Machine vision-based intelligent control system and method for winding machine

Abstract

The application discloses a machine vision-based intelligent control system and control method for a winding machine, and relates to the technical field of industrial automation.The winding machine is equipped with a double-camera machine vision sensing cluster, which synchronously collects video frames of the drum area and the pay-off area of a cable drum, and dynamically adjusts the frame extraction frequency according to the take-up line speed to obtain a video frame sequence.Through image preprocessing, pixel projection segmentation and boundary pixel point analysis of the drum area video frames, the winding quality is determined by peak counting, and a cable arrangement adjustment requirement is generated.Meanwhile, the cable axis trajectory in the pay-off area video frames is analyzed, path abnormalities are identified, and a correction requirement is generated.After priority integration, a comprehensive control instruction is output: when the path is abnormal, the machine is stopped and an alarm is triggered;when the cable arrangement is abnormal, the gain coefficient is dynamically adjusted by an incremental PID controller, the compensation displacement is calculated, and the transverse movement speed of the cable arrangement device is controlled, so that the closed-loop intelligent control of the cable arrangement spacing is realized.

Description

Technical Field

[0001] This invention belongs to the field of industrial automation technology, specifically relating to a machine vision-based intelligent control system and control method for winding machines. Background Technology

[0002] In the field of cable winding and unwinding, the introduction of intelligent sensing and closed-loop control technologies to achieve real-time monitoring and dynamic control of winding quality and running trajectory has become an important development direction for improving production efficiency and ensuring process consistency.

[0003] The automated control and quality inspection of winding machines generally lack automatic feedback mechanisms and still rely heavily on manual parameter setting and process fine-tuning. When the cable type or diameter is changed, the parameters need to be manually reset in the system. If the cable position is offset or the winding is messy during the winding process, the equipment cannot identify and adjust it automatically, and manual shutdown is required. This open-loop control mode results in the equipment having acceptable operating efficiency, but the winding quality is far lower than that of manual methods because the machine cannot observe and judge the winding status in real time like an operator. Secondly, existing technical solutions often only focus on the winding state in the drum area, lacking synchronous monitoring and analysis of the cable's trajectory in the pay-off area. Single-view detection cannot comprehensively assess the overall movement state of the cable during the take-up process, and it is difficult to detect abnormalities in the pay-off end in a timely manner. While simple image classification or defect recognition methods can determine whether there are gaps or overlaps in the winding, they cannot directly link the detection results with the control commands of the actuator, and there is still a gap between detection and control.

[0004] To address the aforementioned problems, this invention proposes an intelligent control system and method for winding machines based on machine vision. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent control system and method for winding machines based on machine vision, solving the problems of disconnect between visual detection and control execution, single detection dimension, and poor environmental adaptability in existing technologies.

[0006] The objective of this invention can be achieved through the following technical solutions: A machine vision-based intelligent control method for winding machines, the method comprising: S1 receives the take-up command, controls the winding machine to perform the take-up operation, activates the machine vision sensor cluster equipped on the winding machine, and collects the first cable video in the winding area of ​​the cable reel and the second cable video in the unwinding area in real time, and performs frame division processing on each, locking the first cable video frame sequence and the second cable video frame sequence. S2, identify and separate the cable in the drum area in the first cable video frame sequence in real time, lock the winding quality, and generate corresponding cable adjustment requirements; S3, combine the second cable video frame sequence to analyze the actual trajectory of the cable and generate the corresponding path correction requirements; S4 integrates the priority of cabling adjustment and path correction requirements, generates comprehensive control instructions, and executes the adjustments.

[0007] As a further aspect of the present invention, in step S1, the machine vision sensor cluster equipped with the winding machine includes a first camera A1 and a second camera A2. The first camera A1 captures the winding area of ​​the cable reel from above; the second camera A2 captures the unwinding area of ​​the cable from the front, excluding the winding area.

[0008] As a further aspect of the present invention, the specific method for real-time acquisition of the first cable video in the reel area and the second cable video in the cable laying area in step S1 is as follows: Extract the first camera A1 and the second camera A2, and perform time calibration and sampling frequency synchronization, wherein the sampling frequency is a preset frequency; Determine the time when the receiving instruction is generated, denoted as t1; At time t1, the first camera A1 and the second camera A2 are activated simultaneously to capture images of the drum area and the cable laying area, respectively, and the first cable video R and the second cable video G captured by the first camera A1 and the second camera A2 are obtained, respectively.

[0009] As a further aspect of the present invention, the specific method for locking the first cable video frame sequence and the second cable video frame sequence in step S1 is as follows: At time t1, the rotary encoder pulse signal of the winding drive motor of the winding machine is acquired, the pulse frequency is analyzed, and the real-time winding speed of the cable is determined, denoted as V(t), where (t) represents the real-time time. Extract the preset speed threshold V max V min , where V max >V min >0; If V(t) ≥ V max Determine the frame decimation frequency F(t) = F max ; If V min ≤V(t)<V max Determine the frame decimation frequency F(t) = F base ; If V(t) < V min Determine the frame decimation frequency F(t) = Fmin ; Among them, F max F base F min All are preset frame extraction frequencies, and F max >F base >F min >0; Video frames are extracted in real time from the first cable video R and the second cable video G at equal time intervals Δt=1 / F(t), where 1 is 1 second; The extracted video frames are arranged in chronological order and denoted as the first cable video frame sequence r1, r2, ..., rj and the second cable video frame sequence g1, g2, ..., gj, where j is the total number of video frames, ri corresponds to gi, i is the counting index, and 1≤i≤j.

[0010] As a further aspect of the present invention, in step S2, the specific method for locking the winding quality and correspondingly generating the wiring adjustment requirements is as follows: Extract any one video frame ri from the first cable video frame sequence r1, r2, ..., rj and perform image preprocessing, including grayscale conversion, median filtering for noise reduction and adaptive contrast enhancement, denoised as the enhanced image ri'. By combining the known pixels on the outer surface of the cable, pixel projection segmentation is performed on the enhanced image ri' to determine the effective winding area of ​​the cable on the surface of the reel. Within the effective winding area, extract the boundary pixels of the upper surface of the cable in the enhanced image ri', and plot them in a two-dimensional coordinate system based on the current position relationship of each pixel to obtain the curve of the upper surface pixels of the cable. Determine the total number of peaks in the curve of the upper surface pixels of the cable, which is denoted as the total number of windings N in the enhanced image ri'. Determine the standard total number of windings N of the cable reel for the current cable outer diameter X. std ; If N=N std No action will be taken. If N < N std The system determines that the winding is sparse and generates a requirement for dense wiring. If N>N std The system determines that the winding is dense and generates a requirement for sparse wiring.

[0011] As a further aspect of the present invention, in step S3, the specific method for generating path correction requirements by analyzing the actual trajectory of the cable in conjunction with the second cable video frame sequence is as follows: Based on the method of determining the enhanced image ri' from the first cable video frame sequence r1,r2,...,rj, the enhanced image gi' is determined from the second cable video frame sequence g1,g2,...,gj; By combining the known outer surface pixels of the cable, pixel projection segmentation is performed on the enhanced image gi' to determine the effective cable area in the laying area; Combine the centerline Z of the cable with the effective cable area to map the actual trajectory of the cable. If the centerline Z intersects, the actual trajectory of the cable is determined to be abnormal and manual correction is required. Conversely, no action is taken.

[0012] As a further aspect of the present invention, the specific method for generating the integrated control command in step S4 is as follows: Extract the cabling adjustment requirement at the current time t, and denote it as requirement D1. Extract the path correction requirement at the current time t, and denote it as requirement D2. If demand D2 exists and is determined to require manual correction, assign the path correction demand the highest priority, generate a shutdown alarm command as a comprehensive control command, suspend the line take-up operation and trigger a manual intervention signal; If demand D2 does not exist and demand D1 exists, assign the highest priority to the cable adjustment demand, generate a cable lateral movement control command as a comprehensive control command, and adjust the lateral movement motor of the cable laying mechanism to adjust the cable arrangement spacing on the cable reel surface. If neither demand D1 nor demand D2 exists, a maintenance instruction is generated as a comprehensive control instruction and remains unchanged.

[0013] As a further aspect of the present invention, in step S4, the specific method by which the transverse motor of the cable-laying mechanism adjusts the spacing of the cables on the surface of the cable reel is as follows: Extract the cable outer diameter X, the effective winding width L of the effective winding area, and the total number of standard windings N. std ; Obtain the total number of windings N detected at the current moment, and calculate the deviation of the total number of windings ΔN=N. std -N; When ΔN≠0, calculate the total lateral displacement H that the cable tray needs to compensate for: H = ΔN × X; Collect the real-time take-up line speed V(t) and the current cable winding radius Q(t) of the drum, and determine the reference lateral speed of the cable guide V0=(V(t)×X) / (2π×Q(t)); Generate a lateral movement control command for the cable guide, instructing the lateral movement motor to drive the cable guide at the corrected lateral movement speed V. adj Run, where V adj =V0+Kp×H, where Kp is the preset position compensation gain coefficient.

[0014] As a further aspect of the present invention, the preset position compensation gain coefficient Kp is dynamically adjusted by an incremental PID controller. The input of the incremental PID controller is the total winding deviation ΔN, and the output is the gain coefficient adjustment amount ΔKp. Kp is corrected in real time to suppress the oscillation of the wire spacing.

[0015] A machine vision-based intelligent control system for a winding machine, the system comprising: The multi-source vision synchronous acquisition end receives the take-up command, controls the winding machine to perform the take-up operation, and activates the machine vision sensor cluster equipped on the winding machine to acquire the first cable video in the winding area of ​​the cable reel and the second cable video in the unwinding area in real time, and performs frame division processing on each, locking the first cable video frame sequence and the second cable video frame sequence. The winding quality assessment end identifies and separates the cable within the drum area in the first cable video frame sequence in real time, locks the winding quality, and generates corresponding cable adjustment requirements. The cable trajectory analysis end combines the second cable video frame sequence to analyze the actual trajectory of the cable and generate the corresponding path correction requirements. The control command decision-making end integrates the priority of routing adjustment needs and path correction needs to generate comprehensive control commands for execution and regulation.

[0016] The beneficial effects of this invention are: This invention uses machine vision technology to deploy dual cameras for visual closed-loop control, simultaneously analyzing winding quality and trajectory and generating comprehensive instructions. It solves the problem of lag and misjudgment in traditional single feedback sources, can actively adapt to dynamic changes such as cable offset and accumulation, reduce manual intervention, ensure tight and neat winding, reduce damage risk, improve automation level and product consistency, and is especially suitable for high-speed continuous winding scenarios. This invention ensures the temporal consistency of video data from dual cameras through time calibration and synchronous sampling, laying the foundation for accurate comparative analysis. Its core advantage lies in dynamically adjusting the frame extraction frequency according to the cable reeling speed. At high speeds, it captures cable movement details with a high frame rate to avoid missing key information, while at low speeds, it reduces the frame rate to reduce data redundancy and storage burden. This significantly optimizes computing resources and energy consumption while ensuring monitoring accuracy, enhancing the system's adaptability to changes in operating conditions, and combining high efficiency with economy. This invention quantifies winding by comparing peak counts, providing clear numerical basis for winding adjustment, completely eliminating the subjectivity of manual visual inspection and improving process consistency. Simultaneously, it employs centerline cross-detection of trajectory anomalies, which is simpler and more reliable than traditional edge tracking, effectively preventing wire jamming damage. A shared preprocessing strategy across two stages ensures feature uniformity and reduces computational load. Anomaly detection is directly mapped to dense, sparse, or manual intervention commands, forming a closed loop from visual perception to execution correction, enhancing the system's real-time response to winding status and its intelligent self-adjustment level. This invention prioritizes safety through a priority shutdown mechanism, triggering timely manual intervention to prevent equipment damage in case of path abnormalities. Simultaneously, it achieves closed-loop precise control of cable spacing based on line speed, coil diameter, and deviation compensation, ensuring uniform and compact cable arrangement and improving coil quality. Furthermore, it utilizes incremental PID real-time gain correction to effectively suppress spacing oscillations, enhancing the system's adaptability to different operating conditions, thereby reducing the frequency of manual calibration and improving production stability and overall efficiency. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart illustrating the method described in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] like Figure 1 As shown, this application provides a machine vision-based intelligent control system and control method for a winding machine; As an embodiment 1 of this application, it specifically includes: S1 receives the take-up command, controls the winding machine to perform the take-up operation, activates the machine vision sensor cluster equipped on the winding machine, and collects the first cable video in the winding area of ​​the cable reel and the second cable video in the unwinding area in real time, and performs frame division processing on each, locking the first cable video frame sequence and the second cable video frame sequence. S2, identify and separate the cable in the drum area in the first cable video frame sequence in real time, lock the winding quality, and generate corresponding cable adjustment requirements; S3, combine the second cable video frame sequence to analyze the actual trajectory of the cable and generate the corresponding path correction requirements; S4 integrates the priority of cabling adjustment and path correction requirements, generates comprehensive control instructions, and executes the adjustments.

[0021] Example 2

[0022] This embodiment is based on Embodiment 1 and provides a machine vision-based intelligent control method for winding machines, applied to the winding and unwinding process of linear materials such as power cables and communication cables. Specifically, it includes the following: The winding machine is equipped with a machine vision sensor cluster and a controller. The controller receives the winding command issued by the host computer (PLC) and executes the following steps: In step S1, a take-up command is received, the cable reel is controlled to perform the take-up operation, and the machine vision sensor cluster equipped on the cable reel is activated to collect the first cable video in the reel area and the second cable video in the unwinding area in real time. Frame segmentation processing is performed on each, and the first cable video frame sequence and the second cable video frame sequence are locked. Specifically: The machine vision sensor cluster includes a first camera A1 and a second camera A2. The first camera A1 is an industrial complementary metal oxide semiconductor camera with a resolution of 1920×1080. It is installed directly above the cable reel, with the lens optical axis perpendicular to the reel axis. The captured image covers the entire reel area and looks down on the reel area.

[0023] The second camera, A2, is a CMOS camera of the same model. It is mounted in front of the wire-laying frame, with its optical axis parallel to the ground and aligned with the wire-laying area. It looks directly at the wire-laying area and the captured image does not include the roll area.

[0024] At time t1 when the line receiving command is generated, time calibration and sampling frequency synchronization are performed on the first camera A1 and the second camera A2. The time calibration is achieved through the IEEE 1588 Precision Time Protocol, and the sampling frequency is preset to 30Hz.

[0025] At time t1, the first camera A1 and the second camera A2 are activated simultaneously to continuously capture images of the drum area and the cable unwinding area, respectively, to obtain the first cable video R and the second cable video G.

[0026] The specific method for locking the first cable video frame sequence and the second cable video frame sequence is as follows: At time t1, the rotary encoder pulse signal of the winding drive motor of the winding machine is acquired. The rotary encoder is a 1024-line incremental photoelectric encoder, installed at the tail end of the winding motor shaft. The controller analyzes the pulse frequency using the formula V(t)=(π×D×f pulse The real-time take-up linear speed V(t) of the cable is determined by f / (P×c), where D is the current winding diameter of the drum, and f is the current winding speed of the cable. pulsedenoted as pulse frequency, P as the number of pulses per encoder revolution, c as the reduction ratio, and (t) as the real-time time.

[0027] Extract the preset speed threshold V max =2.0m / s, V min =0.2m / s, where, V max >V min >0.

[0028] If V(t) ≥ V max Determine the frame decimation frequency F(t) = F max F max =30Hz; If V min ≤V(t)<V max Determine the frame decimation frequency F(t) = F base F base =15Hz; If V(t) < V min Determine the frame decimation frequency F(t) = F min F min =5Hz.

[0029] F max F base F min All of these are frame extraction frequencies preset by the operators based on actual needs, and F max >F base >F min >0.

[0030] Video frames are extracted in real time from the first cable video R and the second cable video G at equal time intervals Δt = 1 second / F(t) using the determined frame extraction frequency F(t).

[0031] The extracted video frames are arranged in chronological order and denoted as the first cable video frame sequence r1, r2, ..., rj and the second cable video frame sequence g1, g2, ..., gj, where j is the total number of video frames, and ri and gi are a pair of video frames at the same sampling time, i is the counting index, and 1≤i≤j.

[0032] Next, in step S2, the cable within the reel area in the first cable video frame sequence is identified and separated in real time, the winding quality is locked, and corresponding cable adjustment requirements are generated, specifically: Extract any one video frame ri from the first cable video frame sequence r1, r2, ..., rj and perform image preprocessing, wherein the image preprocessing includes: Grayscale conversion: The RGB color space is converted to a grayscale image using a weighted average method, with grayscale value Gray = 0.299R + 0.587G + 0.114B; Median filtering noise reduction: Median filtering is performed using a 3×3 square template to remove salt-and-pepper noise; Adaptive contrast enhancement: A limited contrast adaptive histogram equalization algorithm is used to enhance the contrast between the cable and the background.

[0033] The preprocessed image is denoted as the enhanced image ri'.

[0034] Based on the known outer surface pixels of the cable, pixel projection segmentation is performed on the enhanced image ri'. The known outer surface pixels of the cable are obtained by collecting cable samples offline, extracting color histograms, and establishing segmentation threshold ranges. In this embodiment, the cable is a black polyvinyl chloride sheath, and the pixel grayscale threshold range is [0, 45].

[0035] The enhanced image ri' is binarized, and pixels with values ​​within the threshold range are set as foreground pixels, while the rest are set as background pixels. Then, the effective winding area of ​​the cable on the reel surface is segmented by a combination of horizontal and vertical projection. This area is a rectangular envelope containing the continuous cable winding.

[0036] Within the effective winding area, the upper surface boundary pixels of the cable in the enhanced image ri' are extracted. The Canny edge detection operator is used to extract the edges, and the continuous edge point set located at the top layer is selected by using the cable stacking relationship, which is the upper surface boundary pixel.

[0037] Each pixel is plotted in a two-dimensional coordinate system based on its current position relationship. The horizontal axis represents the position along the axial direction of the drum, and the vertical axis represents the radial height. This yields the pixel curve on the upper surface of the cable. The peaks are determined by finding the local maxima of the curves, and the total number of peaks is the total number of windings N in the enhanced image ri'.

[0038] Determine the standard total number of windings N of the cable reel for the current cable outer diameter X. std In this embodiment, the cable outer diameter X is known to be 12mm, and the axial effective winding width L of the effective winding area is 600mm, then N std =floor(L / X)=50.

[0039] If N=N std No action will be taken. If N < N std The system determines that the winding is sparse and generates a requirement for dense wiring. If N>N std The system determines that the winding is dense and generates a requirement for sparse wiring.

[0040] Then, in step S3, the actual trajectory of the cable is analyzed by combining the second cable video frame sequence, and the corresponding path correction requirements are generated, specifically including: Based on the same preprocessing method used to determine the enhanced image ri' from the first cable video frame sequence, the enhanced image gi' is determined from the second cable video frame sequence g1, g2, ..., gj. Then, combined with the known outer surface pixels of the cable, the enhanced image gi' is subjected to the same pixel projection segmentation as S2 to determine the effective cable area of ​​the cable in the laying area.

[0041] The centerline Z of the cable is extracted by combining the effective cable area. The centerline Z is then used to extract the single-pixel skeleton of the cable area through an iterative thinning algorithm and remove burrs to obtain the centerline representing the actual trajectory of the cable.

[0042] If the centerline Z intersects, that is, if the centerline has a self-intersection point, it is determined that the actual trajectory of the cable is abnormal, and there is a risk of knotting or tangling, which requires manual correction. Conversely, no action is taken.

[0043] Finally, in step S4, the cabling adjustment requirements and path correction requirements are prioritized and integrated to generate a comprehensive control command, which is then executed, as follows: First, extract the cabling adjustment requirement at the current time t, and denote it as requirement D1; Extract the path correction requirement at the current time t, and denote it as requirement D2.

[0044] If requirement D2 exists and is determined to require manual correction, assign the path correction requirement the highest priority and generate a shutdown alarm command as a comprehensive control command.

[0045] The shutdown alarm command outputs a digital signal to the relay through the controller to disconnect the power supply to the take-up motor. At the same time, it sends alarm information to the monitoring station via Ethernet and triggers the audible and visual alarm, suspends the take-up operation, and triggers a manual intervention signal.

[0046] If demand D2 does not exist and demand D1 exists, assign the highest priority to the cable adjustment demand, generate a cable traverse control command as a comprehensive control command, and adjust the cable traverse motor of the cable traverse mechanism to adjust the cable arrangement spacing on the cable reel surface.

[0047] If neither demand D1 nor demand D2 exists, a maintenance instruction is generated as a comprehensive control instruction and remains unchanged.

[0048] The specific method for adjusting the spacing of the cables on the cable reel surface using the transverse motor of the cable laying mechanism is as follows: Extract the cable outer diameter X=12mm, the effective winding width L=600mm of the effective winding area, and the total number of standard windings N. std =50.

[0049] Obtain the total number of windings N detected at the current moment, and calculate the deviation of the total number of windings ΔN=N. std-N.

[0050] When ΔN≠0, calculate the total lateral displacement H=ΔN×X that the cable winding device needs to compensate. At the same time, collect the real-time take-up line speed V(t) and the current cable winding radius Q(t) of the drum.

[0051] Q(t) is calculated from the number of winding layers obtained by converting the cumulative pulse count of the encoder and the outer diameter of the cable. The reference lateral speed of the cable guide is determined as V0 = (V(t) × X) / (2π × Q(t)).

[0052] Generate a lateral movement control command for the cable guide, instructing the lateral movement motor to drive the cable guide at the corrected lateral movement speed V. adj Run, where V adj =V0+Kp×H, where Kp is the preset position compensation gain coefficient, and in this embodiment, a fixed value Kp=0.5 is taken.

[0053] The lateral movement control command for the cable guide is transmitted via a pulse direction signal output from the controller to the stepper motor driver, which drives the lateral stepper motor to move the cable guide left and right.

[0054] As described above, this embodiment reduces computational load by adaptive frame rate extraction, and combines simultaneous acquisition by dual cameras with winding quality assessment and trajectory anomaly detection to achieve real-time control of wire density and path anomalies during the take-up process, effectively reducing wire tangling and pressing defects and improving take-up quality.

[0055] Example 3

[0056] This embodiment is based on embodiment 2, but differs in that it further optimizes the method of obtaining the position compensation gain coefficient Kp in the lateral movement speed control of the cable. The key point is that the preset position compensation gain coefficient Kp in S4 is dynamically adjusted by an incremental proportional-integral-derivative controller, as detailed below: First, an incremental PID controller is constructed, with the input being the total winding deviation ΔN, the setpoint being 0, and the output being the gain coefficient adjustment ΔKp. The controller's discretization period is synchronized with the frame extraction period. The PID control algorithm is as follows: ΔKp(k)=Kp_coe×(e(k)-e(k-1))+Ki_coe×e(k)+Kd_coe×(e(k)-2e(k-1)+e(k-2)); Where e(k)=ΔN(k) is the total number of winding deviations at the current sampling time, and Kp_coe=0.08, Ki_coe=0.01, Kd_coe=0.02 are PID parameters, which were adjusted by the operator through experiments.

[0057] The real-time correction Kp = Kp_base + ΔKp, where Kp_base is 0.5 as the base gain value.

[0058] Substitute the corrected Kp into V adj =V0+Kp×H, generates the speed command for the transverse motor.

[0059] When the cabling is sparse or dense, ΔN fluctuates. The incremental PID controller dynamically adjusts Kp to accelerate deviation convergence while suppressing overshoot and preventing periodic oscillations in the cabling spacing.

[0060] When ΔN approaches zero, ΔKp approaches zero, Kp returns to the base gain, and the cable tray maintains the reference lateral speed.

[0061] This embodiment introduces incremental PID dynamic adjustment of position compensation gain, which optimizes the lateral movement speed correction in real time according to the winding deviation, effectively suppressing the oscillation of the wire spacing. Especially in the process of high wire speed and variable diameter winding, it can maintain the uniformity of wire arrangement and further reduce the risk of cable surface damage and misalignment.

[0062] Example 4

[0063] This embodiment provides a machine vision-based intelligent control system for winding machines, used to execute the aforementioned embodiments. It includes a multi-source vision synchronous acquisition terminal, a winding quality evaluation terminal, a pay-off trajectory analysis terminal, and a control command decision terminal, as detailed below: The multi-source visual synchronous acquisition terminal includes a first camera A1, a second camera A2, an image acquisition card, and a synchronization controller; Both the first camera A1 and the second camera A2 are global shutter CMOS industrial cameras with an image sensor target surface size of 1 / 1.8 inches and a pixel size of 4.8μm×4.8μm. They are connected to the image acquisition card via a gigabit Ethernet interface.

[0064] The synchronization controller sends hardware trigger signals to the two cameras based on the IEEE 1588PTP protocol to achieve microsecond-level synchronized exposure. The image acquisition card is a four-channel interface card that transmits the acquired raw video stream in parallel to the winding quality evaluation end and the laying trajectory analysis end.

[0065] The multi-source vision synchronous acquisition terminal also obtains the cable take-up command and rotary encoder pulse signal from the PLC, completes the frame division processing, and outputs the first cable video frame sequence and the second cable video frame sequence.

[0066] The winding quality assessment end and the laying trajectory analysis end are deployed on the same edge computing device, which is equipped with a graphics processor and runs a Linux real-time operating system.

[0067] The winding quality assessment end receives the first cable video frame sequence and performs image preprocessing, pixel projection segmentation, upper surface boundary extraction and winding total calculation according to the method described in Example 2 to generate the cabling adjustment requirement D1.

[0068] The cable laying trajectory analysis end receives the second cable video frame sequence, performs centerline extraction and intersection judgment, and generates path correction requirement D2.

[0069] It should be noted that the two ends of generating cable adjustment requirement D1 and generating path correction requirement D2 are processed in parallel, and the processing results are updated to the status table through shared memory at a refresh rate of 1kHz.

[0070] The control command decision-making end is part of the PLC control system. It communicates with the edge computing device through the EtherCAT industrial Ethernet bus and periodically reads the D1 and D2 states.

[0071] The control command decision terminal has embedded priority arbitration logic: when D2 exists, an emergency stop command is immediately sent to the frequency converter via the bus, and the alarm output is set at the same time; When only D1 exists, calculate the traverse motor speed command and send it to the cable servo driver via EtherCAT to drive the traverse servo motor to adjust the position of the cable tray. When there is no need, send a no-operation command.

[0072] All data in the formulas described above have been calculated with dimensions removed. Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.

[0073] The above description is merely an example and illustration of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the invention or exceed the scope defined in the claims, they should all fall within the protection scope of the present invention.

[0074] It should be stated that all user data collected in this application was collected with the user's consent and authorization. Furthermore, the uses of user data are legal and compliant, and the use and processing of user data comply with the relevant laws, regulations, and standards of the relevant regions.

Claims

1. A machine vision-based intelligent control method for winding machines, characterized in that, The method includes: S1 receives the take-up command, controls the winding machine to perform the take-up operation, activates the machine vision sensor cluster equipped on the winding machine, and collects the first cable video in the winding area of ​​the cable reel and the second cable video in the unwinding area in real time, and performs frame division processing on each, locking the first cable video frame sequence and the second cable video frame sequence. S2, identify and separate the cable in the drum area in the first cable video frame sequence in real time, lock the winding quality, and generate corresponding cable adjustment requirements; S3, combine the second cable video frame sequence to analyze the actual trajectory of the cable and generate the corresponding path correction requirements; S4 integrates the priority of cabling adjustment and path correction requirements, generates comprehensive control instructions, and executes the adjustments.

2. The method according to claim 1, characterized in that, In step S1, the machine vision sensor cluster equipped with the cable reel includes a first camera A1 and a second camera A2. The first camera A1 captures the cable reel's winding area from above; the second camera A2 captures the cable unwinding area from the front, excluding the reel area.

3. The method according to claim 1, characterized in that, In step S1, the specific method for real-time acquisition of the first cable video in the cable reel area and the second cable video in the cable laying area is as follows: Extract the first camera A1 and the second camera A2, and perform time calibration and sampling frequency synchronization, wherein the sampling frequency is a preset frequency; The time when the receiving instruction is generated is determined and denoted as t1; At time t1, the first camera A1 and the second camera A2 are activated simultaneously to capture images of the drum area and the cable laying area, respectively, and the first cable video R and the second cable video G captured by the first camera A1 and the second camera A2 are obtained, respectively.

4. The method according to claim 3, characterized in that, In step S1, the specific method for locking the first cable video frame sequence and the second cable video frame sequence is as follows: At time t1, the rotary encoder pulse signal of the winding drive motor of the winding machine is acquired, the pulse frequency is analyzed, and the real-time winding speed of the cable is determined, denoted as V(t), where (t) represents the real-time time. Extract the preset speed threshold V max V min , where V max >V min >0; If V(t) ≥ V max Determine the frame decimation frequency F(t) = F max ; If V min ≤V(t)<V max Determine the frame decimation frequency F(t) = F base ; If V(t) < V min Determine the frame decimation frequency F(t) = F min ; Among them, F max F base F min All are preset frame extraction frequencies, and F max >F base >F min >0; Video frames are extracted in real time from the first cable video R and the second cable video G at equal time intervals Δt=1 / F(t), where 1 is 1 second; The extracted video frames are arranged in chronological order and denoted as the first cable video frame sequence r1, r2, ..., rj and the second cable video frame sequence g1, g2, ..., gj, where j is the total number of video frames, ri corresponds to gi, i is the counting index, and 1≤i≤j.

5. The method according to claim 4, characterized in that, In step S2, the specific method for locking the winding quality and generating the corresponding wiring adjustment requirements is as follows: Extract any one video frame ri from the first cable video frame sequence r1, r2, ..., rj and perform image preprocessing, including grayscale conversion, median filtering for noise reduction and adaptive contrast enhancement, denoised as the enhanced image ri'. By combining the known pixels on the outer surface of the cable, pixel projection segmentation is performed on the enhanced image ri' to determine the effective winding area of ​​the cable on the surface of the reel. Within the effective winding area, extract the boundary pixels of the upper surface of the cable in the enhanced image ri', and plot them in a two-dimensional coordinate system based on the current position relationship of each pixel to obtain the curve of the upper surface pixels of the cable. Determine the total number of peaks in the curve of the upper surface pixels of the cable, which is denoted as the total number of windings N in the enhanced image ri'. Determine the standard total number of windings N of the cable reel for the current cable outer diameter X. std ; If N=N std No action will be taken. If N < N std The system determines that the winding is sparse and generates a requirement for dense wiring. If N>N std The system determines that the winding is dense and generates a requirement for sparse wiring.

6. The method according to claim 4, characterized in that, In step S3, the specific method for generating path correction requirements by analyzing the actual trajectory of the cable in conjunction with the second cable video frame sequence is as follows: Based on the method of determining the enhanced image ri' from the first cable video frame sequence r1,r2,...,rj, the enhanced image gi' is determined from the second cable video frame sequence g1,g2,...,gj; By combining the known outer surface pixels of the cable, pixel projection segmentation is performed on the enhanced image gi' to determine the effective cable area in the laying area; Combine the centerline Z of the cable with the effective cable area to map the actual trajectory of the cable. If the centerline Z intersects, the actual trajectory of the cable is determined to be abnormal and manual correction is required. Conversely, no action is taken.

7. The method according to claim 6, characterized in that, In step S4, the specific method for generating the integrated control command is as follows: Extract the cabling adjustment requirement at the current time t, and denote it as requirement D1. Extract the path correction requirement at the current time t, and denote it as requirement D2. If demand D2 exists and is determined to require manual correction, assign the path correction demand the highest priority, generate a shutdown alarm command as a comprehensive control command, suspend the line take-up operation and trigger a manual intervention signal; If demand D2 does not exist and demand D1 exists, assign the highest priority to the cable adjustment demand, generate a cable lateral movement control command as a comprehensive control command, and adjust the lateral movement motor of the cable laying mechanism to adjust the cable arrangement spacing on the cable reel surface. If neither demand D1 nor demand D2 exists, a maintenance instruction is generated as a comprehensive control instruction and remains unchanged.

8. The method according to claim 7, characterized in that, In step S4, the specific method for adjusting the spacing of the cables on the cable reel surface using the transverse motor of the cable laying mechanism is as follows: Extract the cable outer diameter X, the effective winding width L of the effective winding area, and the total number of standard windings N. std ; Obtain the total number of windings N detected at the current moment, and calculate the deviation of the total number of windings ΔN=N. std -N; When ΔN≠0, calculate the total lateral displacement H that the cable tray needs to compensate for: H = ΔN × X; Collect the real-time take-up line speed V(t) and the current cable winding radius Q(t) of the drum, and determine the reference lateral speed of the cable guide V0=(V(t)×X) / (2π×Q(t)); Generate a lateral movement control command for the cable guide, instructing the lateral movement motor to drive the cable guide at the corrected lateral movement speed V. adj Run, where V adj =V0+Kp×H, where Kp is the preset position compensation gain coefficient.

9. The method according to claim 8, characterized in that, The preset position compensation gain coefficient Kp is dynamically adjusted by an incremental PID controller. The input of the incremental PID controller is the total winding deviation ΔN, and the output is the gain coefficient adjustment amount ΔKp. Kp is corrected in real time to suppress the oscillation of the wire spacing.

10. A machine vision-based intelligent control system for a winding machine, characterized in that, The system includes: The multi-source vision synchronous acquisition end receives the take-up command, controls the winding machine to perform the take-up operation, and activates the machine vision sensor cluster equipped on the winding machine to acquire the first cable video in the winding area of ​​the cable reel and the second cable video in the unwinding area in real time, and performs frame division processing on each, locking the first cable video frame sequence and the second cable video frame sequence. The winding quality assessment end identifies and separates the cable within the drum area in the first cable video frame sequence in real time, locks the winding quality, and generates corresponding cable adjustment requirements. The cable trajectory analysis end combines the second cable video frame sequence to analyze the actual trajectory of the cable and generate the corresponding path correction requirements. The control command decision-making end integrates the priority of routing adjustment needs and path correction needs to generate comprehensive control commands for execution and regulation.

Images