Positioning control method based on numerical control cooperative control

Abstract

The invention discloses a positioning control method based on numerical control cooperative control, and particularly relates to the technical field of numerical control cooperation, comprising the steps of triggering a pulse light source by an encoder and synchronously acquiring images by a global exposure camera, extracting feature points by using a neural network, and calculating a pattern pose in combination with a PnP algorithm or affine transformation; calculating a multi-dimensional position deviation by comparing a standard pose with a real-time pose, and quantifying a color register deviation in a CIELAB color space; a multivariable decoupling program control algorithm and a fuzzy PID strategy are adopted to generate a cooperative control instruction of a registration motor and a color adjusting valve; finally, continuous adjustment is achieved through a high-dynamic closed-loop program control system, and disturbance resistance and out-of-tolerance early warning capacity are achieved. The pattern positioning and color register precision is improved, and the method is suitable for the high-speed numerical control production environment.

Description

Technical Field

[0001] This invention relates to the field of program control and digital control technology, and more specifically, to a positioning control method based on numerical control collaborative control. Background Technology

[0002] Existing industrial CNC systems mostly adopt a "unit-independent control mode." For example, in a printing production line, actuators such as pattern matching motors, color adjustment valves, and conveyor rollers are driven by independent controllers. Each unit adjusts based solely on local sensor data, lacking a unified CNC collaborative framework. Information silos: The position compensation signal of the matching motor cannot be synchronized to the color adjustment valve, resulting in a coordination deviation of "color density mismatch after position correction"; Action delay: The control cycles of each unit are inconsistent (e.g., motor adjustment cycle is 20ms, valve adjustment cycle is 50ms), resulting in a lag in the overall line response, which cannot adapt to high-speed production. No global optimization: It is impossible to dynamically allocate the adjustment priority of each actuator based on the overall production rhythm (such as fabric transport speed, number of printing heads), resulting in a conflict between local optimization and overall accuracy.

[0003] There is an urgent need for a CNC collaborative control scheme based on unified sensing input to achieve synchronized action of multiple actuators, information exchange, and overall line accuracy control. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a positioning control method based on numerical control collaborative control, which solves the problems mentioned in the background art through the following scheme.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a positioning control method based on numerical control collaborative control, comprising: S1: Image Acquisition and Transmission: The displacement of the target material is monitored in real time by a displacement sensor. When the displacement reaches the preset sampling interval, a trigger command is generated. This trigger command synchronously triggers the pulse light source and the global exposure camera to acquire a clear real-time image of the moving material surface. S2: Pattern feature recognition and localization calculation: The real-time image is processed, and the processed image is used to identify multiple feature points of the pattern through a pre-trained feature extraction neural network model. Based on the spatial relationship of the feature points, the current actual pose of the pattern is calculated using the PnP algorithm or affine transformation. S3: Multi-dimensional error fusion calculation: The current actual pose is compared with the pre-stored standard pattern pose to calculate the multi-dimensional positional deviation of the pattern; at the same time, the pixel set of the color block area of ​​the pattern in the real-time image is extracted and converted to the CIELAB color space. The color difference between the real-time image and the standard color reference is calculated to quantify the color registration deviation value. S4: Intelligent control command generation: Based on the multi-dimensional position deviation and the color matching deviation value, a multi-variable decoupling program control algorithm is used to generate a collaborative control command for synchronously adjusting multiple position actuators and color adjustment mechanisms along the entire line; S5: Closed-loop feedback control: The processing is adjusted based on the cooperative control command. Then, S1 is executed again, and S2 to S5 are executed in sequence, thereby forming a real-time dynamic closed-loop feedback control loop integrated into the overall factory control system.

[0006] Preferably, the displacement sensor includes an encoder and is configured to generate a synchronization trigger signal when the displacement of the target material reaches a preset sampling interval, so as to synchronously control the actions of the pulse light source and the global exposure camera.

[0007] Preferably, the real-time image processing includes: smoothing the real-time image using a Gaussian filter; and enhancing the contrast between the pattern and the background, and between different color regions within the pattern, using a contrast-limited adaptive histogram equalization algorithm.

[0008] Preferably, the current actual pose includes: When using the PnP algorithm, the three-dimensional point set of the standard pattern model, the corresponding two-dimensional point set of the image, and the pre-calibrated camera intrinsic parameter matrix are used as inputs. The EPnP algorithm is used to solve for the rotation vector and translation vector, and the rotation vector is converted into Euler angles to obtain the current actual pose. When using affine transformation, the transformation matrix is ​​solved using a direct linear transformation algorithm or the RANSAC algorithm, and the translation and rotation angles of the pattern in the two-dimensional plane are decomposed from the transformation matrix.

[0009] Preferably, the quantization of color registration deviation includes: calculating the Euclidean distance between the mean value of the pixel set of the color patch region in the CIELAB color space and the LAB value of the standard color reference to obtain the color difference. This serves as the color deviation value.

[0010] Preferably, the multivariable decoupling program control algorithm includes: constructing a decoupling matrix to eliminate the coupling effect between different motion dimensions of the position actuator; the cooperative control command includes a pulse control signal for driving the position actuator to perform translation compensation, and an analog or digital control signal for adjusting the output of the color adjustment mechanism.

[0011] Preferably, the position actuator includes: a motor for achieving lateral and longitudinal translation compensation; the color adjustment mechanism includes a regulating valve for adjusting the colorant flow rate; the coordinated control command consists of a motor drive signal and a valve adjustment signal generated based on the deviation calculation result of the same control cycle, so as to achieve coordinated adjustment of pattern positioning and color density.

[0012] Preferably, the real-time dynamic closed-loop feedback control system includes: continuous operation with a fixed sampling period; and within each control period, predicting the deviation range of the next sampling moment based on the equipment response model; if the actual sampling deviation deviates from the predicted value by more than a preset threshold or the deviation continues to increase, triggering an out-of-tolerance alarm mechanism.

[0013] Preferably, the closed-loop feedback control includes: employing a composite control strategy combining feedforward control and feedback control to suppress external disturbances; and determining that the control process is complete when all deviations are less than their respective preset tolerance thresholds within multiple consecutive control cycles.

[0014] The technical effects and advantages of this invention are as follows: 1. A highly integrated intelligent control framework was constructed: machine vision, multi-dimensional information fusion and modern control theory were deeply combined to form a general software-defined control scheme, which improved the intelligence level of the system; 2. Improved the perception accuracy and robustness of the control system: Through synchronous trigger imaging technology and advanced image preprocessing algorithms, the accuracy and reliability of input information are ensured, laying the foundation for high-quality control; 3. Multivariable collaborative and decoupled control is achieved: The multivariable decoupled program control algorithm effectively eliminates the coupling interference between different control loops and degrees of freedom within the system, improving the accuracy and stability of control; 4. A highly dynamic and strong disturbance-resistant closed-loop mechanism has been established: Through a fixed short-cycle closed-loop feedback and a feedforward-feedback composite control strategy, the system has the ability to quickly respond to dynamic deviations and suppress external disturbances, ensuring the stability and reliability of the control process.

[0015] 5. Achieve whole-line CNC collaboration: Establish a unified collaborative control center to eliminate information silos among various actuators. Through deviation transmission modeling and decoupling algorithms, the coupling interference rate of multi-mechanism adjustment is reduced. Support collaborative replacement when actuators are abnormal. Adapt to flexible production: The collaborative control center can be quickly adapted to different production lines through parameter configuration. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall structure of the present invention.

[0017] Figure 2 This is a schematic diagram of the image acquisition synchronization triggering mechanism of the present invention.

[0018] Figure 3 This is a schematic diagram of the high dynamic closed-loop feedback and fault-tolerant mechanism structure of the present 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] refer to Figures 1-3 The positioning control method based on CNC collaborative control shown includes: S1: Image Acquisition and Transmission: The displacement of the target material is monitored in real time by a displacement sensor. When the displacement reaches the preset sampling interval, a trigger command is generated. This trigger command synchronously triggers the pulse light source and the global exposure camera to acquire a clear real-time image of the moving material surface. S2: Pattern feature recognition and localization calculation: The real-time image is processed, and the processed image is used to identify multiple feature points of the pattern through a pre-trained feature extraction neural network model. Based on the spatial relationship of the feature points, the current actual pose of the pattern is calculated using the PnP algorithm or affine transformation. S3: Multi-dimensional error fusion calculation: The current actual pose is compared with the pre-stored standard pattern pose to calculate the multi-dimensional positional deviation of the pattern; at the same time, the pixel set of the color block area of ​​the pattern in the real-time image is extracted and converted to the CIELAB color space. The color difference between the real-time image and the standard color reference is calculated to quantify the color registration deviation value. S4: Intelligent control command generation: Based on the multi-dimensional position deviation and the color matching deviation value, a multi-variable decoupling program control algorithm is used to generate a collaborative control command for synchronously adjusting multiple position actuators and color adjustment mechanisms along the entire line; S5: Closed-loop feedback control: The processing is adjusted based on the cooperative control command. Then, S1 is executed again, and S2 to S5 are executed in sequence, thereby forming a real-time dynamic closed-loop feedback control loop integrated into the overall factory control system.

[0021] S1 is the starting point of the entire closed-loop control, and its synchronization accuracy directly determines the accuracy of subsequent control decisions. The encoder, as a position feedback sensor, ensures strict synchronization between image acquisition and material displacement through its trigger pulses, which is the foundation for achieving high dynamic control.

[0022] The fabric moves continuously under the drive of the printing equipment, and an encoder that rotates synchronously monitors the fabric displacement in real time. When the displacement reaches a preset sampling interval, the encoder immediately generates a trigger pulse signal; this trigger pulse signal is simultaneously sent to the industrial camera and the LED light source controller.

[0023] Upon receiving the pulse, the camera immediately initiates a global exposure, beginning the acquisition of one frame. Simultaneously, the LED light source controller receives the pulse and immediately drives the light source to emit a high-brightness pulse of light lasting 0.1ms. Within the extremely short exposure time of the camera, the pulsed light source illuminates the moving fabric. At this moment, the light signal sensed by the camera sensor is the clear image of the fabric pattern at that instant, without any motion blur.

[0024] After the exposure is complete, the camera transmits the image data captured by the sensor to the memory of the industrial computer in real time through a high-speed data interface; the image acquisition card driver in the industrial computer receives and parses the data stream, stores a complete frame of image in the designated memory buffer, and sets the image ready flag; then the real-time image acquisition is complete.

[0025] It should be further explained that the preset sampling interval is 10 mm. In high-speed printing, if the sampling interval is too large (50 mm), details of local registration deviations may be missed, resulting in a lag in control response. If the interval is too small (1 mm), redundant images will be generated, exceeding the system's processing capacity and causing data congestion and delay. The 10 mm interval can ensure that meaningful pose changes are captured under the printing accuracy requirements, while ensuring the real-time performance of the system. The selection of this parameter reflects the classic trade-off between sampling period and control performance in the control system.

[0026] S2: After receiving the real-time image, the computer first performs preprocessing: A Gaussian filter with a standard deviation of σ=1.5 is used to smooth the image, effectively suppressing noise introduced during acquisition and transmission. The contrast-limited adaptive histogram equalization algorithm is used to enhance the contrast between the pattern and the background, as well as between different color regions within the pattern, making the features more prominent. Edge detection algorithms such as Canny can be used to obtain the contour information of the image, providing assistance for subsequent feature recognition; The preprocessed image is then fed into a pre-trained feature extraction neural network model.

[0027] This neural network employs a keypoint detection architecture based on a deep convolutional neural network. Its backbone network uses the lightweight HRNet, which maintains high-resolution representation throughout the process and accurately locates keypoints at pixel-level coordinates. The network ultimately outputs a heatmap of the same size as the input image for each feature point, with a Gaussian peak appearing at the corresponding location of the feature point in the heatmap.

[0028] During training, 5000 labeled images of printed patterns (including images from different angles, under different lighting conditions, and with varying deformation states) were used. The pixel coordinates of all feature points in each image were manually and precisely labeled using a labeling tool to form the ground truth. The mean squared error loss function was used as the optimization objective to measure the difference between the predicted and ground truth heatmaps. A stochastic gradient descent optimizer was employed, combined with a multinomial learning rate decay strategy. Data augmentation was applied to improve the model's generalization ability. Training concluded when the model's localization accuracy on the validation set reached an error of less than 0.5 pixels.

[0029] After the network forward propagation, the output heatmap is post-processed. By finding the coordinates of the maximum point in the heatmap of each channel and using quadratic function fitting for sub-pixel precise localization, the precise pixel coordinates of all feature points in the current image can be obtained. ,in This represents the horizontal coordinate of the pixel in the image coordinate system; The vertical coordinate of the pixel in the image coordinate system; This serves as an index identifier. By introducing a pre-trained neural network model, the robustness of the control system to complex operating conditions (such as deformation and changes in illumination) is improved, providing more reliable state observations for control decisions.

[0030] Three-dimensional coordinates; Camera intrinsic parameter matrix It was obtained through precise pre-calibration using Zhang Zhengyou's calibration method, and includes the camera's focal length. and principal point coordinates This describes the projection relationship from the camera coordinate system to the pixel coordinate system: .

[0031] The algorithm calculates the actual 3D pose of the pattern: 3D point set of standard pattern model and the corresponding two-dimensional point set of the image and camera intrinsic parameter matrix Solve for the rotation vector Translation vector ;use Iterative solution to minimize reprojection error:

[0032] After solving, the rotation vector is converted into Euler angles to obtain the current actual pose of the pattern: position coordinates. and rotation angle .

[0033] Affine transformation is used to calculate the actual 2D pose of the pattern. When the fabric is flat and the camera is mounted approximately perpendicular to the plane, the pattern deformation can be approximated as a planar affine or perspective transformation; the solution can be obtained using the direct linear transformation algorithm or the RANSAC optimization algorithm: The translation of the pattern in the two-dimensional plane can be decomposed from the matrix. Rotation angle .

[0034] It's important to further explain that the essence of Gaussian filtering is to find the optimal balance between suppressing noise and preserving key image details. When σ is less than 0.5, the filter kernel window is too small, resulting in insufficient suppression of high-frequency noise, and residual noise can interfere with the accuracy of subsequent feature recognition. When σ is greater than 2.5, it leads to over-smoothing, blurring the edges and details of the printed pattern, which also reduces the accuracy of feature point localization. σ=1.5 effectively suppresses common Gaussian noise and random noise introduced during image acquisition and transmission, while maintaining the sharpness of feature points.

[0035] S3: Position deviation calculation: Compare the current actual 3D pose with the pre-stored standard pattern pose (ideal position). (Compare) The calculated translation vector The three-dimensional position of the pattern center in the camera coordinate system is given; it is necessary to transform the coordinates based on the relative geometric relationship between the camera and the fabric plane. Converted into longitudinal and transverse displacements on the two-dimensional plane of the fabric. This transformation process utilizes the camera's extrinsic parameters (the camera's pose relative to the world coordinate system, which has been pre-calibrated). Calculated Euler angles middle: (Yaw angle): Directly corresponds to the rotational deviation of the pattern within the fabric plane. .

[0036] (pitch angle) and (Roll Angle): These two angles reflect the tilt or wrinkling of the fabric plane itself. In high-precision systems, they can be used as compensation factors for correction. The calculation improves accuracy.

[0037] Calculate the lateral translation deviation: Longitudinal translation deviation: Rotation angle deviation: .

[0038] Multidimensional comprehensive positional deviation: .

[0039] Position deviation calculation: Compare the current two-dimensional actual pose with the pre-stored standard pattern pose (ideal position). Compare and calculate the lateral translation deviation. Longitudinal translation deviation Rotational deviation ; Multidimensional comprehensive positional deviation: .

[0040] Color registration deviation calculation: Based on the pre-stored color swatch template position, the image regions of each color block are accurately cropped from the acquired real-time image (e.g., color blocks are segmented using the HSV color space), and the extracted color block image pixel values ​​are converted to the CIELAB color space; assuming the LAB value of the standard color swatch is... The current average pixel value of the color block is Calculate color difference:

[0041] like If the color is within the perceptible threshold for human eyes, then it is determined that the color plate has a color registration error. The error value for each color plate is calculated sequentially. .

[0042] It needs to be further explained that, The threshold used to determine the existence of color registration deviation is a scientifically selected factor based on human visual science and industry best practices. This ensures that the color registration control system can efficiently and accurately eliminate defective products based on human visual perception, thereby reliably improving the consistency of color quality in the final printed products. The position and color information of the pattern are fused into a unified multi-dimensional deviation vector, providing precise feedback input for subsequent multivariate control algorithms. Position deviation and color deviation together constitute the controlled variable of the control system.

[0043] S4: Employs a multivariable decoupling program control algorithm, taking position deviation and color deviation as inputs, and outputting numerical control commands for synchronous adjustment of the matching motor and color adjustment valve.

[0044] Flower control: Three inputs A multivariable system with two outputs (motor X-axis displacement adjustment and Y-axis displacement adjustment); since the movement of the three axes may have mechanical coupling, it is necessary to first eliminate the interference between dimensions through a decoupling matrix; The rotation center of the printing frame to , The distances between the motor actuators are respectively and (Obtained from mechanical design drawings or calibrated on-site). The equivalent position deviation after decoupling is calculated as follows:

[0045] For small angular deviations ( ),for:

[0046] The equivalent deviation after decoupling and The inputs are fed into two independent digital PID controllers, which calculate the corrections respectively. and Direction and position control quantity and The formula for calculating discrete PID is: ,in It is the first The equivalent bias of the second sampling; This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; For the system's control cycle, This is a circular index variable.

[0047] Similarly, Directional control quantity ,in It is the first The equivalent bias of the second sampling; This is the proportionality coefficient; The integral coefficient; is the differential coefficient.

[0048] Color control: according to The value adjusts the RGB mixing ratio of the color valve. Fuzzy PID control is used to set... The membership function, and the rule table are as follows: like Between, the change in valve opening No adjustment is required; like Between, the change in valve opening Increase by 2%; like Between, the change in valve opening Increase by 5% to 8%; like Between, the change in valve opening Increase by 12% like If it is negative (i.e., the color is too dark), then the rule is... It is a negative value.

[0049] Output valve opening change ,in This is the membership function.

[0050] The final generated coordinated control command is: number of motor stepping pulses. ; ( (Lead screw lead), valve opening percentage .

[0051] It should be further explained that in the rule table, color difference is divided by setting four thresholds (0, 1.5, 2.5, 4.0); when This represents the ideal state, indicating that the current color block is completely consistent with the standard color plate and requires no adjustment. This rarely occurs in actual production, but it exists as a theoretical benchmark. The threshold that is perceptible to the human eye Color difference is usually imperceptible to the human eye. If it exceeds this value, it means that color difference has begun to appear and needs to be adjusted. Therefore, 1.5 is set as the starting threshold for whether intervention is needed. There is a noticeable color difference. At this point, the color difference is already clearly visible, which is a moderate deviation. Moderate adjustment is required to prevent the color from deviating further. This indicates a significant color difference. This indicates a very obvious color difference, which is a serious deviation and requires urgent adjustment. Otherwise, it will affect the quality of the finished product. This value usually corresponds to an emergency adjustment or alarm threshold.

[0052] The use of multivariable decoupling control actively eliminates the internal coupling of the system, while the introduction of fuzzy PID enhances the ability to handle nonlinear and time-varying characteristics, which demonstrates the effective application of modern control theory in complex industrial processes.

[0053] S5: After real-time adjustment of the printing process based on the aforementioned coordinated control command, the system immediately starts a new cycle, forming a highly dynamic closed-loop program control loop. The system operates at a fixed sampling period ( The system operates continuously; after the control command for the current cycle is completed, the system does not simply wait, but immediately triggers the image acquisition for the next cycle (returning to S1). The timing of the entire closed-loop process is strictly scheduled by the real-time operating system of the industrial computer to ensure the precise issuance and execution of CNC commands from image acquisition and processing.

[0054] To ensure the rigor of the control logic, the system performs performance prediction and out-of-tolerance alarms in each cycle. Specifically, after issuing a control command, the system predicts the approximate range of deviation at the next sampling time based on the known equipment response model. If the actual sampled deviation deviates significantly from the predicted value, or if the deviation continues to increase, an alarm mechanism (audible and visual alarm, system shutdown) is immediately triggered, and fault data is recorded, indicating potential anomalies such as mechanical jamming, ink clogging, or camera defocusing, thereby preventing production accidents. This out-of-tolerance early warning mechanism is a crucial component of the advanced control system's reliability assurance.

[0055] A feedforward and feedback composite control strategy is used to suppress disturbances (fabric tension changes, transmission system errors, ink viscosity changes); until all deviations are less than the threshold (position tolerance ±0.1mm, rotation tolerance ±0.5°, color difference tolerance) for 5 consecutive cycles. Once the printing is complete, the system can then keep the parameters ready or automatically switch to the matching and color matching parameters for the next pattern according to the production plan, starting a new control cycle.

[0056] It should be further explained that the positional tolerance is ±0.1mm. Under normal viewing distance, the human eye's limit for distinguishing pattern misalignment is approximately between 0.1mm and 0.2mm. Setting the tolerance to ±0.1mm means that pattern misalignment on the finished product will be completely imperceptible to the human eye, thus ensuring extremely high visual quality. At the same time, ±0.1mm is a precision standard generally pursued by high-end printing. Misalignment below this value will become obvious on delicate patterns and be regarded as a defective product; while pursuing higher precision will increase the cost of equipment and control systems. A rotation tolerance of ±0.5° is the critical threshold to prevent macroscopically visible pattern defects caused by the geometric amplification of rotational deviation.

[0057] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other. In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A positioning control method based on numerical control collaborative control, characterized in that, include: S1: Image acquisition and transmission: The displacement of the target material is monitored in real time by a displacement sensor. When the displacement reaches the preset sampling interval, a trigger command is generated. The trigger command synchronously triggers the pulsed light source and the global exposure camera to acquire clear real-time images of the moving material surface; S2: Pattern feature recognition and localization calculation: The real-time image is processed, and the processed image is used to identify multiple feature points of the pattern through a pre-trained feature extraction neural network model. Based on the spatial relationship of the feature points, the current actual pose of the pattern is calculated using the PnP algorithm or affine transformation. S3: Multi-dimensional error fusion calculation: The current actual pose is compared with the pre-stored standard pattern pose to calculate the multi-dimensional positional deviation of the pattern; at the same time, the pixel set of the color block area of ​​the pattern in the real-time image is extracted and converted to the CIELAB color space. The color difference between the real-time image and the standard color reference is calculated to quantify the color registration deviation value. S4: Intelligent control command generation: Based on the multi-dimensional position deviation and the color matching deviation value, a multi-variable decoupling program control algorithm is used to generate a collaborative control command for synchronously adjusting multiple position actuators and color adjustment mechanisms along the entire line; S5: Closed-loop feedback control: The processing is adjusted based on the cooperative control command. Then, S1 is executed again, and S2 to S5 are executed in sequence, thereby forming a real-time dynamic closed-loop feedback control loop integrated into the overall factory control system.

2. The positioning control method based on numerical control collaborative control according to claim 1, characterized in that, The displacement sensor includes an encoder and is configured to generate a synchronous trigger signal when the displacement of the target material reaches a preset sampling interval, so as to synchronously control the actions of the pulse light source and the global exposure camera.

3. The positioning control method based on numerical control collaborative control according to claim 1, characterized in that, The real-time image is processed by: smoothing the real-time image using a Gaussian filter; and enhancing the contrast between the pattern and the background, and between different color regions within the pattern, using a contrast-limited adaptive histogram equalization algorithm.

4. The positioning control method based on numerical control collaborative control according to claim 1, characterized in that, The current actual pose includes: When using the PnP algorithm, the three-dimensional point set of the standard pattern model, the corresponding two-dimensional point set of the image, and the pre-calibrated camera intrinsic parameter matrix are used as inputs. The EPnP algorithm is used to solve for the rotation vector and translation vector, and the rotation vector is converted into Euler angles to obtain the current actual pose. When using affine transformation, the transformation matrix is ​​solved using a direct linear transformation algorithm or the RANSAC algorithm, and the translation and rotation angles of the pattern in the two-dimensional plane are decomposed from the transformation matrix.

5. The positioning control method based on numerical control collaborative control according to claim 1, characterized in that, The quantized color registration deviation value includes: calculating the Euclidean distance between the mean value of the pixel set of the color block region in the CIELAB color space and the LAB value of the standard color reference, to obtain the color difference. This serves as the color deviation value.

6. The positioning control method based on numerical control collaborative control according to claim 1, characterized in that, The multivariable decoupling program control algorithm includes: constructing a decoupling matrix to eliminate the coupling effect between different motion dimensions of the position actuator; the cooperative control command includes a pulse control signal for driving the position actuator to perform translation compensation, and an analog or digital control signal for adjusting the output of the color adjustment mechanism.

7. The positioning control method based on numerical control collaborative control according to claim 6, characterized in that, The position actuator includes a motor for achieving lateral and longitudinal translation compensation; the color adjustment mechanism includes a regulating valve for adjusting the colorant flow rate; the coordinated control command consists of a motor drive signal and a valve adjustment signal generated based on the deviation calculation result of the same control cycle, so as to achieve coordinated adjustment of pattern positioning and color density.

8. The positioning control method based on numerical control collaborative control according to claim 1, characterized in that, The real-time dynamic closed-loop feedback control system includes: continuous operation with a fixed sampling period; and within each control period, predicting the deviation range of the next sampling moment based on the equipment response model. If the actual sampling deviation deviates from the predicted value by more than a preset threshold or the deviation continues to increase, an out-of-tolerance alarm mechanism is triggered.

9. The positioning control method based on numerical control collaborative control according to claim 1, characterized in that, The closed-loop feedback control includes: using a composite control strategy that combines feedforward control and feedback control to suppress external disturbances; and determining that the control process is complete when all deviations are less than their respective preset tolerance thresholds within multiple consecutive control cycles.

Images