A composite yarn production parameter intelligent control system

Abstract

The application discloses a kind of composite yarn production parameter intelligent control systems, it is related to production control technical field, including composite yarn body, still including acquisition module and analysis module, composite yarn body includes core yarn, inner cladding and outer cladding, and the twist direction of inner cladding and outer cladding is opposite, control system still includes acquisition module and analysis module, real-time detection data in the production process of composite yarn is obtained by acquisition module, the inner and outer layer fibers are locked by the reverse twisting structure in the application, even in severe stretching or bending, stainless steel fiber slip or breakage can be effectively prevented, the long-term stability of shielding effectiveness under electric field is ensured, the gap area between fiber and substrate is identified, the area ratio of suspended fiber is taken as an independent defect dimension into the evaluation system, the technical defects that existing two-dimensional image detection technology can only judge whether fiber covers or not, but it is difficult to judge whether fiber is in close contact with the lower layer.

Description

Technical Field

[0001] The present invention relates to the technical field of production control, and specifically to an intelligent control system for production parameters of composite yarns. Background Art

[0002] At present, with the regular development of live working on UHV transmission lines, the performance requirements for protective clothing are becoming increasingly stringent. The traditional live working shielding clothing mainly achieves the functions of conductivity and shielding through the overall blending or post-treatment of the fabric. The production method is simple blending or single-layer wrapping of stainless steel fibers and flame-retardant fibers such as aramid. This method requires increasing the proportion of stainless steel fibers, which will result in a hard, heavy and poorly breathable fabric, seriously affecting the activity flexibility and wearing comfort of operators. Therefore, it is difficult for existing yarns to optimize the uniformity and bonding fastness of the conductive layer while ensuring high strength and flame retardancy.

[0003] With the rapid development of emerging fields such as intelligent textiles, wearable electronic devices and electromagnetic shielding materials, composite yarns have become an important research direction in the field of textile materials. However, during the production process of such composite yarns, stainless steel fibers are prone to slipping or breaking, and the long-term stability of the shielding effectiveness under ultra-high voltage electric fields is easily affected. At the same time, it is necessary to accurately control multiple process parameters such as the wrapping coverage rate and yarn diameter of the inner and outer layers of stainless steel fibers. Otherwise, the conductive performance of the yarn may decrease. At present, the quality inspection during the production process of composite yarns mainly relies on manual sampling inspection or the method of judging by the threshold of a single sensor, through conventional means such as fiber coverage rate detection by a line scanning camera and laser displacement sensor detection.

[0004] This method mostly relies on detecting single parameters such as coverage rate by two-dimensional images or diameter by laser, and judges whether it exceeds the standard by setting a fixed threshold. However, two-dimensional images can only reflect whether the fiber covers the detection area and cannot identify whether the fiber is tightly adhered to the lower layer. In the wrapping structure of composite yarns, stainless steel fibers may be in a suspended state. Although visually covering the area, they do not actually participate in the conductive network. This ineffective coverage cannot be detected at all in conventional two-dimensional detection, resulting in a large number of yarn segments with conductive performance defects being misjudged as qualified products. Second, it lacks the ability to diagnose the root cause of defects. Existing technologies can only judge whether a certain parameter is abnormal and cannot analyze the correlation between different parameters, resulting in operators being difficult to judge the real source of the problem when facing alarm signals, leading to low efficiency in troubleshooting faults. Therefore, the existing composite yarn control system still needs to be improved. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent control system for production parameters of composite yarns, which solves the problems mentioned in the above background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent control system for composite yarn production parameters, comprising a composite yarn body, and further comprising a data acquisition module and an analysis module, wherein the composite yarn body comprises a core yarn, an inner covering layer and an outer covering layer, and the twisting directions of the inner covering layer and the outer covering layer are opposite;

[0007] The control system also includes a data acquisition module and an analysis module. The data acquisition module acquires real-time detection data during the composite yarn production process, including single-point detection data, intra-segment cumulative data, and risk fusion data. The acquired detection data is input into the analysis module, which includes a single-point detection unit, an intra-segment cumulative unit, and a risk fusion unit. The analysis module outputs a comprehensive defect risk warning coefficient.

[0008] It also includes an early warning module, which outputs a comprehensive defect risk early warning coefficient R in the analysis module. k Then, the early warning module outputs the alarm level according to the hierarchical alarm rules.

[0009] Optionally, the single-point detection related data includes the inner layer stainless steel fiber coverage rate, the outer layer stainless steel fiber coverage rate, the yarn diameter, the suspended fiber area, the total area of ​​the detection area, the inner layer standard coverage rate, the outer layer standard coverage rate, the standard yarn diameter, and the suspension penalty coefficient.

[0010] Optionally, the cumulative correlation data within the segment includes the number of detection points per evaluation segment, relative deviation of inner layer coverage, relative deviation of outer layer coverage, average deviation of inner layer within the segment, average deviation of inner and outer layers within the segment, cross-correlation of inner and outer layer deviations, correlation threshold, and correlation penalty intensity.

[0011] Optionally, the risk fusion-related data includes the average yarn surface temperature, standard process temperature, coefficient of thermal expansion of stainless steel, coefficient of thermal expansion of aramid, total length of stainless steel fibers within the window, length of core yarn within the window, normalized reference length, and minimum value for preventing and eliminating zero.

[0012] Optionally, the single-point detection unit outputs a single-point wrapping comprehensive deviation index based on the inner layer stainless steel fiber coverage, inner layer standard coverage, outer layer stainless steel fiber coverage, outer layer standard coverage, yarn diameter, standard yarn diameter, suspended fiber area, total detection area, and suspension penalty coefficient in the single-point detection related data. The single-point wrapping comprehensive deviation index is used to comprehensively evaluate the inner and outer layer coverage deviation, diameter deviation, and fiber suspension ratio at the i-th detection point to determine the severity of the local defect at that point.

[0013] Optionally, the segment accumulation unit outputs the segment defect accumulation coefficient based on the single-point wrapping comprehensive deviation index, the number of detection points per evaluation segment, the relative deviation of inner layer coverage, the relative deviation of outer layer coverage, the average deviation of inner layer within the segment, the average deviation of inner and outer layers within the segment, the cross-correlation of inner and outer layer deviations, the correlation threshold, and the correlation penalty intensity in the segment accumulation correlation data. The overall quality level of the segment is evaluated by the segment defect accumulation coefficient.

[0014] Optionally, the risk fusion unit outputs a comprehensive defect risk warning coefficient based on the most severe single-point defect within the window, the cumulative defect coefficient within the segment, the average yarn surface temperature, the standard process temperature, the thermal expansion coefficient of stainless steel, the thermal expansion coefficient of aramid, the total length of stainless steel fibers within the window, the core yarn length within the window, the normalized reference length, and the prevention and removal of zero minimum values ​​in the risk fusion related data. The most severe single-point defect, the cumulative defect within the segment, and the thermal expansion mismatch effect are fused into a comprehensive risk level through the comprehensive defect risk warning coefficient.

[0015] The most severe single-point defect within the window represents the maximum value of the single-point wrapping comprehensive deviation index within window k.

[0016] Optionally, the hierarchical alarm rule is as follows:

[0017] S1: When the comprehensive defect risk warning coefficient is less than 1.0, the alarm level is green, which means normal, and the system response is to continue production;

[0018] S2: When the comprehensive defect risk warning coefficient is greater than or equal to 1.0 and less than 2.0, the alarm level is yellow (i.e., pay attention), and the system response is to record data;

[0019] S3: When the comprehensive defect risk warning coefficient is greater than or equal to 2.0 and less than 3.0, the alarm level is orange, i.e., a warning, and the system response is to automatically fine-tune the tension or speed.

[0020] S4: When the comprehensive defect risk warning coefficient is greater than or equal to 3.0, the alarm level is red, which means an alarm is triggered. The system response is to immediately stop the machine and mark the location of the defect segment.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0022] I. This invention utilizes an aramid fiber core and a double stainless steel fiber reverse-winding structure, forming a denser and more uniform conductive network. The reverse twisting structure ensures that the inner and outer fiber layers are interlocked, effectively preventing the stainless steel fibers from slipping or breaking even under severe stretching or bending, thus ensuring long-term stability of shielding effectiveness under ultra-high voltage electric fields.

[0023] Meanwhile, aramid, acting as the core yarn, bears most of the mechanical load, preventing the stainless steel fibers from breaking due to direct stress. Reverse twisting balances the yarn torque, resulting in a smoother yarn, easier subsequent weaving processes, and better fabric resistance to pilling. Compared to traditional blended yarns, this wrapping structure can achieve or even exceed the same conductivity with fewer metal fibers. The metal fibers are concentrated on the surface, resulting in higher conductivity, thus retaining more of the aramid's flexibility, making the fabric softer and more breathable while maintaining rigidity.

[0024] Second, this invention identifies the void area between the fiber and the substrate, incorporating the area of ​​suspended fibers as an independent defect dimension into the evaluation system. This overcomes the limitation of existing two-dimensional image detection technologies, which can only determine whether a fiber covers the area but cannot determine whether the fiber is in close contact with the underlying layer. Although suspended fibers visually cover the area, they do not actually participate in the conductive network. This ineffective coverage is completely undetectable in conventional two-dimensional detection and can only be discovered through three-dimensional morphology reconstruction. Furthermore, suspended fibers may break or fray due to friction during subsequent weaving or wearing, forming potential sources of breakage in the conductive network. Early identification of this type of defect is of significant quality control importance.

[0025] Third, this invention upgrades the detection dimension from single-parameter detection to parameter relationship detection by introducing cross-correlation correction for inner and outer layer defects. The synergistic information of inner and outer layer coverage in the specific process context of double reverse wrapping has diagnostic value that conventional single-parameter detection cannot reveal. For example, if both inner and outer layers are simultaneously too high or too low, it may indicate a fluctuation in core yarn speed; when the core yarn slows down, the number of wrapping turns for both layers increases, and the coverage increases synchronously. When the inner layer is too high and the outer layer is too low, or vice versa, it may indicate an uneven yarn supply. Therefore, the system can not only identify the existence of defects but also preliminarily determine the possible causes of defects through correlation patterns, providing clear directional information for operators to adjust process parameters and significantly shortening troubleshooting time.

[0026] Fourth, by introducing the relevant effects of thermal expansion mismatch, this invention enables the system to distinguish between tension fluctuations caused by temperature and tension abnormalities caused by equipment failure, avoiding misjudging temperature effects as equipment failures and causing unnecessary downtime for inspection and equipment maintenance. At the same time, the system can identify the synergistic deterioration effect of temperature fluctuations and defects within the section, that is, the defects of yarns that are already of poor quality will be significantly aggravated under temperature fluctuations. Attached Figure Description

[0027] Figure 1 This is a flowchart of the present invention;

[0028] Figure 2 This is a hierarchical alarm rule diagram of the present invention;

[0029] Figure 3 This is a schematic diagram of the composite yarn body of the present invention.

[0030] In the diagram: 1. Core yarn; 2. Inner covering layer; 3. Outer covering layer. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Example 1:

[0033] Please see Figures 1 to 3 The present invention provides an intelligent control system for composite yarn production parameters, including a composite yarn body, which includes a core yarn 1, an inner covering layer 2 and an outer covering layer 3, wherein the twisting directions of the inner covering layer 2 and the outer covering layer 3 are opposite.

[0034] The inner covering layer 2 is made of a first stainless steel microfiber tightly wrapped around the core yarn 1 with a first twist direction. This layer mainly constructs the inner conductive network and initially fixes the core yarn 1. The first twist direction can be an S twist.

[0035] The outer coating layer 3 is formed by wrapping a second stainless steel microfiber filament around the yarn after the first step of treatment with a second twist direction opposite to the first twist direction. This layer not only further supplements the conductive path, but more importantly, it balances the internal stress of the yarn by reverse twisting, so that the two layers of stainless steel fibers interweave and are more tightly bound together. The second twist direction can be Z twist.

[0036] This structure, using aramid as the core and double stainless steel fibers twisted in reverse, creates a denser and more uniform conductive network. The reverse twisting structure interlocks the inner and outer fibers, effectively preventing slippage or breakage of the stainless steel fibers even under severe stretching or bending, thus ensuring long-term stability of shielding effectiveness under ultra-high voltage electric fields.

[0037] Meanwhile, aramid, acting as the core yarn, bears most of the mechanical load, preventing the stainless steel fibers from breaking due to direct stress. Reverse twisting balances the yarn torque, resulting in a smoother yarn, easier subsequent weaving processes, and better fabric resistance to pilling. Compared to traditional blended yarns, this wrapping structure can achieve or even exceed the same conductivity with fewer metal fibers. The metal fibers are concentrated on the surface, resulting in higher conductivity, thus retaining more of the aramid's flexibility, making the fabric softer and more breathable while maintaining rigidity.

[0038] Furthermore, the control system also includes a data acquisition module and an analysis module. The data acquisition module obtains real-time detection data during the composite yarn production process, including single-point detection data, segment-accumulated data, and risk fusion data.

[0039] The single-point detection data includes data obtained by using a line-scanning industrial camera installed after the inner layer wrapping process. The yarn travels at a constant speed, and the line-scanning camera scans line by line to obtain high-resolution images. After binarization, the inner layer stainless steel fiber coverage C is obtained. in,i Using a line-scanning industrial camera installed after the outer layer wrapping process, and employing the same acquisition principle as the inner layer, the outer layer stainless steel fiber coverage C is obtained after processing. out,i The yarn diameter d is directly acquired by a laser displacement sensor installed on the yarn's travel path. i The suspended fiber area A is calculated by acquiring three-dimensional contour data through a high-speed laser triangulation sensor installed after the outer layer is wrapped. susp,i The specific processing method is as follows: The yarn surface contour lines are continuously acquired, and the three-dimensional morphology of the yarn surface is reconstructed through continuous contour splicing. Then, the gap areas between the fibers and the underlying substrate are identified. When there is a gap exceeding one-third of the fiber diameter below the fiber, that fiber segment is determined to be suspended. The projected areas of all suspended fibers are summed. The total detection area A is measured by a high-speed laser triangulation sensor. total,i .

[0040] It also includes the preset inner layer standard coverage C in,0 Outer layer standard coverage C out,0 The standard yarn diameter d0 and the suspension penalty coefficient η are all calibrated by the process engineer and entered into the standard parameter library. The suspension penalty coefficient η ranges from 1.0 to 5.0, with a larger value for high-grade products and a smaller value for general products.

[0041] Furthermore, the cumulative relevant data within each segment includes the number of inspection points (n) per evaluation segment, set by the process engineer based on production speed and inspection accuracy requirements. The relative deviation of the inner layer coverage (ΔC) is calculated from this data. in,i Its calculation formula is ΔC in,i Equals C in,i Subtract C in,0 Difference divided by C in,0 The relative deviation ΔC of the outer layer coverage was obtained through calculation. out,i Its calculation formula is ΔC out,i Equals C out,i Subtract C out,0 Difference divided by C out,0 The average deviation CP of the inner layer within the segment was obtained through calculation. inIts calculation formula is CP in It equals one-n times ΔC in,i The summation from i = 1 to n. The average deviation CP between the inner and outer layers of the segment is obtained through calculation. out Its calculation formula is CP out It equals one-n times ΔC out,i Summation from i=1 to n. (By ΔC) in,i ΔC out,i CP in and CP out The calculated cross-correlation ρ between inner and outer layer deviations in,out,k The calculation method uses the Pearson correlation coefficient formula.

[0042] It also includes a correlation threshold ρ calibrated by engineers based on sensor noise levels and process requirements. th , and the correlation penalty intensity λ, which is determined based on the actual degree of harm of the correlation anomaly;

[0043] Correlation threshold ρ th The value ranges from 0.2 to 0.5. It is calibrated by engineers based on the sensor noise level and process requirements. The calibration method is as follows: Collect a segment of data while the production line is running normally without faults, and calculate ρ. in,out,k The statistical distribution of ρ th The value is set to the 95th percentile or higher of the normal distribution, so that under normal conditions, less than 5% of the windows will trigger the correlation penalty. In this embodiment, 0.3 can be used.

[0044] Furthermore, risk fusion-related data includes direct acquisition via infrared temperature sensors installed in the wrapping area. The acquisition method involves aligning the sensor with the yarn surface and using a non-contact, real-time measurement of the yarn surface temperature. After outputting a continuous analog signal, the arithmetic mean of n temperature data points is calculated to obtain the average yarn surface temperature T. yarn,k The standard process temperature T is specified according to the product design specifications. ref From the Handbook of Materials Science, the material constants determined by the actual use of stainless steel fiber materials, including the coefficient of thermal expansion α of stainless steel. SS Material constants from the material supplier's technical data sheet: aramid's coefficient of thermal expansion α Kevlar The total length L of stainless steel fibers within the window, calculated based on yarn structure parameters. SS The length L of the core yarn inside the window is calculated based on the yarn speed and the window length. Kevlar And the preset normalized baseline length L0 and the minimum value ε to prevent the removal of zero.

[0045] The collected detection data is input into the analysis module, which includes a single-point detection unit, an intra-segment accumulation unit, and a risk fusion unit. The analysis module sequentially outputs the single-point wrapping comprehensive deviation index U. i Intra-segment defect accumulation coefficient S k and the comprehensive defect risk warning coefficient R k .

[0046] The single-point detection unit is based on the inner layer stainless steel fiber coverage C from the single-point detection related data. in,i Inner layer standard coverage C in,0 C, outer layer stainless steel fiber coverage out,i Outer layer standard coverage C out,0 Yarn diameter d i Standard yarn diameter d0, suspended fiber area A susp,i Total area A of the detection area total,i And the suspension penalty coefficient η outputs the single-point wrapping comprehensive deviation index U i The suspension penalty coefficient η is calibrated by the process engineer according to the product grade. The calibration method is as follows: a batch of samples with a known suspension ratio is manufactured on the production line, and their actual conductivity is measured by adjusting the wrapping tension or speed. Then, the value of the suspension penalty coefficient is used to deduce the single-point wrapping comprehensive deviation index U. i It has the strongest correlation with electrical conductivity. The comprehensive deviation index U is obtained through single-point wrapping. i The severity of local defects at the i-th detection point is determined by comprehensively evaluating the deviation of inner and outer layer coverage, diameter deviation, and fiber suspension ratio.

[0047] Specifically, the processing logic of the single-point detection unit is as follows:

[0048]

[0049] Single-point wrapping comprehensive deviation index U i The processing formula consists of the product of three factors. The first factor is the Euclidean distance of the coverage deviation, which is obtained by taking the square root of the sum of the squares of the relative deviations of the inner and outer layer coverage. The relative deviation of the inner layer coverage is ΔC. in Relative deviation ΔC from outer layer coverage outThe calculation method has been described above. The squaring operation converts both positive and negative deviations into positive values, treating both excessive and insufficient coverage as defects. Furthermore, the squaring operation assigns higher weight to large deviations, meaning a serious defect is more noteworthy than two minor defects, which aligns with the practical needs of defect detection. Summation is used to combine the inner and outer layer deviations, reflecting that the deviations in inner and outer layer coverage can compensate for each other. If the inner layer has more coverage than the outer layer, the sum of squares may still be relatively small, reflecting the reality that the total amount of fibers in the inner and outer layers may simply be unevenly distributed rather than an abnormal total amount. Taking the square root restores the sum of squares to the same dimension as the original deviation, avoiding the numerical inflation caused by the squaring operation.

[0050] Another point worth emphasizing is that the reason for choosing Euclidean distance over Manhattan distance is that Euclidean distance assigns higher weight to larger deviations. In defect detection, a serious defect is more worthy of attention than multiple minor defects. Under Euclidean distance, the serious defect contributes more, while under Manhattan distance, the two are equal and cannot be distinguished. This characteristic of prioritizing large deviations meets the actual needs of defect detection.

[0051] The second factor is the diameter deviation correction factor, which is obtained by adding 1 to the relative deviation of the yarn diameter. The relative deviation of the yarn diameter is calculated as d. i Subtract the absolute value of d0 and divide by d0. Adding 1 ensures that the diameter is perfectly normal, i.e., d... i When d = 0, the factor is one and does not change U. i When the diameter deviates, the factor is greater than 1, and the magnification U is increased. i This design, which penalizes deviations only and does not penalize under normal conditions, avoids introducing unnecessary noise under normal conditions. The reason for choosing additive correction instead of traditional multiplicative correction is that diameter deviation is a concomitant signal of coverage defects, not an independent defect. When the diameter is perfectly normal, U should not be reduced simply because there is no diameter deviation. i Nor should additional penalties be introduced; the design of adding 1 ensures that the factor remains constant at 1 under normal conditions, amplifying U only under abnormal conditions. i This achieves the effect of conditional triggering.

[0052] The third factor is the fiber suspension penalty factor, which is obtained by adding 1 to the suspension penalty coefficient η and multiplying by the suspension ratio. The suspension ratio is calculated as A. susp,i Divide by A total,i This normalizes the absolute suspended area, making it unaffected by the size of the detection area. η controls the suspended proportion on U. i The amplification intensity increases with increasing η, resulting in a higher U value for the same suspended proportion. iThe higher the amplification factor, the more flexibly the system can adjust the detection sensitivity according to the quality requirements of different products. The reason for choosing linear penalty instead of nonlinear penalty is that the relationship between the proportion of dangling objects and the loss of conductivity is approximately linear, and using square penalty would excessively suppress slight dangling objects and underestimate their actual impact.

[0053] Multiplying the three factors instead of adding them means that defects in each dimension will amplify each other. When a detection point has both coverage deviation and diameter anomaly, U i The increase is greater than the sum of the values ​​of the two when they exist individually. Through the above processing, the system can effectively identify composite defects with simultaneous anomalies across multiple dimensions.

[0054] Furthermore, by identifying the void area between the fiber and the substrate, the proportion of suspended fiber area is incorporated into the evaluation system as an independent defect dimension. This addresses the limitation of existing two-dimensional image detection technologies, which can only determine whether a fiber covers the area but cannot determine whether the fiber is in close contact with the underlying layer. Although suspended fibers visually cover the area, they do not actually participate in the conductive network. This ineffective coverage is completely undetectable in conventional two-dimensional detection and can only be discovered through three-dimensional morphology reconstruction. At the same time, suspended fibers may break or fray due to friction during subsequent weaving or wearing, forming potential sources of breakage in the conductive network. The above treatment has significant quality control implications for the early identification of this type of defect.

[0055] Furthermore, the intra-segment cumulative unit is based on the single-point entangling comprehensive deviation index U in the intra-segment cumulative related data. i Number of detection points per evaluation segment (n), relative deviation of inner layer coverage (Δ) Cin,i ΔC, relative deviation of outer layer coverage out,i Average deviation CP within the inner layer of the segment in Average deviation CP between inner and outer layers of segment out Cross-correlation ρ between inner and outer layer deviations in,out,k Correlation threshold ρ th Correlation penalty intensity λ and defect accumulation coefficient S within the output segment k Through the segment defect accumulation coefficient S k Assess the overall quality level of this segment. Specifically, the processing logic for the cumulative units within the segment is as follows:

[0056]

[0057] in:

[0058]

[0059] in:

[0060]

[0061]

[0062] The above processing formula consists of the product of a basic term and a correction term. The basic term is the root mean square value, which is calculated by multiplying one-nth by U. i It is obtained by taking the square root of the sum of the squares of i from 1 to n. The squaring operation is similar to the first factor in Formula 1, and for larger values ​​of U... i Assign higher weights. Summation accumulates the defect information of all detection points within the segment. Divide by n and take the average, making S... k Unaffected by the segment length n, S under different n values k It is comparable. Taking the square root restores the result to be comparable to U. i They have the same units of measurement. The reason for choosing the root mean square (RMS) instead of the arithmetic mean is that the arithmetic mean is often dragged down by a large number of good points, making serious deficiencies almost invisible, while the RMS significantly reflects serious deficiencies. The RMS reflects both the level of the mean and the degree of fluctuation. The greater the fluctuation, the greater the deviation of the RMS from the arithmetic mean. This avoids the problem of the arithmetic mean being dragged down by good points and masking serious deficiencies, which better meets the needs of intra-segment quality assessment.

[0063] The correction term is a penalty factor for the cross-correlation between inner and outer layer defects, obtained by adding 1 to λ and multiplying by the degree to which the correlation exceeds the threshold. The calculation of the degree to which the correlation exceeds the threshold is divided into three layers:

[0064] The first layer calculates the Pearson correlation coefficient ρ for the inner-outer layer bias. in,out,k The numerator calculates the covariance of the inner and outer layer deviations. When both inner and outer layers are simultaneously large or small, the product is positive, and the covariance is positive. When one layer is large and the other is small, the product is negative, and the covariance is negative. When the changes in the two layers are independent, the positive and negative products cancel each other out, and the covariance approaches zero. The denominator is the product of the standard deviations of the inner and outer layer deviations, used for normalization. Dividing by the standard deviation yields ρ. in,out,k It is not affected by the magnitude of the deviation.

[0065] In the double reverse wrapping process, the correlation between the inner and outer layer coverage deviations is essentially linear. Changes in core yarn speed will cause synchronous linear changes in the inner and outer layer coverage, while uneven yarn supply will cause inverse linear changes in the inner and outer layer coverage. The Pearson correlation coefficient can effectively capture this linear relationship, and it is simple to calculate and has a clear physical meaning, making it suitable for real-time computation by edge computing units.

[0066] The second layer calculates the degree to which the correlation exceeds a threshold, i.e., ρ. in,out,k The absolute value minus ρ th Difference divided by 1 minus ρ th The difference, the excess part is normalized to the interval between zero and one, when the absolute value equals ρ. th The value is zero when the absolute value is zero, and 1 when the absolute value is 1. This normalization makes ρ thMeasures of excess are comparable when taking different values.

[0067] The third layer uses the max function to truncate negative values ​​to zero when the absolute value is less than or equal to ρ. th At that time, it was assumed that there was no meaningful correlation between the inner and outer layer deviations, and the correction term degenerates to zero, S k Unaffected. When the absolute value is greater than ρ th When a meaningful association is considered to exist, the correction term is greater than zero, S k The magnitude is amplified. The reason for taking the absolute value is that both strong positive and strong negative correlations indicate the existence of systemic root causes, albeit of different types. Positive correlations suggest core yarn speed fluctuations, while negative correlations suggest uneven yarn supply distribution; both need to be identified and penalized, hence the absolute value is used for unified processing. Furthermore, the use of the max function for truncation is precisely to account for values ​​below the threshold ρ. th The correlation is considered to be a random correlation caused by random noise and has no physical meaning. Even if the absolute value is slightly lower than ρ, it will not be meaningful if it is not truncated. th It also produces a slight correction effect and introduces unnecessary noise. The max function ensures that the penalty is triggered only when the absolute value clearly exceeds the threshold, thus achieving a hard boundary for threshold triggering.

[0068] The correlation penalty strength λ typically ranges from 0.5 to 2.0, and is calibrated by engineers based on the actual severity of the correlation anomaly. The calibration method involves analyzing the correlation strength between the window of correlation anomaly in historical data and the actual product quality, such as conductivity test results, and adjusting λ to ensure that the defect accumulation coefficient S within the segment is within acceptable limits. k It has the strongest correlation with product quality.

[0069] In the above processing, a cross-correlation correction for inner and outer layer defects is introduced, achieving a dimensional upgrade from single-parameter detection to detection of relationships between parameters. ρ in,out,k The system is calculated using two conventional parameters: inner and outer layer coverage. However, the synergistic information between these two parameters in the specific context of double reverse wrapping reveals diagnostic value that conventional single-parameter detection cannot uncover. For example, if both inner and outer layers are simultaneously too high or too low, it may indicate a fluctuation in the core yarn speed. When the core yarn speeds up, the number of wrapping turns for both layers increases, and the coverage increases synchronously. When the inner layer is too high and the outer layer is too low, or vice versa, it may indicate an uneven yarn supply. Therefore, the system can not only identify the existence of defects but also preliminarily determine the possible causes of defects through correlation patterns, providing operators with clear directional information for adjusting process parameters and significantly shortening troubleshooting time.

[0070] Furthermore, the risk fusion unit is based on the most severe single-point defect U within the window from the risk fusion-related data. k max Intra-segment defect accumulation coefficient S kAverage surface temperature of yarn (T) yarn,k Standard process temperature T ref The coefficient of thermal expansion of stainless steel is α SS Aramid thermal expansion coefficient α Kevlar The total length L of stainless steel fiber inside the window SS Length L of the inner core yarn of the window Kevlar The normalized baseline length L0 and the comprehensive defect risk warning coefficient R output by eliminating the zero minimum value ε are used to determine the overall defect risk warning coefficient. k By using the comprehensive defect risk early warning coefficient R k The most severe single-point defect, cumulative defects within a segment, and thermal expansion mismatch effect are combined into a comprehensive risk level, where the most severe single-point defect within the window is U. k max This is the single-point wrapping comprehensive deviation index U. i The maximum value within window k.

[0071] Specifically, the processing logic of the risk fusion unit is as follows:

[0072]

[0073] in: , ;

[0074] , , ;

[0075]

[0076] The above processing logic consists of multiplying the basic risk factor and the conditional amplification factor. The basic risk factor φ(x) will multiply U k max After nonlinear transformation, it is mapped to the basic risk value. First, U k max Through x equals U k max Divide by 1 plus U k max The mapping from zero to positive infinity is compressed into an interval; this mapping is a nonlinear compression, U k max When U is small, the mapping is approximately linear. k max When it is large, each increase of one unit of U k maxThe increase in x becomes smaller and smaller. Then, x is amplified from zero to 1 back to positive infinity through a mapping where φ(x) equals x divided by 1 minus x plus ε. When x is small, φ(x) is approximately equal to x; when x approaches 1, the denominator approaches zero, and φ(x) increases sharply. ε is used to prevent the denominator from becoming zero. This composite transformation of compression followed by amplification achieves adjustable nonlinear amplification.

[0077] In yarn production, a minor single-point defect has almost no impact on the overall performance of the yarn, and the risk is approximately proportional to the defect. However, a serious single-point defect may mean that the fiber at that point is close to breaking or has been severely detached. Its harm is far greater than the sum of multiple minor defects because the breaking strength of the yarn is determined by the weakest point, not by the average value. The nonlinear amplification of φ(x) simulates this short-plank effect.

[0078] The conditional amplification factor is composed of 1 plus D(y,z), where D(y,z) is the cooperative triggering function. First, let S... k By y equals S k Divide by 1 plus S k The mapping from zero to positive infinity is compressed to the interval between zero and one, similar to the mapping of x. Then the thermal expansion mismatch ΔL is... thermal,k Dividing the absolute value of y by the normalized reference length L0 yields the dimensionless relative value z. Finally, the product of y and z is calculated, and then mapped to the interval from zero to one by y×z divided by 1 plus y×z to obtain D(y,z).

[0079] The core design logic for this process is to use multiplicative coupling instead of additive coupling. When the quality within a segment is good, i.e., S... k When y is approximately zero, even if the temperature fluctuates greatly (i.e., z is large), y multiplied by z is approximately zero, D is approximately zero, the conditional amplification factor is 1, and R... k Unaffected, this reflects the physical fact that high-quality yarn can withstand temperature fluctuations; that is, tightly wrapped fibers will not slip significantly even with temperature changes. When the temperature is constant, ΔL thermal,k When z is approximately zero, even if the defect within the segment is severe (i.e., y is large), y multiplied by z is approximately zero, D is approximately zero, the conditional amplification factor is 1, and R... k Unaffected, this reflects the logic of treating defects as basic risks during periods of stable temperature; without the drive of thermal expansion, defects will not worsen further.

[0080] Only when both intra-segment defects and temperature deviations exist simultaneously, and y multiplied by z is greater than zero, D is greater than zero, the conditional amplification factor is greater than 1, and R... kThe amplification factor simulates the physical law that when yarn of already poor quality encounters thermal expansion mismatch caused by temperature fluctuations, defects will be significantly amplified. In other words, the superposition of two minor problems can lead to serious malfunctions. If additive coupling is used, even if there are no defects within a segment, as long as temperature fluctuations exist, the conditional amplification factor will be greater than 1, causing the system to alarm due to temperature fluctuations even when the yarn quality is good, which is prone to false alarms.

[0081] The thermal expansion mismatch ΔL in the above processing logic thermal,k The calculation method is based on the coefficient of thermal expansion α of stainless steel. SS Multiply by the total length L of the stainless steel fibers inside the window SS Multiply by the average yarn surface temperature T yarn,k Compared with standard process temperature T ref The difference, minus the thermal expansion coefficient α of aramid. Kevlar Multiply by the length L of the inner core yarn of the window Kevlar Multiply by T yarn,k With T ref The difference. Due to α SS A positive value indicates that stainless steel elongates when heated, α Kevlar A negative value indicates that aramid shrinks when heated. As temperature rises, the first term represents positive elongation, and the second term represents negative shrinkage. Subtracting a negative number is equivalent to adding its absolute value; the two effects are superimposed, resulting in ΔL. thermal,k The absolute value increases. This reverse thermal expansion effect is a unique physical phenomenon of double-reverse-wound composite yarns. The reason for choosing the yarn surface temperature instead of the ambient temperature is that the yarn surface temperature directly reflects the temperature state of the material itself. During the wrapping process, the frictional heat generated between the fiber and the core yarn, the deformation heat caused by the wrapping tension, etc., will make the yarn surface temperature higher than the ambient temperature. Using the yarn surface temperature can more accurately reflect the actual driving force of the thermal expansion effect.

[0082] In the above process, by introducing the relevant effects of thermal expansion mismatch, the system can distinguish between tension fluctuations caused by temperature and tension abnormalities caused by equipment failure, thus avoiding misjudging temperature effects as equipment failures and causing unnecessary downtime for inspection and equipment maintenance. At the same time, the system can identify the synergistic deterioration effect of temperature fluctuations and defects within the section, that is, the defects of yarns that are already of poor quality will be significantly aggravated under temperature fluctuations.

[0083] Example 2, based on the above examples:

[0084] See Figure 1 and Figure 2 The control system also includes an early warning module, which outputs a comprehensive defect risk early warning coefficient R from the analysis module. k Then, the early warning module outputs the alarm level according to the following hierarchical alarm rules:

[0085] S1: When Rk When the value is less than 1.0, the alarm level is green, indicating normal operation. The system response is to continue production and record data. k A value less than 1.0 means that the most serious single-point defect is minor and the quality and temperature conditions within the segment are normal, which is a typical state for normal production operation.

[0086] S2: When R k When the value is greater than or equal to 1.0 and less than 2.0, the alarm level is yellow (i.e., require attention), and the system response is to record data and prompt the operator to check. k U equals 2.0 k max A value of 2.0 indicates a severe defect without synergistic amplification, or U. k max If the value is equal to 1.0 and there is moderate synergistic amplification, the yellow level indicates that there is a defect worth noting but has not yet reached the level requiring immediate intervention. The operator should check the data records to determine whether process parameters need to be adjusted.

[0087] S3: When R k When the value is greater than or equal to 2.0 and less than 3.0, the alarm level is orange, i.e., a warning, and the system response is to automatically fine-tune the tension or speed. k 3.0 corresponds to U k max A value equal to 3.0 indicates a severe defect with no synergistic amplification, or U. k max When the value is equal to 1.5 and there is strong synergistic amplification, the orange level means that the defect is already quite serious and the system needs to intervene automatically, such as fine-tuning the wrapping tension or rotation speed, or suggesting a speed reduction. The design goal of this threshold is to intervene before the defect develops to the point of being irreversible.

[0088] S4: When R k When the value is greater than or equal to 3.0, the alarm level is red, indicating an alarm. The system response is to immediately shut down and mark the location of the defective segment. k A value greater than or equal to 3.0 means that the most severe single-point defect is extremely severe, i.e., U. k max A score of 3.0 or higher, or a moderate defect with a significant synergistic amplification effect, indicates that the yarn segment is likely to have an irreversible quality defect. Continuing production will only produce more scrap. The system should be stopped immediately, and the defective segment should be marked for subsequent removal.

[0089] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent control system for composite yarn production parameters, comprising a composite yarn body, characterized in that: It also includes a data acquisition module and an analysis module. The composite yarn body includes a core yarn (1), an inner covering layer (2) and an outer covering layer (3), and the twisting directions of the inner covering layer (2) and the outer covering layer (3) are opposite. The control system also includes a data acquisition module and an analysis module. The data acquisition module acquires real-time detection data during the composite yarn production process, including single-point detection data, intra-segment cumulative data, and risk fusion data. The acquired detection data is input into the analysis module, which includes a single-point detection unit, an intra-segment cumulative unit, and a risk fusion unit. The analysis module outputs a comprehensive defect risk warning coefficient. It also includes an early warning module. After the analysis module outputs the comprehensive defect risk early warning coefficient, the early warning module outputs the alarm level according to the graded alarm rules.

2. The intelligent control system for composite yarn production parameters according to claim 1, characterized in that: The single-point detection data includes the inner layer stainless steel fiber coverage, outer layer stainless steel fiber coverage, yarn diameter, suspended fiber area, total detection area, inner layer standard coverage, outer layer standard coverage, standard yarn diameter, and suspension penalty coefficient.

3. The intelligent control system for composite yarn production parameters according to claim 2, characterized in that: The cumulative correlation data within each segment includes the number of detection points per evaluation segment, relative deviation of inner layer coverage, relative deviation of outer layer coverage, average deviation of inner layer within the segment, average deviation of inner and outer layers within the segment, cross-correlation of inner and outer layer deviations, correlation threshold, and correlation penalty intensity.

4. The intelligent control system for composite yarn production parameters according to claim 3, characterized in that: The risk fusion-related data includes the average yarn surface temperature, standard process temperature, coefficient of thermal expansion of stainless steel, coefficient of thermal expansion of aramid, total length of stainless steel fibers within the window, length of core yarn within the window, normalized baseline length, and minimum value for preventing and eliminating zero.

5. The intelligent control system for composite yarn production parameters according to claim 4, characterized in that: The single-point detection unit outputs a single-point wrapping comprehensive deviation index based on the single-point detection related data, including the inner layer stainless steel fiber coverage rate, inner layer standard coverage rate, outer layer stainless steel fiber coverage rate, outer layer standard coverage rate, yarn diameter, standard yarn diameter, suspended fiber area, total detection area, and suspension penalty coefficient. The single-point wrapping comprehensive deviation index is used to comprehensively evaluate the inner and outer layer coverage rate deviation, diameter deviation, and fiber suspension ratio at the i-th detection point to determine the severity of the local defect at that point.

6. The intelligent control system for composite yarn production parameters according to claim 5, characterized in that: The segment accumulation unit outputs the segment defect accumulation coefficient based on the single-point wrapping comprehensive deviation index, the number of detection points per evaluation segment, the relative deviation of inner layer coverage, the relative deviation of outer layer coverage, the average deviation of inner layer within the segment, the average deviation of inner and outer layers within the segment, the cross-correlation of inner and outer layer deviations, the correlation threshold, and the correlation penalty intensity in the segment accumulation correlation data. The overall quality level of the segment is evaluated by the segment defect accumulation coefficient.

7. The intelligent control system for composite yarn production parameters according to claim 6, characterized in that: The risk fusion unit outputs a comprehensive defect risk warning coefficient based on the most severe single-point defect within the window, the cumulative defect coefficient within the segment, the average yarn surface temperature, the standard process temperature, the thermal expansion coefficient of stainless steel, the thermal expansion coefficient of aramid, the total length of stainless steel fibers within the window, the core yarn length within the window, the normalized reference length, and the prevention and removal of zero minimum values ​​in the risk fusion related data. The comprehensive defect risk warning coefficient is used to fuse the most severe single-point defect, the cumulative defect within the segment, and the thermal expansion mismatch effect into a comprehensive risk level. The most severe single-point defect within the window represents the maximum value of the single-point wrapping comprehensive deviation index within window k.

8. The intelligent control system for composite yarn production parameters according to claim 7, characterized in that: The tiered alarm rules are as follows: S1: When the comprehensive defect risk warning coefficient is less than 1.0, the alarm level is green, which means normal, and the system response is to continue production; S2: When the comprehensive defect risk warning coefficient is greater than or equal to 1.0 and less than 2.0, the alarm level is yellow (i.e., pay attention), and the system response is to record data; S3: When the comprehensive defect risk warning coefficient is greater than or equal to 2.0 and less than 3.0, the alarm level is orange, i.e., a warning, and the system response is to automatically fine-tune the tension or speed. S4: When the comprehensive defect risk warning coefficient is greater than or equal to 3.0, the alarm level is red, which means an alarm is triggered. The system response is to immediately stop the machine and mark the location of the defect segment.

Images