Round needle yarn tension self-adaptive control method and system

Abstract

This invention discloses a method and system for adaptive yarn tension control of a circular knitting machine, belonging to the field of program control system technology. To address the problems of lag and difficulty in handling yarn state fluctuations in existing feedback control, this invention employs a feedforward-feedback composite control scheme based on machine vision prediction. This method sets a visual detection point upstream of the tension sensor on the yarn feeding path, acquiring yarn images in real time and extracting the diameter fluctuation rate and fly hair area ratio. These are input into a pre-trained support vector regression model to generate a feedforward speed compensation amount. Simultaneously, the deviation between the measured value and the set value of the tension sensor is acquired, and a PID controller generates a feedback speed adjustment amount. Then, based on the fluctuation degree of the state characteristic parameters, the two are dynamically weighted and fused to drive the yarn feeding motor. This invention transforms "passive adjustment" into "active prediction," effectively reducing control lag and significantly improving the tension control accuracy and stability under high-speed knitting conditions.

Description

Technical Field

[0001] This invention relates to the field of program control system technology, specifically to a method and system for adaptive control of yarn tension in a circular needle machine. Background Technology

[0002] During the knitting process of a circular knitting machine (circular weft machine), the stability of yarn tension directly affects the fabric quality. Excessive tension fluctuations can lead to defects such as yarn breakage, horizontal stripes on the fabric, and uneven loops, seriously affecting product quality and production efficiency.

[0003] Existing tension control schemes are mainly divided into two categories. One is the mechanical passive type, which uses tension springs or damping plates to apply constant resistance to the yarn. This type of scheme cannot actively adjust according to changes in working conditions and has poor adaptability. The other is the closed-loop feedback type, which maintains constant tension by detecting the yarn tension value in real time and comparing it with the set value, and using a PID controller to adjust the speed of the yarn feeding motor.

[0004] However, existing closed-loop feedback control schemes have inherent flaws: their adjustment actions inevitably lag behind tension disturbances. From the tension sensor detecting fluctuations, to the PID controller calculating the adjustment amount, and then to the yarn feed motor changing its speed, the cumulative delays in each link of this control chain lead to a considerable length of defective fabric under high-speed weaving conditions. Simultaneously, quality fluctuations such as yarn thickness variations and fly waste can enter the weaving area unpredictably, and the resulting tension disturbances rely entirely on downstream delayed detection for response, making it difficult to fundamentally improve control quality.

[0005] Therefore, eliminating the lag in tension control and achieving advance prediction and compensation for changes in the yarn's own state during high-speed weaving has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To solve the above-mentioned technical problems, the technical solution provided by the present invention is as follows: An adaptive control method for yarn tension on a circular needle knitting machine includes the following steps: (S1) On the yarn feeding path, continuously acquire yarn images at the detection point upstream of the tension sensor; (S2) Through image processing, the diameter fluctuation rate and fly hair area ratio of the yarn are extracted in real time as state feature parameters; (S3) Input the diameter fluctuation rate and fly hair area ratio into the pre-trained tension disturbance prediction model to generate a feedforward rotation speed compensation amount to compensate for tension fluctuations caused by changes in the yarn's own state. (S4) Obtain the actual tension value measured by the tension sensor, and generate a feedback speed adjustment amount through the PID controller based on the deviation between the actual tension value and the target tension set value; (S5) The feedforward speed compensation amount and the feedback speed adjustment amount are dynamically weighted and fused according to the fluctuation degree of the state characteristic parameters to generate the final control signal and drive the yarn feeding motor.

[0007] Furthermore, the dynamic weighted fusion is performed according to the following formula: V out = V base + α·ΔV ff + (1-α)·ΔV fb Among them, V out For the final control signal, V base Based on the rotational speed, ΔV ff ΔV is the feedforward speed compensation amount. fb The feedback speed adjustment amount is α, which is a weighting coefficient and its value ranges from 0 to 1; α is positively correlated with the combined fluctuation amplitude of the diameter fluctuation rate and the fly feather area ratio.

[0008] Furthermore, when the overall fluctuation amplitude exceeds a preset threshold, α takes a value greater than 0.5; when the overall fluctuation amplitude is lower than the preset threshold, α takes a value less than 0.5.

[0009] Furthermore, the tension disturbance prediction model is a support vector regression model; the generated feedforward speed compensation amount includes: the tension disturbance prediction value output by the support vector regression model within a future preset time window, and then converting the tension disturbance prediction value into the feedforward speed compensation amount according to the dynamic characteristic model of the yarn feeding motor.

[0010] Furthermore, the training sample construction method of the support vector regression model is as follows: the diameter fluctuation rate and the area ratio of fly feathers collected in the historical operation are used as input features, and the tension value measured by the tension sensor after the corresponding time delay is used as the output label to construct the training sample set.

[0011] Furthermore, the yarn tension adaptive control method of the circular needle machine also includes an online update step (S6): periodically adding new samples by combining the real-time collected diameter fluctuation rate, fly hair area ratio, and the corresponding time-delayed measured tension value, and incrementally learning the support vector regression model.

[0012] Furthermore, the frame rate of the acquired yarn image is higher than the frequency corresponding to the knitting speed of the circular needle machine.

[0013] An adaptive control system for yarn tension of a circular needle machine, wherein the adaptive control method for yarn tension of the circular needle machine includes: The visual prediction unit is installed on the yarn feeding path and located upstream of the tension sensor. It is used to collect yarn images and extract the yarn diameter fluctuation rate and fly hair area ratio in real time. The feedforward controller has a built-in pre-trained support vector regression model to generate feedforward rotation speed compensation based on the diameter fluctuation rate and the fly feather area ratio. The feedback controller is used to generate a feedback speed adjustment amount based on the deviation between the actual tension value measured by the tension sensor and the target tension set value. The composite control module has its input terminals connected to the output terminals of the feedforward controller and the feedback controller, respectively. It is used to fuse the two control signals according to a preset dynamic weighting rule and output the final control signal to drive the yarn feeding motor.

[0014] The advantages of this invention compared to the prior art are: 1. By using machine vision to "pre-aim" at the yarn state fluctuation upstream of the tension detection point, and generating compensation instructions in advance through the tension disturbance prediction model, tension fluctuations are greatly reduced, and dynamic control accuracy under high-speed weaving is greatly improved. 2. Feedforward control is responsible for handling predictable disturbances caused by yarn-related factors, while feedback control is responsible for handling random disturbances such as loop vibration and sudden friction changes. Each has its own function. Furthermore, through a dynamic weighting mechanism, the weights of the two are adaptively adjusted according to the degree of yarn condition fluctuation, making the system more robust. 3. The support vector regression model can be continuously optimized through online incremental learning. After long-term operation, the prediction accuracy continues to improve. It can adapt to the characteristics of different batches of yarn raw materials and reduce the debugging cost after changing yarn or batch. 4. The visual inspection point only requires the installation of an industrial camera on the existing yarn feeding path, without changing the main mechanical structure of the circular needle machine. The modification cost is low and it is easy to promote and implement on existing equipment. Attached Figure Description

[0015] Figure 1 This is a schematic diagram illustrating the principle of the yarn tension adaptive control system for the circular needle machine of the present invention; Figure 2 This is a flowchart illustrating the adaptive control method for yarn tension in a circular needle machine according to the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0017] Example: The yarn tension adaptive control system for a circular needle machine according to an embodiment of the present invention, such as... Figure 1As shown, specifically, it includes a visual prediction unit, a feedforward controller, a tension sensor, a feedback controller, and a composite control module.

[0018] The visual prediction unit includes an industrial camera and an image processing module. The industrial camera is installed on the yarn feeding path, located between the tension sensor and the yarn bobbin, i.e., upstream of the tension sensor, preferably 10 to 15 centimeters away. The image acquisition frame rate of the industrial camera is higher than the frequency corresponding to the knitting speed of the circular needle machine to ensure that rapid changes in the yarn state are not missed. The image processing module is used to perform real-time analysis on each acquired yarn image to extract two key state feature parameters: the diameter fluctuation rate and the area ratio of fly hair and hairiness.

[0019] The feedforward controller incorporates a pre-trained support vector regression model. The training samples for this model are constructed as follows: during system debugging or historical operation, the diameter fluctuation rate and the ratio of fly hair area extracted by the image processing module at each sampling moment are recorded synchronously, along with the actual tension value collected by the tension sensor after a preset time delay (0.1 seconds in this embodiment). The diameter fluctuation rate and fly hair area ratio extracted by the image processing module are used as the model input features, and the actual tension value collected by the tension sensor after the preset time delay is used as the model output label, forming a supervised training sample set for offline training of the support vector regression model. After training, the model can predict the estimated tension disturbance within the specified time window based on the input diameter fluctuation rate and fly hair area ratio. The feedforward controller also incorporates a dynamic characteristic model of the yarn feeding motor, which can convert the estimated tension disturbance into the corresponding feedforward speed compensation amount ΔV. ff .

[0020] The feedback controller employs a conventional PID control algorithm. Its input is connected to a downstream tension sensor, and its output generates a feedback speed adjustment ΔV. fb .

[0021] The composite control module is connected to the output terminals of the feedforward controller and the feedback controller respectively, and executes the dynamic weighted fusion algorithm.

[0022] The yarn tension adaptive control method of the circular needle machine in this embodiment, such as Figure 2 As shown, the specific workflow is as follows: S1: Visual Prediction An industrial camera, positioned 10 to 15 centimeters upstream of the tension sensor along the yarn feeding path, continuously acquires image sequences of the running yarn at a frequency higher than the knitting speed of a circular needle knitting machine. The image processing module analyzes each frame: on one hand, it measures the diameter at multiple points along the yarn axis and calculates the ratio of its standard deviation to the mean to obtain the diameter fluctuation rate, which reflects the severity of changes in yarn thickness; on the other hand, it extracts fly hair and fuzz areas from the image through threshold segmentation and calculates the percentage of their area to the main yarn area to obtain the fly hair and fuzz area ratio, which reflects the amount of adhering substances and fuzz on the yarn surface. These two characteristic parameters are strongly correlated with the tension fluctuation of the subsequent yarn as it passes through the tension detection point.

[0023] S2: Feedforward compensation signal generation The diameter fluctuation rate and fly hair area ratio extracted in real time in step S1 are input into the support vector regression model that has been trained in the feedforward controller. This model outputs the estimated value of the tension disturbance that will occur in the next 0.1 seconds. The feedforward controller calls the built-in dynamic characteristic model of the yarn feeding motor to convert the estimated value of the tension disturbance into the corresponding feedforward speed compensation amount ΔV. ff， The physical meaning of this compensation amount is: by applying this compensation amount to the speed of the yarn feeding motor in advance, the tension disturbance introduced by the changes in the thickness and fineness of the yarn and the state of fly hair can be just offset.

[0024] S3: Feedback adjustment signal generation A downstream tension sensor continuously monitors the actual tension value of the yarn segment after passing the visual detection point. The feedback controller calculates the deviation between this actual tension value and the target tension setpoint, and generates a feedback speed adjustment ΔV through PID controller calculation. fb .

[0025] S4: Dynamic Weighted Composite Control The composite control module receives the feedforward speed compensation amount ΔV ff and feedback speed regulation amount ΔV fb The following formula is used for fusion: V out = V base +α·ΔV ff +(1-α)·ΔV fb Among them, V out To output the final control signal to the yarn feed motor servo driver, V base The base rotational speed is given by α, which is a dynamic weighting coefficient ranging from 0 to 1.

[0026] The dynamic adjustment rule for the weighting coefficient α is as follows: Calculate the combined fluctuation amplitude of the diameter fluctuation rate and the area ratio of fly hair and hair within the current time window. The combined fluctuation amplitude can be the weighted sum of the normalized fluctuation amplitudes of the two, or the larger of the two values. When the combined fluctuation amplitude exceeds the preset threshold, it indicates that the yarn itself is changing drastically. The tension disturbance that will be triggered at this time can be effectively predicted by the feedforward model. The system will set α to a value greater than 0.5 (e.g., between 0.7 and 0.9) to make the control output more reliant on the feedforward compensation amount and achieve advanced adjustment. When the combined fluctuation amplitude is lower than the preset threshold, it indicates that the yarn itself is relatively stable. At this time, the residual tension deviation is mainly caused by random factors such as loop shaking and sudden friction changes in the yarn channel. The system will set α to a value less than 0.5 (e.g., between 0.1 and 0.3) to make the control output return to the mode of feedback control and avoid amplifying the small errors of the feedforward model.

[0027] Finally, the control signal V out The servo driver sends the signal to the yarn feeding motor to precisely adjust the speed of the yarn feeding motor, thereby achieving adaptive feedforward-feedback composite control of yarn tension.

[0028] S6: Online Update During actual continuous operation, the system can periodically (e.g., after producing 1 kilometer of fabric or after each yarn bobbin replacement) initiate an online update program. This program uses the newly collected diameter fluctuation rate, fly hair area ratio, and actual tension value recorded by the tension sensor after a corresponding time delay to form new training samples, which are then used to perform incremental learning on the support vector regression model. Through this mechanism, the model can gradually adapt to the differences in the characteristics of different batches of yarn raw materials, as well as the slight drift in the mechanical state of the equipment after long-term operation, so that the prediction accuracy continues to improve with the accumulation of usage time.

[0029] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A method for adaptive control of yarn tension in a circular needle machine, characterized in that, Includes the following steps: (S1) On the yarn feeding path, continuously acquire yarn images at the detection point upstream of the tension sensor; (S2) Through image processing, the diameter fluctuation rate and fly hair area ratio of the yarn are extracted in real time as state feature parameters; (S3) Input the diameter fluctuation rate and fly hair area ratio into the pre-trained tension disturbance prediction model to generate a feedforward rotation speed compensation amount to compensate for tension fluctuations caused by changes in the yarn's own state. (S4) Obtain the actual tension value measured by the tension sensor, and generate a feedback speed adjustment amount through the PID controller based on the deviation between the actual tension value and the target tension set value; (S5) The feedforward speed compensation amount and the feedback speed adjustment amount are dynamically weighted and fused according to the fluctuation degree of the state characteristic parameters to generate the final control signal and drive the yarn feeding motor.

2. The adaptive yarn tension control method for a circular needle machine according to claim 1, characterized in that, The dynamic weighted fusion is performed according to the following formula: V out = V base + a·ΔV ff + (1-a)·ΔV fb Among them, V out For the final control signal, V base Based on the rotational speed, ΔV ff ΔV is the feedforward speed compensation amount. fb The feedback speed adjustment amount is α, which is a weighting coefficient and its value ranges from 0 to 1; α is positively correlated with the combined fluctuation amplitude of the diameter fluctuation rate and the fly feather area ratio.

3. The adaptive yarn tension control method for a circular needle machine according to claim 2, characterized in that, When the overall fluctuation amplitude exceeds the preset threshold, α takes a value greater than 0.5; when the overall fluctuation amplitude is lower than the preset threshold, α takes a value less than 0.

5.

4. The adaptive yarn tension control method for a circular needle machine according to claim 1, characterized in that, The tension disturbance prediction model is a support vector regression model; the generated feedforward speed compensation amount includes: the tension disturbance prediction value output by the support vector regression model within a future preset time window, and then the tension disturbance prediction value is converted into the feedforward speed compensation amount according to the dynamic characteristic model of the yarn feeding motor.

5. The yarn tension adaptive control method for a circular needle machine according to claim 4, characterized in that, The training sample set for the support vector regression model is constructed by using the diameter fluctuation rate and the ratio of fly feather area collected during historical operation as input features, and the tension value measured by the tension sensor after the corresponding time delay as the output label.

6. The yarn tension adaptive control method for a circular needle machine according to claim 5, characterized in that, It also includes an online update step (S6): periodically adding new samples by combining the real-time collected diameter fluctuation rate, fly feather area ratio, and the corresponding time-delayed measured tension value, and incrementally learning the support vector regression model.

7. The adaptive yarn tension control method for a circular needle machine according to claim 1, characterized in that, The frame rate of the acquired yarn image is higher than the frequency corresponding to the knitting speed of the circular needle machine.

8. A yarn tension adaptive control system for a circular needle machine, used to execute the yarn tension adaptive control method for a circular needle machine as described in any one of claims 1 to 7, characterized in that, include: The visual prediction unit is installed on the yarn feeding path and located upstream of the tension sensor. It is used to collect yarn images and extract the yarn diameter fluctuation rate and fly hair area ratio in real time. The feedforward controller has a built-in pre-trained support vector regression model to generate feedforward rotation speed compensation based on the diameter fluctuation rate and the fly feather area ratio. The feedback controller is used to generate a feedback speed adjustment amount based on the deviation between the actual tension value measured by the tension sensor and the target tension set value. The composite control module has its input terminals connected to the output terminals of the feedforward controller and the feedback controller, respectively. It is used to fuse the two control signals according to a preset dynamic weighting rule and output the final control signal to drive the yarn feeding motor.

Images