Spinning quality control method based on linear array image texture feature analysis
By using linear array image texture feature analysis and delayed timing masking logic, the problem of lag between motor speed adjustment and yarn tension stability in spinning machines is solved, thereby achieving stability and adaptability in yarn quality detection and ensuring production safety.
Patent Information
- Application Number
- CN202610424153.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-01
- Publication Date
- 2026-07-10
AI Technical Summary
There is a time lag between the adjustment of the motor speed of the spinning machine and the stabilization of the yarn tension, which leads to mechanical oscillation and misjudgment. Furthermore, the fixed parameter detection model is difficult to adapt to the differences in raw material batches, affecting the detection accuracy and production continuity.
A linear array image texture feature analysis method is adopted. The yarn image is captured line by line by line through a rotary encoder. Brightness compensation and Gaussian smoothing are performed, and the backbone width and contrast values are statistically analyzed. Combined with delay timing and weight adaptive correction, yarn quality control is achieved.
It effectively suppresses control feedback oscillations, enables automatic batch adaptation of raw materials, improves detection accuracy and production safety, and prevents mechanical damage.
Smart Images
Figure CN122367897A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of online detection and intelligent control technology for textile quality, and more specifically, to a spinning quality control method based on linear array image texture feature analysis. Background Technology
[0002] In existing textile production processes, line scan cameras are often used to continuously acquire images of high-speed moving yarns, and image processing algorithms are used to identify defects such as thick spots, thin spots, and fuzz. Traditional monitoring systems typically use fixed image discrimination thresholds. When a quality fluctuation is detected, the control unit executes an alarm or shutdown action according to preset logic, or performs simple open-loop speed compensation.
[0003] Currently, a substantial technical problem exists in the field of automated control for spinning: as a highly inertial mechanical system, the spinning machine experiences a significant time lag between the adjustment of its motor speed and the stabilization of the actual yarn tension. Existing monitoring methods, after executing speed adjustment actions, often result in the vision system acquiring image signals in a non-steady state due to mechanical oscillations and transient response fluctuations, leading to continuous misjudgments and system control oscillations. Simultaneously, due to the inherent differences in the physical properties between batches of spinning raw materials (such as cotton fiber), fixed-parameter detection models struggle to effectively separate raw material background noise from actual quality defects without interrupting production calibration, making it difficult to simultaneously achieve both high detection accuracy and operational continuity in complex production environments. Summary of the Invention
[0004] In view of the aforementioned existing problems, the present invention is proposed.
[0005] Therefore, this invention provides a spinning quality control method based on linear array image texture feature analysis, which solves the problems of misjudgment of speed regulation oscillation caused by mechanical inertia in existing monitoring systems, and the difficulty in automatically adapting to the background differences of raw material batches.
[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:
[0007] This invention provides a spinning quality control method based on linear array image texture feature analysis, which includes the following steps:
[0008] S1. The electrical pulse signal emitted by the rotary encoder triggers the line scan camera to perform line-by-line shooting to obtain continuous yarn unfolding images;
[0009] S2. Extract the average brightness value of the background area of the yarn unfolded image and perform brightness compensation on the main area of the yarn to obtain a preprocessed yarn image;
[0010] S3. Calculate the width of the main image representing the yarn thickness and the contrast value of the surface pixel brightness in the preprocessed yarn image, and combine them to form feature data.
[0011] S4. Perform weighted operations on the feature data based on the preset feature mapping relationship to obtain the index value representing the true quality state of the yarn;
[0012] S5. If the value of the index exceeds the preset qualified range, a speed adjustment signal is sent to the drive motor of the spinning machine and a preset delay time is started, during which the speed adjustment judgment is blocked.
[0013] S6. After the delay timer ends, the monitoring loop from image acquisition to quality assessment resumes.
[0014] As a preferred embodiment of the spinning quality control method based on linear array image texture feature analysis described in this invention, step S1 specifically includes: the rotary encoder emitting an electrical pulse signal for each fixed angle rotation of the rotating shaft; the linear array camera being in a waiting state, and capturing a line of images of the moving yarn whenever an electrical pulse signal is received, ensuring that the actual physical length of the yarn corresponding to each line of images captured by the linear array camera remains consistent, and stitching the continuously captured images together to form a continuous yarn unfolding image.
[0015] As a preferred embodiment of the spinning quality control method based on linear array image texture feature analysis described in this invention, step S2 specifically includes: selecting blank background areas where no yarn passes through the edges in the yarn unfolded image; obtaining an average brightness value based on the brightness values of all pixels in the blank background area; comparing the average brightness value with a preset standard brightness value; if the average brightness value is lower than the standard brightness value, calculating the difference between the standard brightness value and the average brightness value, and superimposing the difference onto the pixels in the main yarn area for brightness compensation; subsequently, using Gaussian smoothing filtering to blur the entire image to eliminate instantaneous noise, resulting in a preprocessed yarn image.
[0016] As a preferred embodiment of the spinning quality control method based on linear array image texture feature analysis described in this invention, step S3 specifically includes: scanning line by line on the preprocessed yarn image to find the darkest and densest part of the image as the yarn backbone; counting the number of pixels the backbone spans horizontally to obtain the backbone image width representing the actual physical thickness of the yarn; scanning the pixels with alternating light and dark at the edges of the backbone and calculating the degree of light and dark variation of the pixels to obtain a contrast value representing the degree of surface hairiness of the yarn; and combining the backbone image width and the contrast value to form feature data.
[0017] As a preferred embodiment of the spinning quality control method based on linear array image texture feature analysis according to the present invention, step S4 specifically includes: multiplying the backbone image width and the contrast value by the corresponding preset weight ratio to obtain the weighted backbone width and weighted contrast; within a unit yarn length, counting the number of times the weighted backbone width exceeds a preset upper limit threshold as the thick spot occurrence rate, and counting the number of times the weighted backbone width is lower than a preset lower limit threshold as the fine spot occurrence rate; obtaining the average value of all the weighted contrast values within a unit yarn length to obtain the hairiness index; and combining the thick spot occurrence rate, the fine spot occurrence rate, and the hairiness index into an index value.
[0018] As a preferred embodiment of the spinning quality control method based on linear array image texture feature analysis according to the present invention, step S4 further includes a weight adaptive correction step: continuously acquiring the average backbone image width and average contrast value of the spinning machine within a preset stable period; calculating the width deviation between the average backbone image width and the preset standard yarn width, and calculating the contrast deviation between the average contrast value and the preset standard hairiness index; generating a compensation factor for the preset weight ratio based on the width deviation and the contrast deviation, and dynamically updating the preset weight ratio using the compensation factor to achieve detection adaptation for different batches of raw material yarn.
[0019] As a preferred embodiment of the spinning quality control method based on linear array image texture feature analysis described in this invention, step S5 specifically includes: comparing the index value with a preset qualified numerical range; if the index value exceeds the qualified numerical range, calculating a new operating speed and converting the new operating speed into a speed regulation voltage signal and sending it to the drive motor; simultaneously sending the speed regulation voltage signal, starting a delay timing program; before the running time of the delay timing program reaches a preset delay duration, disabling the speed regulation control logic of the spinning machine, so that the mechanical components smoothly transition to the new operating speed.
[0020] As a preferred embodiment of the spinning quality control method based on linear array image texture feature analysis according to the present invention, step S6 includes: after the delay timing program runs for the preset delay time, removing the shielding of the speed control logic and restoring the pulse trigger response of the linear array camera, acquiring the new yarn image after speed adjustment and stabilization; converting the new yarn image into the latest index value, continuously monitoring the change trend of the index value within a preset time window, and executing subsequent operation instructions according to the change trend.
[0021] As a preferred embodiment of the spinning quality control method based on linear array image texture feature analysis described in this invention, the subsequent operation instructions include: if the index value shows a downward trend within the preset time window and is within the preset qualified numerical range, then the stable operation instruction is executed, the voltage signal for changing the speed is stopped, and the spinning machine is controlled to maintain the currently adjusted operating speed for stable operation; if the index value shows a continuous abnormal trend within the preset time window and exceeds the preset qualified numerical range, then the audible and visual alarm instruction is executed, the current operating speed is maintained, and the alarm circuit is activated to drive the audible and visual alarm to issue an alarm signal.
[0022] As a preferred embodiment of the spinning quality control method based on linear array image texture feature analysis described in this invention, the subsequent operation instructions further include: executing a synchronous stop instruction, sending an instruction to the control mechanism of the spinning machine to control each execution component to synchronously reduce its operating speed until it stops completely.
[0023] The beneficial effects of this invention are:
[0024] 1. Effectively suppress control feedback oscillations: By introducing delayed timing and shielding logic after speed regulation, the physical instability period caused by mechanical transmission inertia is avoided, preventing continuous misjudgments caused by visual detection response fluctuations and improving the stability of the closed-loop control system.
[0025] 2. Achieve automatic batch adaptation of raw materials: Introduce a sensitivity dynamic balancing strategy and use a weight adaptive correction mechanism to lock the detection benchmark at the average level of the current batch in real time. This effectively removes the background noise caused by the natural properties of the raw materials and solves the problem of frequent manual adjustment after changing raw materials.
[0026] 3. Balancing detection sensitivity and production safety: Employing a statistical judgment model based on sampling probability, it ensures accurate capture of continuous anomalies while filtering out occasional noise.
[0027] 4. A comprehensive equipment synchronization protection mechanism: In conjunction with the synchronous deceleration and shutdown logic, it ensures the motion synchronization of multi-axis actuators in fault conditions, prevents yarn breakage and secondary mechanical damage, and ensures the safety of the production process. Attached Figure Description
[0028] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0029] Figure 1 This is a flowchart of the spinning quality control method based on linear array image texture feature analysis.
[0030] Figure 2 The flowchart of the speed regulation delay shielding and oscillation suppression logic provided in the embodiments of the present invention is shown.
[0031] Figure 3 The flowchart for weight adaptive correction and sensitivity balancing provided in the embodiments of the present invention. Detailed Implementation
[0032] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0033] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.
[0034] Secondly, the term "one embodiment" or "example" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the invention. The appearance of an embodiment in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that mutually excludes other embodiments.
[0035] Example 1
[0036] Reference Figure 1 , Figure 2 This is the first embodiment of the present invention, which provides a spinning quality control method based on linear array image texture feature analysis. This embodiment takes a ring spinning machine with a digital speed control interface and high-precision guide rollers as an example. The maximum designed spindle speed of this spinning machine is 25000 r / min. Its drive motor is vector-controlled by a frequency converter and supports dynamic speed adjustment via 0-10V standard analog voltage signal or RS485 communication protocol. The method includes the following steps:
[0037] S1. The linear array camera is triggered by the electrical pulse signal emitted by the rotary encoder to take line-by-line pictures and obtain continuous yarn unfolding images.
[0038] Every time the rotating shaft rotates a fixed angle, the rotary encoder sends out an electrical pulse signal. The line scan camera is in a standby state. Whenever it receives an electrical pulse signal, it takes a line of images of the moving yarn, so that the actual physical length of the yarn corresponding to each line of images taken by the line scan camera remains consistent. The continuously captured images are stitched together to form a continuous image of the unfolded yarn.
[0039] S2. Extract the average brightness value of the background area of the yarn unfolded image and perform brightness compensation on the main area of the yarn to obtain a preprocessed yarn image.
[0040] In the yarn unfolding image, blank background areas where no yarn passes through the edges are selected. The average brightness value is obtained based on the brightness values of all pixels in the blank background area. The average brightness value is compared with a preset standard brightness value. If the average brightness value is lower than the standard brightness value, the difference between the standard brightness value and the average brightness value is calculated and superimposed on the pixels of the main yarn area for brightness compensation. Then, Gaussian smoothing filtering is used to blur the entire image to eliminate instantaneous noise, resulting in a preprocessed yarn image.
[0041] During the operation of the spinning machine, a high-precision rotary encoder is installed on a rotating shaft that rotates synchronously with the guide roller. The rotary encoder is an incremental encoder with a resolution of 2048 P / R. When the roller rotates, the rotary encoder outputs a 5V electrical pulse signal every 0.5° of rotation. A line scan camera receives the electrical pulse signal through an external trigger interface. The line scan camera has a pixel size of 7μm × 7μm and supports a line frequency of no less than 50kHz. Each received signal constitutes a line scan acquisition of the high-speed moving yarn below.
[0042] Because the scanning frequency of the line scan camera is hard-synchronized with the yarn linear velocity via electrical pulses, each line of the image captured by the camera corresponds to a physical length of 0.1 mm on the yarn surface, regardless of changes in vehicle speed. The system stitches together 2048 consecutively acquired lines of images to form a continuous yarn unfolding image with a total pixel count of 2048×2048.
[0043] The system then extracts a 20×20 pixel blank background area without yarn from the image edge and calculates its average gray value. In this step, the system introduces a brightness compensation formula to eliminate interference from ambient light fluctuations, as follows:
[0044]
[0045] in,
[0046] : Compensated pixel grayscale value;
[0047] : The grayscale value of the pixels in the main yarn area of the original image;
[0048] The system's preset standard reference brightness value (set to 200 in this embodiment);
[0049] : The average gray value of the measured background area (180 in this embodiment).
[0050] The system first calculates the difference between the standard brightness and the measured background brightness. The difference is then used as a gain compensation and superimposed on each pixel in the main area of the yarn, so that the brightness level of the entire image is uniformly aligned with the standard reference value.
[0051] The entire image is then blurred using a 3×3 window Gaussian smoothing filter to eliminate instantaneous noise and obtain a pre-processed yarn image.
[0052] S3. Statistically preprocess the yarn image by combining the width of the main image representing the yarn thickness and the contrast value of the surface pixel brightness to form feature data.
[0053] A line-by-line scan is performed on the preprocessed yarn image to find the darkest and densest part of the image as the yarn backbone; the number of pixels that the backbone spans horizontally is counted to obtain the backbone image width, which represents the actual physical thickness of the yarn; the pixels with alternating light and dark areas at the edges of the backbone are scanned and the degree of light and dark variation of the pixels is calculated to obtain the contrast value, which represents the degree of fuzziness on the yarn surface; the backbone image width and the contrast value are combined to form feature data.
[0054] S4. Perform weighted calculations on the feature data based on the preset feature mapping relationship to obtain the index value representing the true quality state of the yarn;
[0055] Multiply the backbone image width and contrast values by the corresponding preset weight ratios to obtain the weighted backbone width and weighted contrast. Within a unit yarn length, count the number of times the weighted backbone width exceeds a preset upper threshold as the thick spot occurrence rate, and count the number of times the weighted backbone width is below a preset lower threshold as the detail occurrence rate. Obtain the average value of all weighted contrast values within a unit yarn length to obtain the hairiness index. Combine the thick spot occurrence rate, detail occurrence rate, and hairiness index into an index value.
[0056] The system performs a line-by-line grayscale scan of the preprocessed image, identifying the continuous pixel region with the lowest grayscale value (darkest color) as the yarn backbone. If the backbone spans 12 pixels horizontally, the width of the backbone image is recorded as [value missing]. Simultaneously, the contrast ratio between the main edge pixels and the background is calculated. .
[0057] Based on a preset feature mapping relationship, the system combines the following weighted calculation formulas to form feature data. :
[0058]
[0059] in,
[0060] : The weighted feature data of the current slice;
[0061] : Width of the backbone image extracted from the current row of scans (in pixels);
[0062] : Edge contrast value extracted from the current row scan;
[0063] : Preset weight ratio of the backbone width (in this embodiment, it is initially set to 0.6);
[0064] : The preset weighting ratio of the contrast value (in this embodiment, it is initially set to 0.4).
[0065] The system multiplies the geometric features (width) and texture features (contrast) by their respective weighting coefficients and sums them to obtain a single value that comprehensively represents the yarn quality at that location.
[0066] Within a unit yarn length of 10 meters, if the weighted feature data F repeatedly exceeds the preset upper limit threshold (e.g., 18), the system automatically accumulates and outputs the thick spot occurrence rate. Finally, the thick spot occurrence rate, thin spot occurrence rate, and hairiness index are integrated into the current yarn's index value.
[0067] S5. If the index value exceeds the preset qualified range, a speed adjustment signal is sent to the drive motor of the spinning machine and a preset delay timer is started. The speed adjustment judgment is blocked within the delay timer.
[0068] The index value is compared with the preset qualified number range. If the index value exceeds the qualified number range, a new operating speed is calculated and converted into a speed regulation voltage signal and sent to the drive motor. At the same time as sending the speed regulation voltage signal, a delay timing program is started. Before the running time of the delay timing program reaches the preset delay time, the speed regulation control logic of the spinning machine is disabled, so that the mechanical parts smoothly transition to the new operating speed.
[0069] S6. After the delay timer ends, resume the monitoring loop from image acquisition to quality assessment.
[0070] After the delay timing program reaches the preset delay time, the shielding of the speed control logic is released and the pulse trigger response of the line scan camera is restored to acquire the new yarn image after speed adjustment and stabilization; the new yarn image is converted into the latest index value, the trend of index value change within the preset time window is continuously monitored, and subsequent operation instructions are executed according to the trend.
[0071] In this embodiment, the acceptable range for the indicator value is set to [0, 15]. When the vision system detects that the current indicator value rises to 18 due to fluctuations in the raw material tension, the system determines that the quality exceeds the standard.
[0072] The control unit immediately calculates the required speed reduction increment and adjusts the original 10V drive voltage to 8.5V before sending it to the drive motor, or adjusts the motor operating frequency from 50Hz to 42.5Hz via the RS485 bus. At the instant the voltage signal is sent, the system synchronously triggers the internal delay timing program, with a preset delay duration of 3000ms (this duration is set according to the rotational inertia of the mechanical transmission system).
[0073] During the 3000ms countdown, even though the mechanical components (rollers, motors) are decelerating due to inertia and the yarn surface is in a period of unstable physical oscillation, the vision system strictly blocks all new speed adjustment commands for the spinning machine to ensure that the actuator does not generate high-frequency oscillation judgments due to temporary fluctuations during the transition period.
[0074] Subsequent operation instructions include: if the indicator value shows a downward trend within the preset time window and is within the preset qualified number range, then execute the stable operation instruction, stop sending the voltage signal to change the speed, and control the spinning machine to maintain the current adjusted operating speed for stable operation; if the indicator value shows a continuous abnormal trend within the preset time window and exceeds the preset qualified number range, then execute the audible and visual alarm instruction, maintain the current operating speed, and connect the alarm circuit to drive the audible and visual alarm to issue an alarm signal.
[0075] Simultaneously with activating the audible and visual alarm circuit, subsequent operation instructions include: executing a synchronous stop instruction, sending instructions to the control mechanism of the spinning machine to control each actuator to synchronously reduce its operating speed until it stops completely.
[0076] Once the delay timing program has run for 3000ms, the mechanical components have smoothly transitioned to the new rotational speed. The system then de-shields the logic, restores the pulse trigger response of the line scan camera, and begins acquiring yarn images at the new rotational speed.
[0077] After a preset 5-second monitoring window:
[0078] If the calculated new index value falls back to 12 (within the qualified range of [0,15]), the system determines that the index is trending downward and executes a stable operation command to maintain continuous operation at 8.5V voltage.
[0079] If the indicator value is still outside the qualified range, it is determined that the indicator continues to be abnormal. The system will activate the alarm circuit to drive the audible and visual alarm to issue an alarm signal, prompting manual intervention. In specific serious fault scenarios, the system will execute a shutdown command simultaneously, controlling each execution component to decelerate synchronously until it stops completely, in order to protect the mechanical equipment and yarn quality.
[0080] Example 2
[0081] Reference Figure 3 This is the second embodiment of the present invention, which provides a spinning quality control method based on linear array image texture feature analysis. This embodiment focuses on how the present invention achieves detection adaptation through a weighted adaptive correction step when dealing with different batches of raw material yarn. Specifically, it includes:
[0082] Step S4 also includes a weight adaptive correction step: continuously acquiring the average backbone image width and average contrast value of the spinning machine within a preset stable period; calculating the width deviation between the average backbone image width and the preset standard yarn width, and calculating the contrast deviation between the average contrast value and the preset standard hairiness index; generating a compensation factor for the preset weight ratio based on the width deviation and contrast deviation, and dynamically updating the preset weight ratio using the compensation factor to achieve detection adaptation for different batches of raw yarn.
[0083] In this embodiment, the spinning machine changed the batch of raw materials (e.g., the maturity of the cotton fibers changed). The system initially runs at a preset weight ratio, i.e., the main width weight. The contrast weight is 0.6. It is 0.4.
[0084] The system continuously acquires the operating data of the spinning machine within a preset stable period (300s in this embodiment). Statistical analysis shows that the average main yarn image width within this period... For 13.5 pixels, the average contrast ratio is It is 8.2.
[0085] 1. Calculation process of deviation and compensation factor:
[0086] The system compares the above measured average values with preset standard physical parameters in real time and calculates the deviation using the following formula:
[0087]
[0088] in,
[0089] These are width deviation and contrast deviation, respectively.
[0090] : Preset standard yarn width (set to 12 in this embodiment);
[0091] : Preset standard feather index (set to 10 in this embodiment).
[0092] The system first identifies the background offset of the raw material's physical properties. Due to a change in the cotton fiber batch, the measured average width... =13.5, then the system calculates the width deviation. Actual average contrast ratio =8.2, then the contrast deviation is .
[0093] Subsequently, the system introduces an adjustment sensitivity coefficient. and The physical deviations of the above different dimensions are transformed into a comprehensive compensation factor under a unified dimension. :
[0094]
[0095] In this embodiment (This represents the contribution of the width deviation per pixel to the adjustment of the weight). (This indicates the contribution of contrast offset to the fine-tuning of weights). This formula reflects the system's collaborative evaluation logic for multi-dimensional deviations. Because the main body of the new batch of yarn is naturally thicker ( (For positive bias), the system tends to generate positive compensation to reduce the sensitivity of that dimension; while the contrast is low ( This variable serves as a reverse adjustment variable to avoid over-adjustment of weights due to fluctuations in a single feature. The calculation yields... .
[0096] 2. Dynamic updates and adaptation results of weights:
[0097] The system uses the calculated compensation factor The original preset weight ratios are dynamically updated, and the update logic is as follows:
[0098]
[0099] This update logic follows a sensitivity dynamic balancing strategy: when the compensation factor... When the value is positive, the system actively reduces the weight of the backbone width. Simultaneously increase the weight of contrast. .
[0100] Updated width weight ;
[0101] Updated contrast weights .
[0102] During the aforementioned weight update process, the system employs a dynamic sensitivity balancing strategy to offset the shift in detection benchmarks caused by batch-to-batch variations in raw materials by adjusting the feature weights of different dimensions. Its core logic comprises the following two levels:
[0103] (1) The ebb and flow of feature sensitivity due to the numerical value of the index It is determined by the width of the main trunk. and edge contrast The weighted summation is obtained when the system detects a systematic deviation in a certain feature (such as width) due to the natural properties of the raw material (i.e., When the value is relatively large, the reliability of this feature in identifying quality anomalies decreases. Therefore, the strategy uses a compensation factor. Actively reduce the weight of the dimension (width) with high deviation. Simultaneously, the weights are compensated to another dimension (contrast). Above. This dynamic balance of subtraction and addition ensures the comprehensive indicators. The overall sensitivity to instantaneous quality defects in yarn remains constant, without any detection blind spots or allergens due to changes in raw material batches.
[0104] (2) Dynamically separating raw material background noise. The physical significance of this strategy lies in effectively separating the inherent physical properties of the yarn from instantaneous quality defects. Inherent properties refer to the natural fineness and natural hairiness levels that vary with different cotton batches; this data constitutes background noise and should not trigger speed adjustments. Quality defects refer to sudden changes in fineness or hairiness caused by mechanical failures or unstable tension during production. Through a dynamic sensitivity balancing strategy, the system locks the detection benchmark at the average operating level of the current batch in real time, ensuring that the detection system remains highly sensitive only to abnormal signals that deviate from the average of the current batch.
[0105] 3. Verification of technical effectiveness:
[0106] Through the aforementioned dynamic updates, the system has achieved detection and adaptation for different batches of raw yarn. This is because the main body of the new batch of yarn is naturally thicker (…). (The width is relatively large), so the system automatically fine-tunes and reduces the width weight. And correspondingly increased the contrast weight. This adaptive correction ensures that, during subsequent calculations of indicator values, the final quality indicators (such as the incidence of coarse sections) will not be systematically misjudged due to background differences in the raw materials themselves. The adjusted indicator values can more accurately reflect instantaneous quality fluctuations during the production process, rather than inherent differences in the properties of the raw materials.
[0107] Example 3
[0108] This embodiment provides a spinning quality control method based on linear array image texture feature analysis. This embodiment focuses on the graded alarm and synchronous shutdown process of the present invention when the yarn quality index still cannot fall back to the qualified range after speed control and weight correction.
[0109] 1. Continuous monitoring and judgment of abnormal states:
[0110] After the speed control action in Embodiment 1 is executed, the system completes a 3000ms delay masking. Subsequently, the vision processing module continuously acquires yarn images at the new rotation speed within a preset 5-second time window.
[0111] In a specific operating condition of this embodiment, due to severe wear of mechanical parts of the spinning machine (such as travelors) or mechanical entanglement in the yarn path, the index value remains above 20 (far exceeding the preset acceptable range of [0, 15]) even after speed reduction. At this time, the system determines that the index value shows a continuous abnormal trend.
[0112] After the speed regulation shield is removed, the vision processing module continuously acquires yarn images at the new rotation speed within a preset 5-second time window. To eliminate occasional noise interference, the system employs an anomaly detection formula based on sampling probability:
[0113]
[0114] in,
[0115] The total number of samples taken by the system to perform quality assessment within a preset time window;
[0116] : No. The real-time quality index values obtained from the second sampling calculation;
[0117] : The preset upper limit of the qualified range (in this embodiment, the value is 15);
[0118] : A conditional function that takes the value 1 if the condition within the parentheses is true, and 0 otherwise;
[0119] : The preset abnormality detection ratio threshold (set to 80% in this embodiment).
[0120] The formula is expressed in The percentage of times the indicator exceeds the standard during each sampling. The system does not determine an anomaly based on a single instance of exceeding the standard, but rather requires that more than 80% of the sampling points be in an unqualified state within a 5-second time window. This statistical determination method avoids false shutdowns caused by occasional noise from a single instance, ensuring the reliability of the protection action.
[0121] 2. Triggering and executing audible and visual alarm commands:
[0122] Once a persistent abnormal trend in the indicator is determined, the control unit immediately executes an audible and visual alarm command. The specific operation process is as follows:
[0123] Maintain current speed: The system locks the current operating speed and will not make any further invalid speed adjustment attempts to prevent frequent speed changes from aggravating mechanical damage.
[0124] Alarm circuit activation: The control unit activates the external alarm circuit via its output port (such as relay output or PLC register setting), driving the audible and visual alarm installed on top of the spinning machine. The alarm emits a high-decibel buzzing sound and a red flashing signal, alerting workshop personnel to a quality defect that cannot be repaired by automatic speed adjustment.
[0125] 3. Synchronous execution of stop commands:
[0126] While triggering the audible and visual alarm, to prevent the generation of large quantities of waste yarn or mechanical accidents caused by yarn breakage, the system further triggers a synchronous shutdown command based on preset safety logic:
[0127] Multi-axis synchronous reduction: The control unit sends a stop message to the central control mechanism of the spinning machine. Based on the mechanical transmission ratio, the system controls the main spindle motor, front roller motor, and winding mechanism to synchronously reduce their operating speeds.
[0128] Deceleration rate control: The shutdown process follows a preset deceleration rate (e.g., a 10% reduction in speed per second). This refers to the real-time rotational speed of each actuator during the shutdown process. To trigger the initial speed at the moment of shutdown, Let be the deceleration slope coefficient, then the formula for the shutdown control curve is:
[0129]
[0130] in,
[0131] : Calculated from the time the shutdown is triggered. The commanded rotation speed at that moment;
[0132] The operating speed of the currently controlled component (such as the spindle or roller) at the time of the alarm;
[0133] Deceleration slope (set to 0.1 in this embodiment, i.e., 10% deceleration per second).
[0134] The control unit is based on this unified slope function. Simultaneously, the calculations are performed and distributed to all drive motors. Since each component operates at its initial speed... Based on the same proportion The speed decreases synchronously, so the transmission ratio between the main shaft and the roller remains constant throughout the entire deceleration process until it stops, thus physically eliminating the risk of the yarn slack or breakage.
[0135] This embodiment demonstrates the logical integrity of the system's transition from automatic correction to safety defense. By setting up tiered protection, the system first attempts to restore production through speed adjustment and weight correction; after confirming that the anomaly cannot be eliminated, it guides manual intervention through audible and visual alarms and minimizes losses by using synchronized shutdown. This design not only protects cotton fiber raw materials from excessive waste but also effectively prevents severe physical damage caused by continuous operation of mechanical components under high failure conditions, demonstrating the high reliability of this invention in industrial applications.
[0136] In conclusion,
[0137] 1. Effectively suppress control feedback oscillations: By introducing delayed timing and shielding logic after speed regulation, the physical instability period caused by mechanical transmission inertia is avoided, preventing continuous misjudgments caused by visual detection response fluctuations and improving the stability of the closed-loop control system.
[0138] 2. Achieve automatic batch adaptation of raw materials: Introduce a sensitivity dynamic balancing strategy and use a weight adaptive correction mechanism to lock the detection benchmark at the average level of the current batch in real time. This effectively removes the background noise caused by the natural properties of the raw materials and solves the problem of frequent manual adjustment after changing raw materials.
[0139] 3. Balancing detection sensitivity and production safety: Employing a statistical judgment model based on sampling probability, it ensures accurate capture of continuous anomalies while filtering out occasional noise.
[0140] 4. A comprehensive equipment synchronization protection mechanism: In conjunction with the synchronous deceleration and shutdown logic, it ensures the motion synchronization of multi-axis actuators in fault conditions, prevents yarn breakage and secondary mechanical damage, and ensures the safety of the production process.
[0141] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A spinning quality control method based on linear array image texture feature analysis, characterized in that, Includes the following steps: S1. The electrical pulse signal emitted by the rotary encoder triggers the line scan camera to perform line-by-line shooting to obtain continuous yarn unfolding images; S2. Extract the average brightness value of the background area of the yarn unfolded image and perform brightness compensation on the main area of the yarn to obtain a preprocessed yarn image; S3. Calculate the width of the main image representing the yarn thickness and the contrast value of the surface pixel brightness in the preprocessed yarn image, and combine them to form feature data. S4. Perform weighted operations on the feature data based on the preset feature mapping relationship to obtain the index value representing the true quality state of the yarn; S5. If the value of the index exceeds the preset qualified range, a speed adjustment signal is sent to the drive motor of the spinning machine and a preset delay time is started, during which the speed adjustment judgment is blocked. S6. After the delay timer ends, the monitoring loop from image acquisition to quality assessment resumes.
2. The spinning quality control method based on linear array image texture feature analysis according to claim 1, characterized in that, Step S1 specifically includes: for each fixed angle rotation of the rotating shaft, the rotary encoder emits the electrical pulse signal; the line scan camera is in a waiting state, and whenever it receives an electrical pulse signal, it takes a line of images of the moving yarn, so that the actual physical length of the yarn corresponding to each line of images taken by the line scan camera remains consistent, and the continuously captured images are stitched together to form a continuous yarn unfolding image.
3. The spinning quality control method based on linear array image texture feature analysis according to claim 1, characterized in that, Step S2 specifically includes: in the yarn unfolding image, selecting blank background areas where no yarn passes through the edges; obtaining an average brightness value based on the brightness values of all pixels in the blank background area; comparing the average brightness value with a preset standard brightness value; if the average brightness value is lower than the standard brightness value, calculating the difference between the standard brightness value and the average brightness value, and superimposing the difference onto the pixels in the main yarn area for brightness compensation; subsequently, using Gaussian smoothing filtering to blur the entire image to eliminate instantaneous noise, resulting in a preprocessed yarn image.
4. The spinning quality control method based on linear array image texture feature analysis according to claim 1, characterized in that, Step S3 specifically includes: scanning line by line on the preprocessed yarn image to find the darkest and densest part of the image as the main body of the yarn; counting the number of pixels that the main body spans horizontally to obtain the width of the main body image representing the actual physical thickness of the yarn; scanning the pixels with alternating light and dark at the edge of the main body and calculating the degree of light and dark variation of the pixels to obtain a contrast value representing the degree of fuzziness on the yarn surface; and combining the width of the main body image with the contrast value to form feature data.
5. The spinning quality control method based on linear array image texture feature analysis according to claim 1, characterized in that, Step S4 specifically includes: multiplying the backbone image width and the contrast value by the corresponding preset weight ratio to obtain the weighted backbone width and weighted contrast; within a unit yarn length, counting the number of times the weighted backbone width exceeds a preset upper limit threshold as the thick spot occurrence rate, and counting the number of times the weighted backbone width is lower than a preset lower limit threshold as the detail occurrence rate; obtaining the average value of all the weighted contrast values within a unit yarn length to obtain the hairiness index; and combining the thick spot occurrence rate, the detail occurrence rate, and the hairiness index into an index value.
6. The spinning quality control method based on linear array image texture feature analysis according to claim 5, characterized in that, Step S4 further includes a weight adaptive correction step: continuously acquiring the average backbone image width and average contrast value of the spinning machine within a preset stable period; The width deviation between the average backbone image width and the preset standard yarn width is calculated, and the contrast deviation between the average contrast value and the preset standard hairiness index is calculated. Based on the width deviation and the contrast deviation, a compensation factor for the preset weight ratio is generated, and the preset weight ratio is dynamically updated using the compensation factor to achieve detection adaptation for different batches of raw material yarn.
7. The spinning quality control method based on linear array image texture feature analysis according to claim 1, characterized in that, Step S5 specifically includes: comparing the index value with a preset qualified number range; if the index value exceeds the qualified number range, calculating a new operating speed and converting the new operating speed into a speed regulation voltage signal and sending it to the drive motor; while sending the speed regulation voltage signal, starting a delay timing program; before the running time of the delay timing program reaches the preset delay duration, disabling the speed regulation control logic of the spinning machine, so that the mechanical components smoothly transition to the new operating speed.
8. The spinning quality control method based on linear array image texture feature analysis according to claim 1, characterized in that, Step S6 includes: after the delay timing program runs for the preset delay duration, unshielding the speed control logic and restoring the pulse trigger response of the line scan camera, acquiring the new yarn image after speed adjustment and stabilization; converting the new yarn image into the latest index value, continuously monitoring the change trend of the index value within a preset time window, and executing subsequent operation instructions according to the change trend.
9. The spinning quality control method based on linear array image texture feature analysis according to claim 8, characterized in that, The subsequent operation instructions include: if the index value shows a downward trend within the preset time window and is within the preset qualified number range, then execute the stable operation instruction, stop sending the voltage signal to change the speed, and control the spinning machine to maintain the current adjusted operating speed for stable operation; if the index value shows a continuous abnormal trend within the preset time window and exceeds the preset qualified number range, then execute the audible and visual alarm instruction, maintain the current operating speed, and connect the alarm circuit to drive the audible and visual alarm to issue an alarm signal.
10. The spinning quality control method based on linear array image texture feature analysis according to claim 10, characterized in that, Simultaneously with activating the audible and visual alarm circuit, the subsequent operation instructions also include: executing a synchronous stop instruction, sending an instruction to the control mechanism of the spinning machine to control each actuator to synchronously reduce its operating speed until it completely stops.