio is used as the quality selection weight distribution for the parameter arm in the next round of perturbation selection, and this weight is fused with the initial selection weight distribution at the start of this round of adjustment. Weight fusion is calculated separately for each parameter arm using a weighted method: Updated Weight = α × Initial Selection Weight + β × Quality Selection Weight, where α and β are fusion coefficients, α + β = 1, and the fusion coefficient of the quality selection weight distribution is greater than the fusion coefficient of the initial selection weight, such as α = 0.3 and β = 0.7, prioritizing real-time quality feedback to address changes in the production environment and raw material batches. The updated selection weight distribution obtained after fusion is used for the sequential control adjustment of the parameter arms in the next round: parameter arms with higher weights are preferentially selected in the next round of perturbation, thereby achieving dynamic, real-time multi-parameter optimization.

[0058] By providing feedback on the optimization effects of the previous round of adjustments, the priority of the next round of sequential disturbances is dynamically adjusted to ensure that, under real-time loom operation conditions, the optimization direction of each parameter arm always revolves around improving fabric quality and production stability, thereby achieving continuous, minute, and safe parameter optimization and improving the weaving quality and production stability of polyester fabric.

[0059] Furthermore, the method includes: if the quality optimization degree of any parameter arm is lower than a preset threshold in multiple consecutive sequential adjustments or if no disturbance optimization is performed in multiple consecutive sequential adjustments, then the corresponding parameter arm is shielded, and the shielding is removed after one loom start-up and shutdown is completed.

[0060] Specifically, during the sequential adjustment of multiple parameter arms, for each parameter arm, the exploration reward trend and the corresponding quality optimization degree of each round of perturbation are recorded, i.e., the quantitative value of the improvement in weaving quality indicators before and after the perturbation. If a parameter arm's overall quality optimization ratio is consistently lower than the preset optimization threshold (e.g., 0.01) in multiple rounds of sequential perturbation (e.g., set to 3 rounds based on production experience), it indicates that the parameter arm contributes less than 1% to weaving quality in each round. Alternatively, if the perturbation is canceled due to the exploration reward trend not meeting the preset optimization constraints during multiple rounds of perturbation, meaning the parameter arm has not generated any effective optimization, then the parameter arm is marked as an inefficient parameter arm.

[0061] Inefficient parameter arms are shielded, meaning they are temporarily excluded from perturbation in the next round and subsequent sequential adjustments to avoid wasting loom running time or causing unnecessary quality fluctuations. During the shielding period, sequential perturbation and optimization continue for other parameter arms to ensure overall optimization efficiency. The shielding technology can lock or ignore the perturbation commands of the parameter arm through the control system or PLC software, while still collecting its quality index data to monitor production status. To prevent the shielding from affecting the parameter optimization space in the long term, a loom start-up and shutdown operation is specified as the condition for releasing the parameter arm shielding. After the start-up and shutdown, the shielded parameter arm returns to a perturbable state, allowing it to re-participate in the next round of sequential adjustments. Loom start-up and shutdown not only physically reset the equipment but also ensure the reset of the process state, allowing for the rematching of raw material batches or environmental changes, enabling previously shielded parameter arms to be re-optimized under the new steady-state conditions.

[0062] By shielding parameter arms with low optimization effects, the fabric quality fluctuations caused by ineffective disturbances in the production process are avoided. At the same time, the shielding is removed by combining the start and stop of the loom, so that the polyester fabric weaving optimization control has adaptive capabilities. In the dynamic production environment, it continuously focuses on effective parameters, thereby improving the overall efficiency of multi-parameter sequential adjustment and the stability of weaving quality.

[0063] In summary, the integrated construction method for railway signaling equipment provided in this application has the following technical effects: by treating each process parameter in the polyester fabric weaving process as an independent parameter arm, and establishing a preset disturbance constraint for each parameter arm that does not cause fluctuations in the visible fabric quality, the exploration reward trend of each parameter arm is dynamically calculated using iterative sequential disturbance and real-time quality data acquisition, and the parameter arms are continuously adjusted so that each parameter is always finely adjusted near the optimal working condition during the weaving process, thereby achieving continuous optimization of fabric quality and improving the stability and adaptability of the production process, realizing real-time, minute, and safe multi-parameter collaborative optimization.

[0064] Example 2 is based on the same inventive concept as the multi-parameter collaborative polyester fabric weaving control method in the previous examples, such as... Figure 2 As shown, this application provides a polyester fabric weaving control system based on multi-parameter coordination, wherein the polyester fabric weaving control system based on multi-parameter coordination includes: The parameter discretization module 11 is used to discretize the combination space of multiple process parameters in the polyester fabric weaving process into multiple parameter arms; the perturbation constraint establishment module 12 is used to establish multiple preset perturbation constraints for the multiple parameter arms; the parameter arm adjustment module 13 is used to iteratively apply perturbation to the multiple parameter arms in the weaving process based on the multiple preset perturbation constraints. During the perturbation process, the parameter arms are calculated by collecting real-time weaving quality index data of the multiple parameter arms when the perturbation is applied, and the parameter arms are sequentially adjusted.

[0065] Furthermore, the disturbance constraint establishment module 12 is also used to: apply a single step disturbance to the first process parameter corresponding to the first parameter arm with a preset conservative amplitude during the steady-state operation of the loom, and continuously collect the first response sequence of the weaving quality index; based on the first response sequence, calculate the transition time of the weaving quality from the moment of disturbance application to the moment of recovery to steady state, and extract the maximum deviation of the weaving quality index within the transition time; compare the maximum deviation with a preset visible quality fluctuation threshold, and if the maximum deviation is less than the visible quality fluctuation threshold, use the single step disturbance amplitude as the preset disturbance constraint of the first parameter arm, and traverse the multiple parameter arms to establish the multiple preset disturbance constraints.

[0066] Furthermore, the disturbance constraint establishment module 12 is also used to: if the maximum deviation is greater than or equal to the visible quality fluctuation threshold, reduce the single step disturbance by a preset ratio, reapply the step disturbance, until the step disturbance amplitude with the maximum deviation being less than the visible quality fluctuation threshold is obtained, as the preset disturbance constraint of the first parameter arm.

[0067] Furthermore, the disturbance constraint establishment module 12 is also used to: collect multiple normal fabric samples and multiple fabric defect samples containing known defects in advance during the offline calibration process of the same weaving variety; extract the uniformity quantification value of the normal fabric samples and the fabric defect samples, and generate the visible quality fluctuation threshold under the normal-defect separation constraint.

[0068] Furthermore, the parameter arm adjustment module 13 is also used to: perform weaving quality sensitivity analysis on the plurality of parameter arms and establish an initial selection weight distribution; select a first parameter arm from the plurality of parameter arms based on the initial selection weight; apply a perturbation to the first parameter arm according to the corresponding preset perturbation constraint, and collect real-time weaving quality index data during the perturbation process to analyze and explore the reward trend; if the explored reward trend meets the preset optimization constraint, maintain the current perturbation result and continue to apply perturbation to the first parameter arm until the explored reward trend does not meet the preset optimization constraint, cancel the last perturbation adjustment, select a second parameter arm from the plurality of parameter arms based on the initial selection weight, and perform the same perturbation process as the first parameter arm.

[0069] Furthermore, the parameter arm adjustment module 13 is also used to: explore reward trends and determine the changing trend of weaving quality indicators during the disturbance process based on real-time weaving quality indicator data.

[0070] Furthermore, the parameter arm adjustment module 13 is also used to: after completing one round of adjustment of the plurality of parameter arms, perform the adjustment of the initial selection weight distribution: obtain the adjustment record of the plurality of parameter arms in one round; compare the degree of quality optimization of the plurality of parameter arms based on the adjustment record of the one round to obtain the quality optimization ratio; generate a quality selection weight distribution with the quality optimization ratio, and generate an updated selection weight distribution by weight fusion with the initial selection weight distribution; and perform the next round of sequential adjustment of the parameter arms according to the updated selection weight distribution.

[0071] Furthermore, the parameter arm adjustment module 13 is also used to: if the quality optimization degree of any parameter arm is lower than a preset threshold in multiple consecutive sequential adjustments or if no disturbance optimization is performed in multiple consecutive sequential adjustments, then the corresponding parameter arm is shielded, and the shielding is released after one loom start-up and shutdown is completed.

[0072] Furthermore, the parameter arm adjustment module 13 is also used to: when performing weight fusion of the quality selection weight distribution and the initial selection weight distribution, the fusion coefficient of the quality selection weight distribution is greater than the fusion coefficient of the initial selection weight.

[0073] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The polyester fabric weaving control method and specific examples based on multi-parameter coordination in the foregoing embodiment one are also applicable to the polyester fabric weaving control system based on multi-parameter coordination in this embodiment. Through the foregoing detailed description of the polyester fabric weaving control method based on multi-parameter coordination, those skilled in the art can clearly understand the polyester fabric weaving control system based on multi-parameter coordination in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0074] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0075] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for controlling polyester fabric weaving based on multi-parameter coordination, characterized in that, include: The combination space of multiple process parameters in the polyester fabric weaving process is discretized into multiple parameter arms; Multiple preset perturbation constraints are established for the multiple parameter arms; Based on the multiple preset perturbation constraints, the multiple parameter arms are perturbed iteratively applied to the weaving process. During the perturbation process, the reward trend of the parameter arms is calculated by collecting real-time weaving quality index data of the multiple parameter arms when the perturbation is applied, and the sequential adjustment of the parameter arms is executed.

2. The polyester fabric weaving control method based on multi-parameter coordination as described in claim 1, characterized in that, Multiple preset perturbation constraints are established for the multiple parameter arms, including: During the steady-state operation of the loom, a single step disturbance is applied to the first process parameter corresponding to the first parameter arm with a preset conservative amplitude, and the first response sequence of the weaving quality index is continuously collected. Based on the first response sequence, the transition time of the weaving quality from the moment the disturbance is applied to the moment it returns to steady state is calculated, and the maximum deviation of the weaving quality index during the transition time is extracted. The maximum deviation is compared with a preset visible quality fluctuation threshold. If the maximum deviation is less than the visible quality fluctuation threshold, the amplitude of a single step disturbance is used as the preset disturbance constraint of the first parameter arm, and the multiple preset disturbance constraints are established by traversing the multiple parameter arms.

3. The polyester fabric weaving control method based on multi-parameter coordination as described in claim 2, characterized in that, If the maximum deviation is greater than or equal to the visible quality fluctuation threshold, the single step disturbance is reduced by a preset ratio, and the step disturbance is reapplied until the step disturbance amplitude with the maximum deviation is less than the visible quality fluctuation threshold is obtained, which serves as the preset disturbance constraint for the first parameter arm.

4. The polyester fabric weaving control method based on multi-parameter coordination as described in claim 2, characterized in that, The setting of the quality fluctuation threshold includes: In the offline calibration process of the same weaving variety, multiple normal fabric samples and multiple fabric defect samples containing known defects were collected in advance. Extract the uniformity quantification values ​​of the normal fabric sample and the fabric defect sample, and generate the visible quality fluctuation threshold under the normal-defect separation constraint.

5. The polyester fabric weaving control method based on multi-parameter coordination as described in claim 1, characterized in that, Based on the multiple preset perturbation constraints, perturbations are iteratively applied to the multiple parameter arms during the weaving process. During the perturbation process, real-time weaving quality index data of the multiple parameter arms at the time of perturbation are collected to calculate the reward trend of the parameter arms and perform sequential adjustment of the parameter arms, including: Perform weaving quality sensitivity analysis on the multiple parameter arms to establish an initial selection weight distribution; Based on the initial selection weights, the first parameter arm is selected from the plurality of parameter arms; The first parameter arm is subjected to a perturbation according to the corresponding preset perturbation constraint, and real-time weaving quality index data is collected during the perturbation process to analyze and explore the reward trend; If the exploration reward trend satisfies the preset optimization constraint, maintain the current perturbation result and continue to apply perturbation to the first parameter arm until the exploration reward trend no longer satisfies the preset optimization constraint. Then, cancel the last perturbation adjustment, select the second parameter arm from the plurality of parameter arms based on the initial selection weight, and perform the same perturbation process as the first parameter arm.

6. The polyester fabric weaving control method based on multi-parameter coordination as described in claim 5, characterized in that, The reward trend is determined by calculating the changing trend of the weaving quality index during the disturbance process based on real-time weaving quality index data.

7. The polyester fabric weaving control method based on multi-parameter coordination as described in claim 5, characterized in that, After completing one round of adjustment of the multiple parameter arms, the initial selection weight distribution is adjusted: Obtain the adjustment record of one round of the multiple parameter arms; Based on the comparison of the degree of quality optimization performed on the multiple parameter arms according to the one-round adjustment record, the quality optimization ratio is obtained; A quality selection weight distribution is generated based on the aforementioned quality optimization ratio, and an updated selection weight distribution is generated by weight fusion with the initial selection weight distribution. The next round of sequential adjustment of the parameter arms is performed based on the updated selection weight distribution.

8. The polyester fabric weaving control method based on multi-parameter coordination as described in claim 7, characterized in that, include: If the quality optimization degree of any parameter arm is lower than the preset threshold in multiple consecutive sequential adjustments, or if no disturbance optimization is performed in multiple consecutive sequential adjustments, the corresponding parameter arm will be shielded, and the shielding will be lifted after one loom start-up and shutdown.

9. The polyester fabric weaving control method based on multi-parameter coordination as described in claim 7, characterized in that, When the quality selection weight distribution is fused with the initial selection weight distribution, the fusion coefficient of the quality selection weight distribution is greater than the fusion coefficient of the initial selection weight distribution.

10. A polyester fabric weaving control system based on multi-parameter coordination, characterized in that, The steps for implementing the polyester fabric weaving control method based on multi-parameter coordination as described in any one of claims 1 to 9 include: The parameter discretization module is used to discretize the combination space of multiple process parameters in the polyester fabric weaving process into multiple parameter arms; The perturbation constraint establishment module is used to establish multiple preset perturbation constraints for the multiple parameter arms; The parameter arm adjustment module is used to iteratively apply perturbations to the multiple parameter arms during the weaving process based on the multiple preset perturbation constraints. During the perturbation process, the module collects real-time weaving quality index data of the multiple parameter arms when the perturbation is applied to calculate the reward trend of the parameter arms and execute the sequential adjustment of the parameter arms.

Images