Automatic deviation correction control method for stay wire winding machine based on image edge characteristics

Abstract

The invention relates to the technical field of industrial automation control, in particular to an automatic deviation correction control method for a stay wire winding machine based on image edge features. The method comprises the following steps: collecting edge image data of a strip, performing feature extraction, and calculating real-time transverse deviation of the strip; the real-time angular velocity of the reel is collected, the real-time reel diameter of the reel is calculated, and an inertia compensation factor is constructed by using the rigid body dynamics principle; the inertia compensation factor is used for correcting control parameters of the active disturbance rejection controller in real time, the real-time transverse deviation of the strip is combined to calculate the deviation correction control quantity, the deviation correction control quantity is converted into a control signal, and a deviation correction executing mechanism is driven to make related adjustment. Through explicit physical model compensation, the problem that the load inertia fluctuates severely due to the change of the rolling diameter in the rolling process is solved, and the product quality is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology, specifically to an automatic correction control method for a wire winding machine based on image edge features. Background Technology

[0002] In a high-speed packaging material production line, the wire winding machine is responsible for winding the slit strip into a coil. In order to ensure the neatness of the coil, the correction system needs to detect the positional deviation of the strip in real time, and then drive the actuator to eliminate the deviation.

[0003] Existing correction control methods typically employ fixed-parameter PID control or standard active disturbance rejection control. However, in practical applications, the winding process is a typical process of variable mass and variable inertia. As production progresses, the reel diameter gradually increases from empty to full. According to the principles of rigid body dynamics, the total load mass is proportional to the square of the radius. This means that the load inertia at full winding may be several times or even tens of times that at empty winding. This significant change in physical parameters leads to serious control contradictions: If the control parameters are tuned according to the large inertia condition of a full roll, excessively large control parameters will lead to severe overshoot or even high-frequency oscillations in the empty roll stage. If the parameters are tuned according to the small inertia condition of an empty roll, the control force will be insufficient to overcome the huge load in the full roll stage, resulting in sluggish response and a significant decrease in correction accuracy. Although traditional active disturbance rejection control has a certain disturbance rejection capability, it treats all model mismatches as internal disturbances for observation. When model parameters such as inertia change too much, it will consume the bandwidth of the observer, resulting in a decrease in disturbance rejection capability. Therefore, a correction control method that can adapt to drastic changes in inertia across the entire roll diameter is needed. Summary of the Invention

[0004] To address the problem that traditional methods cannot adaptively correct deviations across the entire roll diameter when inertia changes drastically, this invention provides an automatic deviation correction control method for a wire winding machine based on image edge features. This method includes: The system collects and preprocesses edge image data of the strip at the entrance of the wire winding machine, the real-time angular velocity of the reel, and the real-time linear velocity of the traction motor to obtain the actual position of the strip edge. Based on the actual position and a preset reference position, the real-time lateral deviation of the strip is calculated. Based on the real-time angular velocity and the real-time linear velocity, the real-time diameter of the reel is calculated. Using the real-time diameter and the principle of rigid body dynamics, the inertia compensation factor is calculated. The inertia compensation factor is used to correct the control parameters of the active disturbance rejection controller of the wire winding machine in real time. The real-time lateral deviation is input into the corrected active disturbance rejection controller to calculate the correction control quantity. The correction control quantity is sent to the correction actuator to drive the correction actuator to make relevant adjustments.

[0005] This invention constructs an inertia compensation factor that can sense changes in the physical properties of the load in real time by collecting real-time motion data of the reel and combining it with the principle of rigid body dynamics. This factor is used to dynamically correct the core parameters of the active disturbance rejection controller, solving the problems of easy oscillation in the empty winding stage and slow response in the full winding stage of the traditional fixed gain controller. This achieves consistency in the correction accuracy and response speed of the wire winding machine throughout the entire production cycle of the wire diameter.

[0006] Furthermore, the value of the inertia compensation factor is non-linearly positively correlated with the ratio of the real-time roll diameter to the empty roll radius of the roll.

[0007] This invention clarifies that the inertia compensation factor is positively correlated with the square of the real-time winding diameter. This mathematical model, based on rigorous physical derivation, effectively restores the nonlinear growth trend of the total load mass during the winding process of a cylinder. Compared with conventional linear interpolation or piecewise lookup table methods, this model has high physical fidelity under large inertia conditions such as full winding, effectively avoiding control residuals caused by model estimation deviations.

[0008] Furthermore, the control parameters are corrected in real time, including: obtaining the reference control parameters of the active disturbance rejection controller, dividing the reference control parameters by the inertia compensation factor to obtain the control parameters after real-time correction, so that the correction control quantity output by the active disturbance rejection controller increases accordingly when the inertia compensation factor increases.

[0009] This invention employs a correction strategy that divides the reference control parameters by the inertia compensation factor, establishing an adaptive mechanism inversely proportional to the gain and inertia. This means that when the load inertia increases, the controller will automatically output stronger control energy to maintain the predetermined acceleration response capability, ensuring that the final calculation result will not change with the change of the physical state of the strip.

[0010] Further, the preprocessing includes: setting a contrast threshold, binarizing the edge image data of the strip to obtain a binarized image; extracting pixel gradient abrupt change points in the binarized image as edge feature points; and using the least squares method to perform line fitting on the edge feature points to obtain the actual position of the edge of the strip.

[0011] Furthermore, the contrast threshold was calibrated through a light adaptability experiment.

[0012] This invention calibrates the contrast threshold through an illumination adaptability experiment, enabling the algorithm to maintain accurate capture of the strip edge under industrial lighting conditions of different time periods and intensities. This effectively suppresses the jump in detection data caused by ambient light fluctuations or shadow interference, ensuring the accuracy of the correction control.

[0013] Furthermore, the automatic correction control method for the wire winding machine based on image edge features also includes: performing safety limiting processing on the correction control amount, wherein the safety limiting processing includes amplitude limiting and rate of change limiting.

[0014] Furthermore, the amplitude limiting includes setting an upper limit threshold for the correction control quantity, the upper limit threshold being set based on the rated voltage of the correction actuator.

[0015] Furthermore, the correction mechanism is specifically a servo motor module that drives the winding table of the wire winding machine to move axially back and forth.

[0016] Furthermore, the rate of change limiting includes: limiting the rate of change of the correction control quantity to not exceed a preset slope threshold, which is calibrated through a mechanical response test.

[0017] This invention calibrates the slope threshold of the rate of change limit through mechanical response testing, strictly locking the dynamic change range of the control signal outside the physical resonance frequency of the actuator. This effectively prevents high-frequency oscillations induced by sudden changes in the correction command, ensuring efficient correction response while also taking into account the operational safety and service life of the wire winding machine.

[0018] Furthermore, the reference control parameters include: controlling the wire winding machine to be in an empty winding state, inputting a step test signal to the active disturbance rejection controller, adjusting the parameters of the active disturbance rejection controller until the step response curve meets the preset standard, and recording the control parameters at this time as the reference control parameters.

[0019] The present invention has the following beneficial effects: This invention solves the physical contradiction of traditional fixed parameters being unable to push the winding during the full winding stage and over-pushing during the empty winding stage by introducing an inertia compensation factor based on rigid body dynamics. Regardless of the winding stage, the dynamic response characteristics to deviation signals remain highly consistent, ensuring the correction effect across the entire winding diameter. The inertia compensation factor constructed in this invention is derived from physical formulas, eliminating the need for engineers to repeatedly debug based on experience. It has high physical robustness, and the algorithm can automatically adapt as long as the geometric data of the strip is input, thus reducing the debugging threshold. This invention combines environmental adaptability calibration at the vision end with mechanical response boundary locking at the execution end, ensuring the accuracy of results in complex and ever-changing industrial environments, regardless of whether they are facing light interference or sudden load changes. Attached Figure Description

[0020] Figure 1 This is a flowchart of the automatic deviation correction control method for a wire winding machine based on image edge features provided in an embodiment of the present invention; Figure 2 This is a comparison diagram of the correction response of the present invention and the prior art under full-roll conditions, provided by an embodiment of the present invention. Detailed Implementation

[0021] It should be noted that the wire winding machine described in this specification refers to equipment used for winding various long strip materials, including but not limited to flat cables, metal strips, plastic films, and other materials collectively referred to as strips.

[0022] This invention provides an automatic deviation correction control method for a wire winding machine based on image edge features, referring to... Figure 1 This includes steps S1-S4: S1: Data Acquisition and Preprocessing.

[0023] Specifically, the edge image data of the strip at the entrance of the wire drawing and winding machine and the real-time angular velocity of the reel are collected and preprocessed to obtain the actual position of the strip edge. Based on the preset reference position, the real-time lateral deviation of the strip is calculated.

[0024] 1. Hardware environment construction and high-frequency data acquisition To obtain high-quality raw data, this embodiment sets up a detection station in a straight section 300mm to 500mm before the reel inlet of the wire winding machine. Based on backlight projection imaging, a high-brightness LED parallel surface light source is installed below the strip, and an image acquisition device is vertically installed directly above the strip. This results in a high-contrast image with a bright background and a completely black target, eliminating the interference of strip surface texture on edge detection. (1) Image acquisition device: In this embodiment, a high-speed industrial-grade linear CCD camera with a frame rate of 2000fps from the Basler racer series is selected, along with a 50mm low-distortion industrial lens, and the acquisition frequency is set to 1kHz. (2) Motion sensing device: In this embodiment, a 23-bit multi-turn absolute encoder installed at the tail of the winding servo motor is used to collect the real-time angular velocity of the winding reel, and an incremental encoder installed on the traction roller is used to collect the traction linear velocity. (3) Data synchronization: The FPGA-based hardware trigger mode is adopted. The FPGA sends exposure trigger signals to the camera at a 1ms cycle and stores the current value of the encoder to ensure that the acquired data is strictly aligned in the time dimension.

[0025] 2. Image preprocessing and edge feature extraction Image processing unit: In this embodiment, an Advantech IPC-610 industrial control computer equipped with an Intel Core i7-9700K processor and an NVIDIA GeForce RTX 3060 graphics card is selected as the computing platform, in conjunction with an FPGA-based PCIe image acquisition card; The image processing unit receives the raw grayscale image acquired by the image acquisition device and performs preprocessing to ensure that the edges of the strip can still be extracted even under interference from complex industrial environments such as oil mist and dust. Specifically: (1) In order to ensure the real-time performance of the algorithm, this embodiment uses a dynamic tracking strategy to construct the ROI region. For the original grayscale image collected at a certain moment, the horizontal coordinate of the strip edge detected at the previous moment is used as the center, and it is extended to the left and right by 128 pixels. At the same time, the current 1024 rows of scanning data are extracted in the vertical direction to construct a rectangular ROI region with a size of 256×1024 pixels. In order to filter out the noise caused by factors such as the micro-texture of the strip surface or suspended dust in the air, this embodiment uses a Gaussian filter kernel with a size of 5×5 and a standard deviation of 1.2 to perform convolution operation to obtain the filtered ROI image. (2) Set a contrast threshold in advance, traverse each pixel of the filtered ROI image, if its gray value is greater than the contrast threshold, it is determined to be a bright background and marked as 1, otherwise it is determined to be a light-blocking strip and marked as 0, and a binarized image is obtained. It should be noted that the contrast threshold was calibrated through a lighting adaptability experiment, the specific steps of which are as follows: ① With the wire winding machine stopped and the strip stationary, three typical production site lighting environments were constructed using an adjustable brightness industrial simulated light source: High-intensity lighting conditions: Illuminance of 2000 Lux, corresponding to direct midday sunlight or high-intensity workshop lighting; Low-light conditions: Illuminance of 200 Lux, corresponding to nighttime or indoor environments with obstructions; Flicker and shadow conditions: Dynamic alternating illuminance is adopted, with a background base illuminance of 600 Lux, a shadow area illuminance of 150 Lux, and a peak brightness flicker of 1800 Lux; a frequency-converted square wave signal is input to the light source controller using a signal generator to drive the light source to flicker periodically at a frequency that linearly increases from 20 Hz to 60 Hz; a rotating shielding disc with a rotation speed of 120 rpm is set between the light source and the strip. When the blades of the shielding disc pass through the field of view, they will project a fast-moving solid shadow on the surface of the strip.

[0026] Under each of the above operating conditions, 50 frames of images of the ROI region were collected to construct a benchmark sample library containing 150 images.

[0027] ② The industrial camera captures 8-bit grayscale images, and the pixel brightness value range is fixed at [0, 255]. Therefore, an integer variable is defined, whose value range covers [0, 255]. This integer variable starts from 0 and increases by 1 each time until it ends at 255. In each loop, this integer variable is called the temporary threshold. In each loop, each image in the reference sample library is binarized using the current temporary threshold. If the grayscale value of a pixel is greater than the temporary threshold, it is determined to be the background and recorded as 1; otherwise, it is determined to be the strip material and recorded as 0. Finally, 150 binarized images are obtained in the current loop. ③ Scan the 512th row of the binarized image from left to right. When a pixel's logical value changes from 1 to 0 for the first time, record the horizontal coordinate of that position as the left edge coordinate. When a pixel's logical value changes from 0 to 1 for the first time, record the horizontal coordinate of that position as the right edge coordinate. If two left edge coordinates or two right edge coordinates are detected in the same row, discard the frame data. ④ For each binarized image, subtract the left edge coordinate from the right edge coordinate, and calculate the pixel width. For a given loop, repeat the above operation for all binarized images to obtain a pixel width sequence. Calculate the standard deviation of this pixel width sequence. After the temporary threshold iterates from 0 to 255, find the temporary threshold corresponding to the smallest standard deviation, which is the contrast threshold.

[0028] (3) The binarized image is scanned horizontally. In this embodiment, the Sobel operator is used as the edge probe to calculate the gray-level gradient of each pixel in the horizontal direction. The gray-level gradient is the largest at the black-and-white boundary. The pixel with the largest gray-level gradient value in each row is marked as the edge feature point of that row. Since the height of the ROI region is 1024 rows, 1024 coordinate points are finally extracted. These points are connected to obtain the preliminary outline of the strip edge in the image. (4) Due to the pixel grid effect of the camera sensor, the 1024 extracted coordinate points often exhibit uneven micro-serrations. In addition, the slight mechanical vibration during the operation of the device causes these coordinate points to have random jump deviations on the straight line. Therefore, it is necessary to perform linear regression on these 1024 coordinate points to reconstruct a unique physical edge. In this embodiment, the least squares method is used. It is assumed that there is an ideal straight line that passes through these discrete coordinate points. The slope and intercept of this line are continuously adjusted iteratively to minimize the sum of the squares of the vertical distances from these 1024 coordinate points to this ideal straight line, and finally the edge line equation is obtained. (5) In order to eliminate the interference of longitudinal motion on unknown detection, a constant reference observation line needs to be established. Since the ROI region represents the spatial distribution of the strip in a very short time, its geometric center position represents the average position of the strip at the current moment, and this position is least affected by the jitter at both ends of the edge. Therefore, in this embodiment, the geometric center row of the ROI region is selected as the reference observation line. Since the height of the ROI region selected in this embodiment is 1024 rows, the geometric center row is 512 rows. Substitute the coordinates of the reference observation line into the edge line equation to calculate the x-coordinate of the intersection point of the edge line and the reference observation line, which is the actual position of the edge of the strip at the current moment. It should be noted that the actual position specifically refers to the x-coordinate of the point where the strip edge intersects with the reference observation line. Although the physical coordinates of the strip edge have two-dimensional coordinates, the y-coordinate is a preset fixed parameter and does not contain deviation information. Therefore, when outputting to the control algorithm, only the scalar value of the x-coordinate is extracted to represent the actual position of the strip edge.

[0029] As can be seen, although the acquired grayscale image is a discrete pixel grid, the actual position of the strip edge obtained through the above fitting and calculation is a floating-point number with decimal places. Its accuracy breaks through the limitation of physical pixel resolution and achieves sub-pixel level precise positioning.

[0030] 3. Deviation Calculation Through mechanical calibration, the optical center of the image acquisition device is aligned with the mechanical center of the wire winding machine. Since the image acquisition device is centrally located, the horizontal central axis of the ROI region represents the ideal path of the strip. Therefore, the coordinates at half the width of the image of the ROI region are taken as the reference position. The real-time lateral deviation of the strip is obtained by subtracting the reference position from the actual position of the strip edge.

[0031] It should be noted that the specific hardware selection and parameter settings described above are merely preferred embodiments of the present invention and should not be construed as limiting the scope of protection of the present invention. For example, those skilled in the art can replace the linear CCD camera with a high frame rate area array CMOS camera. For high-temperature or black light-absorbing strips, an infrared thermal imager or a laser contour sensor can also be used to obtain the edge position information of the strip. Edge feature extraction is not limited to the Sobel operator and the least squares method. Using the Canny operator, Hough transform, RANSAC algorithm, or semantic segmentation network based on deep learning to achieve edge localization are all equivalent substitutions of the present invention based on the technical concept of image edge features. Any technical solution that obtains the strip position information based on a non-contact visual sensor and calculates the lateral deviation accordingly should be included within the scope of protection of the present invention.

[0032] S2: Real-time roll diameter calculation.

[0033] Specifically, this step aims to calculate the roll diameter parameter, which cannot be directly measured, through kinematic relationships. In order to overcome the calculation divergence problem at low speeds and the data noise caused by mechanical vibration, this invention constructs a data processing link that includes singular value suppression, core calculation, and digital filtering.

[0034] Since the traction line speed is determined not by the wire winding machine but by the rotational speed of the traction roller, the motion sensor reads the real-time linear speed of the traction motor via an industrial fieldbus, denoted as... In this embodiment, the linear velocity read is in meters per minute (m / min), which needs to be converted to the standard unit meters per second (m / s). ,of For a moment The traction line speed; it should be noted that in a constant line speed winding process, It is usually a constant value, but it changes linearly during the equipment start-up and shutdown phases; Similarly, the motion sensor reads the real-time rotational speed of the take-up servo motor at a sampling period of 1ms, denoted as . Considering that wire winding machines are usually connected to speed reducers, this embodiment presets a speed reduction ratio of 5, and calculates the real-time angular velocity of the reel: ,in, For the reel at time Real-time angular velocity, It is pi; Because the real-time angular velocity of the reel approaches 0 at the moment the cable reel is started or stopped, performing division directly would result in a denominator of 0 or extremely small, causing the calculation result to approach infinity. To prevent program crashes or outputting incorrect instructions, this embodiment sets up conditional branch logic triggered by an angular velocity threshold: the minimum effective angular velocity threshold is set to 0.5 rad / s, when... When the real-time roll diameter is less than the minimum effective angular velocity threshold, the empty roll radius is set to the preset value; when... When the actual roll diameter is greater than or equal to the minimum effective angular velocity threshold, the real-time roll diameter of the reel is calculated normally.

[0035] It should be noted that the preset empty roll radius is a constant determined based on the physical specifications of the reel. In this embodiment, the operator inputs the currently used reel model through the human-machine interface of the touch screen, and the control system automatically calls the corresponding radius value as the empty roll radius according to the preset database; the database contains common standard reel models and their corresponding physical radius values.

[0036] Based on the principles of rigid body circular kinematics, linear velocity equals angular velocity multiplied by the radius. Therefore, at time [time missing], the reel... The real-time volume diameter is: ,in, For the reel at time Real-time roll diameter.

[0037] It should be noted that the indirect calculation method of dividing linear velocity by angular velocity used in the embodiments of the present invention is only a preferred solution and should not be construed as a limitation on the scope of protection of the present invention. For example, those skilled in the art can directly measure the distance from the surface of the reel to the sensor by installing a laser rangefinder on the side of the reel to calculate the real-time reel diameter, or calculate it by the thickness of the strip and the total number of rotations of the reel. Any technical means that can obtain the reel diameter value in real time and use it for subsequent inertia compensation are equivalent substitutions for the feature of the present invention of calculating the real-time reel diameter.

[0038] S3: Inertia compensation factor calculation.

[0039] Specifically, this step aims to establish a mapping model connecting geometric dimensions and physical dynamics. Based on the real-time diameter of the reel, and combined with the principle of rigid body dynamics, the mass change rate of the wire winding machine load is calculated in real time to obtain the inertia compensation factor, which is used to correct the control parameters of the wire winding machine's active disturbance rejection controller in real time.

[0040] To eliminate the influence of physical unit dimensions on the control algorithm, the normalized roll diameter ratio is first calculated based on the real-time roll diameter and the preset empty roll radius: ,in, The normalized roll diameter ratio represents the value at time t. The growth rate of the reel's geometry relative to its initial state, for example... Represents the moment The reel is empty; For the reel at time Real-time roll diameter; This is the preset empty volume radius.

[0041] When the correction actuator performs a correction action, its physical essence is to drive the entire winding table to perform axial translational motion on the guide rail. According to Newton's second law in rigid body dynamics, its dynamic response capability depends on the ratio of the driving force to the total load mass. In an ideal situation, regardless of whether the reel is empty or full, the dynamic response capability of the correction actuator is consistent for the same deviation signal. However, as production progresses, the mass of the reel will increase significantly with the increase of the reel diameter. If the driving force of the correction actuator remains unchanged, according to Newton's second law, its dynamic response capability is inversely proportional to the mass of the reel, eventually leading to a gradual slowdown in response.

[0042] Treating the reel as a cylinder of uniform density, and ignoring the minute mass differences in the reel's core, the mass of the reel is proportional to its volume:

[0043] in: For a moment The mass of the reel represents the total load mass of the wire winding machine during the correction process; The density of the strip; This refers to the physical width of the strip. Pi; For the reel at time Real-time roll diameter; The item represents the volume of the reel; because , , All of these are physical constants during the production process, which means that... and The total load mass is directly proportional to the square of the real-time roll diameter of the reel. If the web guiding actuator is to maintain the same dynamic response capability as the empty roll condition when the total load mass gradually increases, its output driving force must increase synchronously with the total load mass.

[0044] Although the reel rotates at high speed during winding, the web guiding motor is not responsible for driving its rotation, but rather for overcoming its mass to move laterally, in conjunction with the aforementioned... and This direct proportionality means that as the diameter of the coil increases, the physical resistance faced by the correction actuator will increase exponentially. To establish the mapping relationship between physical load and control algorithm, this invention analyzes the internal algorithm principle of the active disturbance rejection controller (ADRC): When calculating how to eliminate real-time lateral deviation, the ADRC essentially calculates the ideal linear acceleration that the winding table must generate to eliminate the real-time lateral deviation. This acceleration is determined by the total disturbance and the correction control quantity. According to Newton's second law in rigid body dynamics, under the premise of constant acceleration, the required correction control quantity is proportional to the total load mass of the correction actuator. This means that, under the premise of constant correction control quantity, the actual acceleration capability of the correction actuator is inversely proportional to the total load mass. Therefore, as... As the disturbance density increases, the control parameters of the active disturbance rejection controller should be adjusted according to... The ratio is attenuated; Based on this, an inertia compensation factor is constructed to assess the difficulty of correction by the correction actuator in real time: ,Right now ,in, Normalized volume diameter ratio; It is a moment The inertia compensation factor reflects the moment The degree of difficulty in correcting errors by the corrective enforcement agency is a multiple of that in the case of empty rolls.

[0045] When the cable reel is working normally, the real-time diameter of the reel will not decrease; any... A decrease in the value indicates a measurement error, therefore it is necessary to compare the time intervals. The inertia compensation factor is compared with the value of the inertia compensation factor at the previous moment, and the final output inertia compensation factor with the larger value is used to eliminate the risk of oscillation caused by interference such as sensor jitter.

[0046] Although the total load mass is only quadratically related to the real-time roll diameter, in industrial settings, the roll diameter often varies greatly, for example, from an empty roll of 100mm to a full roll of 1000mm, a tenfold change. This means that the total load mass will fluctuate dramatically by a factor of 100. For such an inertia change spanning two orders of magnitude, traditional fixed-gain PID controllers are completely ineffective, resulting in either oscillation when empty or no movement when full. Even conventional linear gain scheduling struggles to match this non-linear parabolic growth. This invention innovatively introduces an adaptive linearization strategy based on a mass model. Addressing the physical characteristic of the total load mass increasing non-linearly quadratically during the winding process of a wire winding machine, an inertia compensation factor is constructed to match the curve of total load mass change. Combined with the internal model correction of the active disturbance rejection controller, this achieves adaptive control across the entire roll diameter, resolving the physical contradictions of easy oscillation when empty and delayed response when full.

[0047] It should be noted that the use of a square relationship to construct the inertia compensation factor in this embodiment is only a preferred implementation method based on a standard cylindrical model, which aims to achieve accurate inertia compensation and should not be construed as a limitation on the scope of protection of this invention. For example, those skilled in the art can use mass approximation models, polynomial fitting curves, piecewise linear interpolation, etc. to determine the compensation coefficients. As long as the core logic is to increase the control parameters as the roll diameter increases, they all fall within the scope of protection of this invention.

[0048] S4: Calculation of corrective control quantity.

[0049] Specifically, this step aims to solve the problem that traditional fixed-parameter controllers cannot adapt to the huge inertia changes during the winding process. Based on the real-time lateral deviation obtained in S1 and the inertia compensation factor obtained in S3, the correction control quantity used to drive the correction actuator to make relevant adjustments is calculated.

[0050] Traditional fixed-gain active disturbance rejection controllers cannot simultaneously meet the control requirements of wire winding machines when the winding is empty and the winding is full. Therefore, this embodiment adopts an adaptive strategy with gain-inertia ratio to ensure that the active disturbance rejection controller outputs a correction control quantity that matches the current physical load across the entire winding diameter. Obtain the reference control parameters of the active disturbance rejection controller of the wire winding machine, denoted as . This is an unchanging reference standard, representing the optimal control parameters that enable the wire winding machine to fully follow the correction instructions when the winding is empty; Since the inertia compensation factor represents the multiple of the difficulty of correction by the correction actuator relative to the empty roll, the control parameters are calculated based on this: ,in, For a moment Control parameters; For a moment The inertia compensation factor; a division correction strategy is adopted, dividing the reference control parameters by the inertia compensation factor, as... Constant changes The parameters of the active disturbance rejection controller are constantly changing, causing them to be continuously corrected as the operating conditions change. This prevents the active disturbance rejection controller from overcompensating, thus ensuring the accuracy of the observer's estimation of the real disturbance across the entire convolution range without sacrificing the observation bandwidth. This is something that cannot be achieved by simply adjusting the external PID gain.

[0051] It should be noted that the baseline control parameters This was determined through experiments, as detailed below: Remove all the strip from the wire rewinder to ensure the real-time roll diameter of the reel. Equal to the preset empty volume radius In this embodiment, a servo motor module that drives the winding table of the wire winding machine to move axially back and forth is used as the correction actuator. The correction actuator is positioned at the physical midpoint of the winding stroke. A step test signal is input to the setpoint of the active disturbance rejection controller. In this embodiment, a 5mm lateral displacement change is input. The step response curve of the actual position of the correction actuator is monitored in real time using an oscilloscope on a host computer. Keeping other parameters constant, the bandwidth and control parameters of the active disturbance rejection controller are gradually adjusted. The bandwidth parameters include the observer bandwidth and the controller bandwidth. The shape of the step response curve is observed. When the response rise time is less than 100ms, the overshoot is less than 5%, and the steady-state error is 0, the control parameters at this time are taken as the reference control parameters.

[0052] After receiving the real-time lateral deviation, the active disturbance rejection controller (ADRC) calculates the ideal motion acceleration that the winding table should have in order to eliminate the real-time lateral deviation based on its magnitude and trend. In this calculation step, the ADRC does not consider the load of the correction actuator and only focuses on how to eliminate the deviation in geometric position. Meanwhile, the extended state observer inside the active disturbance rejection controller will output a total disturbance estimate in real time. This value represents all negative disturbances that hinder the movement of the winding table, such as slide rail friction, lead screw backlash, eccentric vibration, etc. The active disturbance rejection controller will add the aforementioned ideal motion acceleration to the total disturbance estimate to obtain a net demand driving force. In conventional algorithms, the calculated net demand driving force is often directly used as the control quantity output. However, in this invention, since the control parameters have been corrected by the inertia compensation factor, it is equivalent to pre-embedding a physical lever in the algorithm: when the wire winding machine is in an empty winding state, the inertia compensation factor is 1, and the control parameter is equal to the baseline control parameter; when the wire winding machine is in a fully wound state, the inertia compensation factor will increase by tens of times, causing the control parameter to become significantly smaller than the baseline control parameter. At this time, based on the division operation logic, the active disturbance rejection controller divides the net demand driving force by the control parameter, and the final calculated correction control quantity will be amplified in the opposite direction.

[0053] As can be seen, this invention introduces an inertia compensation factor. This allows the active disturbance rejection controller to sense the load of the web correction actuator in real time. For the same real-time lateral deviation, it will output several times the command when the wire winding machine is fully wound, which is the same as when it is empty. This ensures that the web correction actuator can maintain a high response speed and control accuracy under any winding diameter.

[0054] Furthermore, the correction of control parameters in this invention is fundamentally different from existing PID variable gain technology. Traditional PID variable gain simply amplifies the output ratio outside the controller, essentially using excessive gain to brute-force model error. This often compresses the phase margin of the correction actuator, resulting in poor stability when fully wound. In contrast, this invention corrects the compensation parameters of the extended state observer inside the active disturbance rejection controller. It can ensure the estimation accuracy of the observer for the real disturbance across the entire winding range without sacrificing the observation bandwidth, which cannot be achieved by simply adjusting the external PID gain.

[0055] It should be noted that although the correction control quantity obtained through the above operations theoretically meets the dynamic requirements, in practical applications, the bearing limit of the correction actuator must be considered. To prevent motor overload or mechanical damage caused by excessively large calculated correction control quantities, this embodiment performs dual safety limiting processing on the calculated correction control quantity, as follows: 1. Amplitude Limitation: In this embodiment, the correction actuator uses a servo motor module that drives the winding table of the wire winding machine to move axially back and forth. The hardware manual of the servo motor module is read to obtain its maximum allowable input control voltage. To allow for a safety margin, this embodiment selects 95% of the maximum allowable input control voltage as the upper limit threshold, denoted as... If the corrective control quantity is greater than the upper limit threshold, the upper limit threshold is output; if the corrective control quantity is greater than or equal to the upper limit threshold, the calculated corrective control quantity is output. After amplitude limiting processing, it is ensured that even under extreme working conditions such as full winding of the wire winding machine, even if the control algorithm calculates an extremely high correction control amount, the actual output will not exceed the limit of the correction actuator, thus preventing the motor from overheating and burning out.

[0056] 2. Rate of Change Limit: This limits the rate of change of the calculated corrective control quantity, ensuring that the change within a unit of time does not exceed a preset slope threshold. This slope threshold is calibrated through mechanical response testing. The specific steps are as follows: (1) Put the wire winding machine in the stop state, lock the main shaft, and leave only the correction actuator in the movable state; install a high-sensitivity industrial acceleration sensor at the physical center of gravity of the winding table, and at the same time, collect the actual position feedback signal of the motor in real time through the data monitoring port of the correction actuator; (2) Use the host computer to input a linear sweep frequency signal to the active disturbance rejection controller as the input command signal required for the test. Essentially, it is a sine wave position command whose frequency increases linearly with time. The frequency is set to 1Hz to 200Hz, the amplitude is set to 10% of the rated torque of the correction actuator, and the scanning period is set to 15 seconds. At the same time as inputting the above command, the waveform data collected by the acceleration sensor is recorded synchronously and named the end acceleration response signal. (3) Perform a fast Fourier transform on the input command signal and the end acceleration response signal to obtain the frequency response function, draw the Bode plot, find the peak frequency with the largest amplitude gain, and define it as the first-order natural resonant frequency of the winding stage. (4) Based on the principle of seismic redundancy in engineering, the safe cutoff frequency is set to 0.707 times the first-order natural resonance frequency, denoted as According to the formula for the maximum rate of change of a sine wave, the relationship for the slope threshold is: ,in, The slope threshold is used; it can be seen that when the output control signal is an amplitude value with an upper limit threshold... , frequency is When the signal is a sine wave, its slope reaches its mathematical maximum at the zero-crossing point, that is... As long as the rate of change of the corrective control quantity does not exceed So regardless of whether the corrective control amount is between 0 and... No matter how large the fluctuations are within the specified range, the equivalent frequency component cannot exceed the set safe cutoff frequency. This ensures that mechanical resonance will not occur under any extreme working conditions.

[0057] The corrective control quantity, after the above calculation and limiting processing, is input into the industrial control computer. Communication is carried out using the EtherCAT real-time industrial Ethernet bus. The corrective control quantity is encapsulated into a speed control command data packet conforming to the CIA402 standard and sent to the corrective actuator in real time with a communication cycle of 1ms. After receiving the speed control command, the internal current loop and speed loop of the web correction actuator work together to control the motor output corresponding speed and direction. The motor shaft is mechanically connected to the winding table of the wire winding machine through a precision ball screw pair. The rotational motion of the motor is converted into the axial linear movement of the winding table by the screw. The movement of the winding table causes the reel on it to undergo lateral displacement relative to the strip. This lateral displacement cancels out the real-time lateral deviation, ensuring that the strip is always aligned with the predetermined trajectory of the reel for winding, thus completing a complete automatic web correction closed loop from visual perception to algorithm calculation to mechanical execution.

[0058] Figure 2 This is a comparison diagram of the web correction response of the present invention and the prior art under full-roll conditions, provided by an embodiment of the present invention. As shown in the figure, the prior art exhibits obvious response lag due to insufficient driving force under full-roll conditions, which cannot meet the requirements of high-precision web correction. However, the present invention, by modifying the control parameters, quickly maintains consistency with the empty roll state after physical limiting, proving that the present invention can effectively overcome the huge inertial resistance brought by the roll diameter and achieve a high degree of consistency between the web correction response speed and the empty roll state throughout the entire roll diameter production cycle.

[0059] It should be noted that the specific description of signal conversion and physical drive in the embodiments of the present invention is only a preferred solution and should not be construed as a limitation on the scope of protection of the present invention. For example, the transmission method of the correction control quantity is not limited to using EtherCAT bus. Using other industrial fieldbus protocols, such as PROFINET IRT, CC-Link IE Field, CANopen, etc. for data transmission are all conventional variations of this technical solution. The mechanism by which the correction actuator converts rotary motion into linear motion is not limited to ball screw pairs. Using synchronous belt drive, gear and rack drive, crank and connecting rod mechanism, or directly using a linear motor to drive the winding table to move without intermediate transmission links are all equivalent implementations of the technical feature of driving the winding table to perform axial reciprocating movement.

Claims

1. An automatic deviation correction control method for a wire winding machine based on image edge features, characterized in that, include: The edge image data of the strip at the entrance of the wire winding machine, the real-time angular velocity of the reel, and the real-time linear velocity of the traction motor are collected and preprocessed to obtain the actual position of the edge of the strip. Based on the actual position and the preset reference position, the real-time lateral deviation of the strip is calculated. The real-time diameter of the reel is calculated based on the real-time angular velocity and the real-time linear velocity. Using the real-time roll diameter and combining the principles of rigid body dynamics, the inertia compensation factor is calculated, and the control parameters of the automatic disturbance rejection controller of the wire winding machine are corrected in real time using the inertia compensation factor. The real-time lateral deviation is input to the corrected active disturbance rejection controller, which calculates the correction control quantity and sends it to the correction actuator to drive the correction actuator to make relevant adjustments.

2. The automatic deviation correction control method for a wire winding machine based on image edge features according to claim 1, characterized in that, The value of the inertia compensation factor is non-linearly positively correlated with the ratio of the real-time roll diameter to the empty roll radius of the roll.

3. The automatic deviation correction control method for a wire winding machine based on image edge features according to claim 1, characterized in that, Real-time correction of the control parameters includes: obtaining the reference control parameters of the active disturbance rejection controller, dividing the reference control parameters by the inertia compensation factor to obtain the real-time corrected control parameters, thereby causing the active disturbance rejection controller to output a correspondingly larger correction control quantity when the inertia compensation factor increases.

4. The automatic deviation correction control method for a wire winding machine based on image edge features according to claim 1, characterized in that, The preprocessing includes: setting a contrast threshold, binarizing the edge image data of the strip to obtain a binarized image; extracting pixel gradient abrupt change points in the binarized image as edge feature points; and using the least squares method to perform linear fitting on the edge feature points to obtain the actual position of the edge of the strip.

5. The automatic deviation correction control method for a wire winding machine based on image edge features according to claim 4, characterized in that, The contrast threshold was calibrated through a light adaptability experiment.

6. The automatic deviation correction control method for a wire winding machine based on image edge features according to claim 1, characterized in that, The method further includes: performing a safety limiting process on the correction control quantity, wherein the safety limiting process includes amplitude limiting and rate of change limiting.

7. The automatic deviation correction control method for a wire winding machine based on image edge features according to claim 6, characterized in that, The amplitude limiting includes setting an upper limit threshold for the correction control quantity, which is based on the rated voltage of the correction actuator.

8. The automatic deviation correction control method for a wire winding machine based on image edge features according to claim 1, characterized in that, The correction mechanism is specifically a servo motor module that drives the winding table of the wire winding machine to move axially back and forth.

9. The automatic deviation correction control method for a wire winding machine based on image edge features according to claim 6, characterized in that, The rate of change limit includes: limiting the rate of change of the correction control quantity to not exceed a preset slope threshold, which is calibrated through a mechanical response test.

10. The automatic deviation correction control method for a wire winding machine based on image edge features according to claim 3, characterized in that, The reference control parameters include: controlling the wire winding machine to be in an empty winding state, inputting a step test signal to the active disturbance rejection controller, adjusting the parameters of the active disturbance rejection controller until the step response curve meets the preset standard, and recording the control parameters at this time as the reference control parameters.

Images