Self-adhesion spot drill high-precision positioning control method based on AI vision
By adopting a high-precision positioning control method for self-adhesive point drills based on AI vision, a displacement-pressure correlation model was established and fault tree analysis was performed. This solved the problems of low positioning accuracy and unclear parameter correlation in traditional self-adhesive point drills, achieving high-precision and consistent production, and improving production efficiency and product quality.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- DONGYANG AIKE CRAFTS CO LTD
- Filing Date
- 2026-01-23
- Publication Date
- 2026-04-24
AI Technical Summary
In traditional self-adhesive dot drill positioning control methods, the poor coordination between displacement and pressure parameters leads to low positioning accuracy and unclear correlation between process parameters and positioning deviation, making it difficult to achieve high-precision and consistent production.
A high-precision positioning control method for self-adhesive point drills based on AI vision is adopted. By collecting material properties and workpiece shape data, a displacement-pressure correlation model is established, and the positioning adjustment parameters are optimized by combining fault tree analysis to achieve closed-loop control.
It improved positioning accuracy and production efficiency, reduced rework and scrap rates, enhanced product quality consistency and production adaptability, and reduced raw material losses and equipment maintenance costs.
Smart Images
Figure CN121921374A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of positioning and control technology, specifically a high-precision positioning and control method for self-adhesive dot drills based on AI vision. Background Technology
[0002] Self-adhesive rhinestone application technology is widely used in jewelry processing, clothing decoration, and electronic product aesthetics due to its advantages of convenient operation and strong adhesion. The market has increasingly stringent requirements for the accuracy and consistency of rhinestone application.
[0003] Traditional self-adhesive dot drill positioning control methods often rely on manual experience or simple open-loop control modes, revealing numerous technical pain points in actual production. On one hand, the coordination between displacement and pressure parameters is poor; the adjustment of the motion mechanism's displacement and the control of the dot drill's contact pressure are independent, lacking a quantitative correlation model. When a slight displacement deviation occurs, the optimal contact pressure cannot be matched for compensation, easily leading to problems such as drill tip tilting, adhesive overflow, or poor adhesion, directly causing fluctuations in positioning accuracy. On the other hand, the correlation between process parameters and positioning deviation is unclear. Once a positioning deviation exceeds the standard during production, technicians find it difficult to quickly trace the specific process parameters, resorting to repeated trial and error to adjust parameters such as motion step distance, motor speed, and adsorption pressure. This not only results in a long debugging cycle but also causes raw material waste and reduces production efficiency.
[0004] With the development of machine vision technology, some drilling equipment has introduced vision inspection modules to assist in positioning. However, existing technologies mostly only use vision systems to collect single-point coordinate data, lacking in-depth analysis of time-series positioning data and failing to extract characteristics such as the trend and fluctuation of deviations. At the same time, the correlation analysis between positioning deviations and process parameters remains at the level of simple correlation statistics, failing to clarify the causal logic between parameter anomalies and deviation exceeding standards, making it difficult to formulate accurate parameter optimization solutions.
[0005] Therefore, developing a high-precision positioning control method for self-adhesive dot drills based on AI vision has become a pressing technical challenge in the industry, and is of great significance for improving the quality of self-adhesive dot drill products and promoting technological upgrading in the industry. Summary of the Invention
[0006] To overcome the shortcomings of existing technologies, this invention proposes a high-precision positioning control method for self-adhesive drilling based on AI vision. This invention primarily addresses the problems of low positioning accuracy caused by poor coordination between displacement and pressure in the self-adhesive drilling process, and the unclear correlation between process parameters and positioning deviation.
[0007] The technical solution adopted by this invention to solve its technical problem is: the high-precision positioning control method for self-adhesive dot drills based on AI vision provided by this invention includes:
[0008] Collect material property data of self-adhesive rhinestones and morphological parameter data of workpieces, and formulate basic positioning process parameters in combination with target spot drilling patterns. Use displacement pressure self-adjustment method to calculate positioning adjustment parameters based on the displacement of adjustment mechanism and spot drilling contact pressure.
[0009] The positioning control parameters are obtained by adjusting the basic positioning parameters according to the positioning adjustment parameters. The alignment image data of the workpiece and the drill bit are collected during the drilling process. The alignment image data is processed by the positioning calibration method to obtain the time-series positioning coordinate data.
[0010] Determine whether the time-series positioning coordinate data deviates from the preset positioning coordinates. If so, perform feature extraction on the current time-series positioning coordinate data to obtain the positioning deviation feature of the ornament.
[0011] The fault tree method is used to analyze the correlation between the positioning deviation characteristics of the trim and the basic positioning process parameters to obtain the positioning correlation parameters. The positioning adjustment parameters are then optimized to obtain the positioning correction control parameters.
[0012] The high-precision positioning control method for self-adhesive dot drills based on AI vision provided by this invention includes the following steps for determining basic positioning process parameters:
[0013] The basic properties of the drilled jewelry, the characteristics of the self-adhesive layer, and the adsorption properties are collected as material property data. The overall dimensions, surface curvature, and edge contour of the workpiece are obtained. Combined with the coordinate range, flatness, and spacing between adjacent areas of the target drilled area, morphological parameter data are generated.
[0014] Extract the diamond arrangement density, arrangement pattern, and diamond specification combination of the target diamond pattern, and output the coordinate-based dot matrix diagram of the pattern.
[0015] The drilling cycle of a single diamond is calculated based on a coordinate grid diagram and the preset daily production capacity of the production line, and the operating speed of the mechanism is calculated by combining morphological parameter data.
[0016] Based on the material property data, the motion zone is divided, and combined with the transmission ratio of the motion mechanism, the running speed of the mechanism is converted into the pulse equivalent of the servo motor, and the segmented motion acceleration is set.
[0017] The initial value of the vacuum adsorption pressure is set according to the adsorption characteristics, and the spot drilling contact pressure is determined by combining the characteristics of the self-adhesive layer and the flatness of the workpiece surface.
[0018] Based on the required accuracy of the partition, the initial step distance of the motion mechanism is calculated. For different curved workpieces, the attitude angle of the drill bit is corrected by combining the normal vector angle of the drilling area.
[0019] The high-precision positioning control method for self-adhesive dot drills based on AI vision provided by this invention includes the following steps for calculating positioning adjustment parameters:
[0020] Establish a correlation model between the displacement of the motion mechanism and the contact pressure of the drilling, record the optimal contact pressure value under different displacement deviations, and form a displacement-pressure correlation comparison table.
[0021] The target displacement deviation is calculated based on the target coordinates of the coordinated dot matrix and the actual stopping coordinates of the current motion mechanism. The target displacement is then corrected by considering mechanical clearance and temperature and humidity.
[0022] The target required contact pressure is obtained by matching the initial contact pressure value corresponding to the target required displacement amount according to the displacement pressure correlation table and adjusting it in conjunction with the point drilling contact pressure.
[0023] The target displacement is converted into motor control parameters, and the displacement adjustment parameters are obtained by adjusting them according to the motion scenario and different types of workpieces.
[0024] The target required contact pressure is converted into actuator parameters. The pressure adjustment parameters are obtained by adjusting the adhesive layer viscosity and contact time. These parameters are then integrated with the displacement adjustment parameters to obtain the positioning adjustment parameters.
[0025] The high-precision positioning control method for self-adhesive dot drills based on AI vision provided by this invention includes the following steps in forming a displacement-pressure correlation lookup table:
[0026] The displacement offset of the motion mechanism and the contact pressure of the drill bit are used as independent variables, while the final positioning accuracy and the good rate of the drill-decorated part are used as dependent variables. The displacement deviation range is divided according to the target positioning accuracy threshold.
[0027] The single-variable method was used to screen the independent and dependent variables according to the logic of fixed displacement deviation - adjustment of contact pressure - collection of results to obtain valid data.
[0028] For each combination of displacement deviations, displacement pressure data with the smallest average positioning residual, the smallest residual standard deviation, and qualified fit are selected from the effective data. Based on the displacement pressure data, a correlation model is constructed using multiple linear regression.
[0029] The optimal contact pressure value is obtained by substituting the displacement deviation combination into the correlation model. A displacement-pressure correlation comparison table is formed using the displacement deviation as an index.
[0030] The high-precision positioning control method for self-adhesive dot drills based on AI vision provided by this invention includes the following steps for adjusting the positioning control parameters:
[0031] The positioning adjustment parameters are prioritized according to their adjustment effect and response speed, and divided into first priority, second priority and third priority.
[0032] For the first priority, based on the current pulse equivalent of the servo motor, the displacement compensation pulse number is converted into the actual displacement adjustment amount. It is then determined whether the ratio of the actual displacement adjustment amount to the basic motion step distance is greater than a preset threshold. If so, the basic motion step distance is corrected.
[0033] For the second priority, the positioning adjustment parameters are superimposed on the basic contact pressure threshold to obtain the pressure control range, and the vacuum pressure decay time is corrected according to the viscous force deviation when the drill bit is released.
[0034] For the third priority, in response to temperature and humidity fluctuations, the temperature and humidity compensation coefficient is introduced into the basic positioning parameters to correct the movement step distance and pressure parameters. In response to mechanical clearance wear, the clearance compensation coefficient is superimposed to correct the displacement compensation pulse number.
[0035] The high-precision positioning control method for self-adhesive dot drills based on AI vision provided by this invention includes the following steps for acquiring alignment image data:
[0036] Zhang's calibration method was used to perform hand-eye calibration of the industrial camera and the drilling motion mechanism, and to establish the mapping relationship between pixel coordinates and physical coordinates.
[0037] Select the light source based on the material properties data, and adjust the brightness and color temperature of the light source. Match the exposure time according to the speed of the moving mechanism.
[0038] Based on the coordinate-based dot matrix, the visual detection area is divided within the camera's field of view. The image sampling frequency is calculated according to the speed of the moving mechanism and the target positioning accuracy, and the triggering method is set by the encoder.
[0039] When the motion mechanism carrying the diamond ornament moves to the visual detection area, the trigger mode is reached, and the process image is acquired according to the image sampling frequency.
[0040] Distortion correction, denoising, grayscale conversion, and contrast enhancement are performed on the process image to obtain alignment image data.
[0041] The high-precision positioning control method for self-adhesive dot drills based on AI vision provided by this invention includes the following steps for obtaining time-series positioning coordinate data:
[0042] Based on material property data, feature points of different shaped diamond ornaments are extracted from the alignment image data, and pseudo feature points caused by noise and reflection are identified to obtain the diamond ornament feature points.
[0043] The standard template image of the workpiece marking points is retrieved, and the marking points are matched in the alignment image data using a normalized cross-correlation algorithm to obtain the pixel coordinates of the marking points.
[0044] Using the pixel coordinates of the marked points as a reference, the pixel-level relative pose of the diamond trim to the workpiece is calculated and converted into a physical relative pose. Combined with the theoretical reference coordinates of the workpiece in the motion coordinate system, the real-time absolute positioning coordinates of the diamond trim are calculated.
[0045] Using the pulse signal of the main encoder of the production line as the time reference, timestamps are added to the real-time absolute positioning coordinates corresponding to each frame of the image for synchronous alignment to obtain the timing positioning coordinate data.
[0046] The self-adhesive dot drilling high-precision positioning control method based on AI vision provided by this invention includes the following steps for extracting the positioning deviation features of the ornament:
[0047] Retrieve the corresponding sampling time from the current time-series positioning coordinate data, and calculate the absolute and relative deviations for each sampling time to form a deviation dataset.
[0048] The bias dataset is divided into multiple regionalized subsets based on the visual detection regions, and temporal features are extracted from three dimensions: trend, volatility, and suddenness.
[0049] Spatial and temporal features of each regionalized subset are extracted from three dimensions—location characteristics, morphological characteristics, and correlation characteristics—and integrated to obtain the positioning deviation features.
[0050] The high-precision positioning control method for self-adhesive dot drills based on AI vision provided by this invention includes the following steps for obtaining positioning association parameters:
[0051] The analysis object is determined based on the characteristics of the positioning deviation. The deviation problem that needs to be solved is defined as the top event, and the events related to the basic positioning process parameters are defined as the bottom events.
[0052] For the top event, it is decomposed into intermediate events from four aspects: visual detection, motion execution, stress control, and environmental interference, and a first-level causal chain is established.
[0053] Each intermediate event is further decomposed down until it is associated with the underlying event, the logical gate relationships between different events are determined, and a fault tree logic diagram is drawn.
[0054] Based on the fault tree logic diagram, find the combination of bottom events that led to the top event to form a minimal cut set, and calculate the structural importance coefficient of each bottom event.
[0055] The probability of occurrence of each underlying event is calculated, and the partial derivative of the probability of occurrence of the top event with respect to the probability of occurrence of the underlying event is used to obtain the probability importance. The critical importance is then calculated in combination with the structural importance coefficient.
[0056] The process parameters corresponding to critical importance values greater than the preset importance threshold are taken as influencing process parameters, and the corresponding rules are extracted by combining the fault tree logic diagram to obtain the location association parameters.
[0057] The high-precision positioning control method for self-adhesive dot drills based on AI vision provided by this invention includes the following steps for obtaining positioning correction control parameters:
[0058] A parameter mapping table is established based on the positioning association parameters and positioning adjustment parameters, from which the parameter adjustment direction and adjustment magnitude threshold are determined.
[0059] The adjustment weight factors corresponding to the process parameters are calculated based on their critical importance, and then sorted from high to low to determine the adjustment priority.
[0060] Based on the adjustment priority, the parameter correction amount is calculated using the parameter mapping table, and the results are summarized to obtain the preliminary adjustment parameters.
[0061] Based on the first-level causal chain, coupling compensation coefficients are added to the parameter combinations with coupling relationships, and the positioning correction control parameters are obtained by constraint verification based on the equipment safety threshold.
[0062] The beneficial effects of this invention are as follows:
[0063] 1. This invention eliminates systematic errors in visual inspection by using hand-eye calibration and precise matching of feature points, achieving high-precision conversion from pixel-level to physical-level accuracy. Fault tree analysis clarifies influencing parameters, and combined with coupling compensation correction, ensures stable positioning accuracy reaching a preset threshold, significantly improving positioning control precision. Furthermore, the displacement-pressure correlation model avoids defects such as diamond adsorption tilt, release offset, and adhesive overflow; spatiotemporal dimension analysis and closed-loop optimization of deviation characteristics effectively reduce the proportion of local high-deviation areas, improving the yield of point-drill bonding. Pre-setting and prioritizing process parameters based on production cycle time shortens the parameter debugging cycle and improves the point-drilling efficiency of individual diamonds; the closed-loop control system reduces rework and scrap rates caused by positioning deviations, lowering raw material losses and equipment maintenance costs; it is applicable to diamonds of different materials and shapes, as well as workpieces of different forms, improving production efficiency, reducing production costs, and enhancing process adaptability and robustness. Attached Figure Description
[0064] The invention will now be further described with reference to the accompanying drawings.
[0065] Figure 1 This is a flowchart illustrating the high-precision positioning control method for self-adhesive dot drills based on AI vision provided in an embodiment of the present invention.
[0066] Figure 2 This is a flowchart illustrating the process of obtaining basic positioning parameters in the AI vision-based high-precision positioning control method for self-adhesive dot drilling provided in this embodiment of the invention.
[0067] Figure 3This is a flowchart illustrating the process of obtaining positioning-related parameters in the AI vision-based high-precision positioning control method for self-adhesive dot drills provided in this embodiment of the invention. Detailed Implementation
[0068] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.
[0069] like Figures 1 to 3 As shown in the embodiment of the present invention, the high-precision positioning control method for self-adhesive dot drills based on AI vision includes:
[0070] Collect material property data of self-adhesive rhinestones and morphological parameter data of workpieces, and formulate basic positioning process parameters in combination with target spot drilling patterns. Use displacement pressure self-adjustment method to calculate positioning adjustment parameters based on the displacement of adjustment mechanism and spot drilling contact pressure.
[0071] The steps for determining the basic positioning process parameters include:
[0072] The basic properties of the drilled jewelry, the characteristics of the self-adhesive layer, and the adsorption properties are collected as material property data. The overall dimensions, surface curvature, and edge contour of the workpiece are obtained. Combined with the coordinate range, flatness, and spacing between adjacent areas of the target drilled area, morphological parameter data are generated.
[0073] Basic attribute collection: Record the geometric parameters of the rhinestone jewelry, including diameter / side length, thickness, chamfer size, etc.; material type, including acrylic, rhinestone, crystal, etc.; and surface characteristics, including reflectivity, transparency, etc.
[0074] Self-adhesive layer property data collection: The viscosity coefficient and peel strength of the adhesive layer were tested using a tensile testing machine, and the thickness and uniformity of the adhesive layer were measured. The change curves of the adhesive layer viscosity with temperature and humidity were recorded to determine the optimal temperature and humidity range for drilling.
[0075] Adsorption characteristics collection: Test the adsorption stability of diamond jewelry under different vacuum pressures to determine the vacuum pressure threshold range that prevents the diamond jewelry from tilting or falling off.
[0076] Extract the diamond arrangement density, arrangement pattern, and diamond specification combination of the target diamond pattern, and output the coordinate-based dot matrix diagram of the pattern.
[0077] The drilling cycle of a single diamond is calculated based on a coordinate grid diagram and the preset daily production capacity of the production line, and the operating speed of the mechanism is calculated by combining morphological parameter data.
[0078] Based on the material property data, the motion zone is divided, and combined with the transmission ratio of the motion mechanism, the running speed of the mechanism is converted into the pulse equivalent of the servo motor, and the segmented motion acceleration is set.
[0079] The initial value of the vacuum adsorption pressure is set according to the adsorption characteristics, and the spot drilling contact pressure is determined by combining the characteristics of the self-adhesive layer and the flatness of the workpiece surface.
[0080] Based on the required accuracy of the partition, the initial step distance of the motion mechanism is calculated. For different curved workpieces, the attitude angle of the drill bit is corrected by combining the normal vector angle of the drilling area.
[0081] The steps for calculating positioning adjustment parameters include:
[0082] Establish a correlation model between the displacement of the motion mechanism and the contact pressure of the drilling, record the optimal contact pressure value under different displacement deviations, and form a displacement-pressure correlation comparison table.
[0083] The steps to create a displacement-pressure correlation table include:
[0084] The displacement offset of the motion mechanism and the contact pressure of the drill bit are used as independent variables, while the final positioning accuracy and the good rate of the drill-decorated part are used as dependent variables. The displacement deviation range is divided according to the target positioning accuracy threshold.
[0085] The single-variable method was used to screen the independent and dependent variables according to the logic of fixed displacement deviation - adjustment of contact pressure - collection of results to obtain valid data.
[0086] The steps to obtain valid data may include:
[0087] (1) Set a fixed displacement deviation value. (2) Adjust the contact pressure step by step: start from 0.05N, increase by 0.05N each time, until 0.50N. (3) Repeat the spot drilling experiment 30 times under each pressure (to eliminate random errors), and record the following for each experiment: the final positioning residual of visual inspection and the fitting status of the diamond trim (whether there is glue overflow, detachment, or tilting). (4) Change to the next displacement deviation combination and repeat steps (2)-(3) until all preset displacement deviation ranges are covered.
[0088] For each combination of displacement deviations, displacement pressure data with the minimum average positioning residual, the minimum residual standard deviation, and satisfactory fit are selected from the valid data. A correlation model is then constructed based on the displacement pressure data using multiple linear regression, expressed as follows:
[0089]
[0090] In the formula, It is the optimal contact pressure. b and c are displacement deviations. , , The correlation coefficient, That is the reference pressure.
[0091] The optimal contact pressure value is obtained by substituting the displacement deviation combination into the correlation model. A displacement-pressure correlation comparison table is formed using the displacement deviation as an index.
[0092] Displacement-Pressure Correlation Reference Table:
[0093]
[0094] The target displacement deviation is calculated based on the target coordinates of the coordinated dot matrix and the actual stopping coordinates of the current motion mechanism. The target displacement is then corrected by considering mechanical clearance and temperature and humidity.
[0095] The target required contact pressure is obtained by matching the initial contact pressure value corresponding to the target required displacement amount according to the displacement pressure correlation table and adjusting it in conjunction with the point drilling contact pressure.
[0096] The target displacement is converted into motor control parameters, and the displacement adjustment parameters are obtained by adjusting them according to the motion scenario and different types of workpieces.
[0097] For high-speed motion scenarios, inertia compensation is added: the overshoot deviation caused by inertia is calculated based on the moving speed of the motion mechanism. For curved workpieces, attitude compensation is added: the number of rotational pulses is corrected based on the normal vector angle of the drilling area to ensure that the drill bit is perpendicularly attached to the workpiece surface.
[0098] Displacement adjustment parameters may include: X-axis pulse count, Y-axis pulse count, rotary axis pulse count, displacement adjustment rate, etc.
[0099] The target required contact pressure is converted into actuator parameters. The pressure adjustment parameters are obtained by adjusting the adhesive layer viscosity and contact time. These parameters are then integrated with the displacement adjustment parameters to obtain the positioning adjustment parameters.
[0100] If the contact pressure is controlled by a pneumatic valve, the target required contact pressure F1 is converted into the corresponding pneumatic pressure value according to the pneumatic-pressure calibration curve.
[0101] If controlled by a servo pressure head, the downward displacement of the pressure head is converted into a force sensor for real-time feedback calibration.
[0102] Adjust the air pressure or pressure head displacement according to the ambient temperature—when the temperature rises, the adhesive layer viscosity decreases, so increase the pressure appropriately. When the temperature drops, decrease the pressure.
[0103] Add contact time compensation parameter: After pressure adjustment, match the corresponding pressing and holding time of the diamond jewelry to ensure a firm fit.
[0104] Pressure adjustment parameters may include: target air pressure value, pressure holding time, and pressure adjustment response speed.
[0105] The positioning control parameters are obtained by adjusting the basic positioning parameters according to the positioning adjustment parameters. The alignment image data of the workpiece and the drill bit are collected during the drilling process. The alignment image data is processed by the positioning calibration method to obtain the time-series positioning coordinate data.
[0106] The steps for adjusting the positioning control parameters include:
[0107] The positioning adjustment parameters are prioritized according to their adjustment effect and response speed, and divided into first priority, second priority and third priority.
[0108] First priority: displacement compensation pulse count → directly corrects coordinate deviation, fastest response.
[0109] Second priority: Contact pressure correction value and vacuum pressure decay rate → affect the stability of diamond fitting and indirectly optimize positioning accuracy.
[0110] Third priority: Dynamic compensation coefficients, such as inertia / temperature and humidity / mechanical clearance → long-term stability compensation, adapting to fluctuations in operating conditions.
[0111] For the first priority, based on the current pulse equivalent of the servo motor, the displacement compensation pulse number is converted into the actual displacement adjustment amount. It is then determined whether the ratio of the actual displacement adjustment amount to the basic motion step distance is greater than a preset threshold. If so, the basic motion step distance is corrected.
[0112] For the second priority, the positioning adjustment parameters are superimposed on the basic contact pressure threshold to obtain the pressure control range, and the vacuum pressure decay time is corrected according to the viscous force deviation when the drill bit is released.
[0113] The formula for correcting the vacuum pressure decay time is expressed as follows:
[0114]
[0115] In the formula, This is the corrected vacuum pressure decay time. It is the decay time of the basic vacuum pressure. It is the viscosity compensation coefficient. It is the viscosity coefficient deviation rate.
[0116] For the third priority, in response to temperature and humidity fluctuations, the temperature and humidity compensation coefficient is introduced into the basic positioning parameters to correct the movement step distance and pressure parameters. In response to mechanical clearance wear, the clearance compensation coefficient is superimposed to correct the displacement compensation pulse number.
[0117] The steps for acquiring alignment image data include:
[0118] Zhang's calibration method was used to perform hand-eye calibration of the industrial camera and the drilling motion mechanism, and to establish the mapping relationship between pixel coordinates and physical coordinates.
[0119] Select the light source based on the material properties data, and adjust the brightness and color temperature of the light source. Match the exposure time according to the speed of the moving mechanism.
[0120] Based on the coordinate-based dot matrix, the visual detection area is divided within the camera's field of view. The image sampling frequency is calculated according to the speed of the moving mechanism and the target positioning accuracy, and the triggering method is set by the encoder.
[0121] Based on the distribution of the spot drilling area, it is divided into several sub-inspection areas, such as the pattern area and edge area on the workpiece surface. Each sub-inspection area must cover the alignment range of at least one diamond to avoid blind spots.
[0122] Based on the moving speed of the motion mechanism and the target positioning accuracy, the image sampling frequency is calculated: the sampling frequency f must satisfy f≥2v / L (v is the motion speed, L is the diameter of the diamond ornament) to ensure that the moving trajectory of the diamond ornament can be completely captured between two adjacent frames without missing the alignment key frames.
[0123] When the motion mechanism carrying the diamond ornament moves to the visual detection area, the trigger mode is reached, and the process image is acquired according to the image sampling frequency.
[0124] For curved workpieces or multi-workstation drilling scenarios, multi-camera collaborative acquisition is adopted: cameras at different angles are triggered simultaneously to acquire multi-angle alignment images of the drill bit and the workpiece, and the complete alignment state is restored by image stitching or 3D reconstruction.
[0125] Distortion correction, denoising, grayscale conversion, and contrast enhancement are performed on the process image to obtain alignment image data.
[0126] Distortion correction: Based on the distortion coefficients calibrated by the camera, the image is geometrically corrected to eliminate the shape distortion of the diamond jewelry caused by lens distortion.
[0127] Denoising: Median filtering or Gaussian filtering algorithms are used to remove salt and pepper noise and electronic noise from the image while preserving the edge and contour features of the diamond jewelry.
[0128] Grayscale conversion and contrast enhancement: Convert color images to grayscale images, and enhance the grayscale contrast of the diamond ornament and the workpiece through histogram equalization or adaptive thresholding algorithm to highlight alignment feature points, such as the center of the diamond ornament and the positioning mark point of the workpiece.
[0129] Select valid images that meet the clarity requirements, have an edge gradient ≥20, and are free of ghosting and occlusion. Remove blurry or ghosting images caused by light source failure or mechanical vibration.
[0130] The steps to obtain time-series positioning coordinate data include:
[0131] Based on material property data, feature points of different shaped diamond ornaments are extracted from the alignment image data, and pseudo feature points caused by noise and reflection are identified to obtain the diamond ornament feature points.
[0132] For round diamond ornaments: The Hough circle detection algorithm is used to extract the center pixel coordinates and radius of the ornament.
[0133] For irregularly shaped diamond jewelry: a contour detection + corner extraction algorithm, such as Shi-Tomasi corner detection, is used to obtain the coordinates of the characteristic corners of the diamond jewelry, and then the center coordinates and rotation angle of the diamond jewelry are calculated by using the minimum bounding rectangle.
[0134] The standard template image of the workpiece marking points is retrieved, and the marking points are matched in the alignment image data using a normalized cross-correlation algorithm to obtain the pixel coordinates of the marking points.
[0135] Using the pixel coordinates of the marked points as a reference, the pixel-level relative pose of the diamond trim to the workpiece is calculated and converted into a physical relative pose. Combined with the theoretical reference coordinates of the workpiece in the motion coordinate system, the real-time absolute positioning coordinates of the diamond trim are calculated.
[0136] Using the pulse signal of the main encoder of the production line as the time reference, timestamps are added to the real-time absolute positioning coordinates corresponding to each frame of the image for synchronous alignment to obtain the timing positioning coordinate data.
[0137] Determine whether the time-series positioning coordinate data deviates from the preset positioning coordinates. If so, perform feature extraction on the current time-series positioning coordinate data to obtain the positioning deviation feature of the ornament.
[0138] The steps for extracting the positioning deviation features of the ornaments include:
[0139] Retrieve the corresponding sampling time from the current time-series positioning coordinate data, and calculate the absolute and relative deviations for each sampling time to form a deviation dataset.
[0140] The bias dataset is divided into multiple regionalized subsets based on the visual detection regions, and temporal features are extracted from three dimensions: trend, volatility, and suddenness.
[0141] Trend feature extraction: Calculate the average deviation and cumulative deviation within a preset time period. Use linear regression to fit the deviation change curve over time, and record the trend coefficient and average deviation as trend features.
[0142] Volatility Feature Extraction: Calculate the standard deviation and range of the deviation data to characterize the dispersion of the deviation. Perform a Fast Fourier Transform (FFT) on the deviation time series to extract the main fluctuation frequencies and corresponding amplitudes of the deviation, and identify the periodic patterns of the deviation fluctuations, such as periodic deviations consistent with the vibration frequency of a motor. Record the standard deviation, range, and main fluctuation frequencies as volatility features.
[0143] Sudden feature extraction: Set a deviation threshold and identify deviation peaks that exceed the threshold.
[0144] Statistically analyze the duration and frequency of exceeding the threshold. Calculate the deviation change between adjacent sampling points and extract the maximum change as the deviation mutation feature. Record the peak value, duration, frequency, and maximum change as the sudden change feature.
[0145] Spatial and temporal features of each regionalized subset are extracted from three dimensions—location characteristics, morphological characteristics, and correlation characteristics—and integrated to obtain the positioning deviation features.
[0146] Location characteristic extraction: A spatial heatmap of deviations is drawn: using a two-dimensional plane of the workpiece as a base, color intensity represents the average deviation magnitude of each region, identifying high-deviation areas. The area proportion and center coordinates of high-deviation areas are calculated to locate key areas where deviations are concentrated.
[0147] Morphological characteristic extraction: For angular deviations, the distribution ratio of angular deviations in different regions is statistically analyzed to identify high-incidence areas of diamond jewelry tilt. The anisotropy coefficient of the deviation is calculated.
[0148] Correlation feature extraction: Analyze the spatial correlation between deviation and drilling process parameters: such as comparing the contact pressure and movement step distance of high deviation areas with the corresponding areas to identify areas where parameters do not match.
[0149] Calculate the correlation coefficient of the deviation between adjacent drilling positions to determine whether the deviation has spatial transitivity, such as if an excessive deviation at a certain point leads to a chain reaction of increased deviations at adjacent points.
[0150] The fault tree method is used to analyze the correlation between the positioning deviation characteristics of the trim and the basic positioning process parameters to obtain the positioning correlation parameters. The positioning adjustment parameters are then optimized to obtain the positioning correction control parameters.
[0151] The steps to obtain the location association parameters include:
[0152] The analysis object is determined based on the characteristics of the positioning deviation. The deviation problem that needs to be solved is defined as the top event, and the events related to the basic positioning process parameters are defined as the bottom events.
[0153] For the top event, it is decomposed into intermediate events from four aspects: visual detection, motion execution, stress control, and environmental interference, and a first-level causal chain is established.
[0154] The top event is an excessive deviation in planar position. First-level intermediate events can be decomposed into:
[0155] ME1: Displacement execution deviation of the motion mechanism.
[0156] ME2: Jewelry adsorption-release posture shift.
[0157] ME3: Visual coordinate transformation deviation.
[0158] ME4: Mechanical deformation deviation caused by ambient temperature and humidity.
[0159] Each intermediate event is further decomposed down until it is associated with the underlying event, the logical gate relationships between different events are determined, and a fault tree logic diagram is drawn.
[0160] Event mapping table:
[0161]
[0162] .
[0163] Based on the fault tree logic diagram, identify the combination of bottom events that led to the top event to form a minimal cut set, and calculate the structural importance coefficient of each bottom event, expressed by the formula:
[0164]
[0165] In the formula, It is the structural importance coefficient. It is the total number of minimum cut sets. It is the minimum cut set number containing the base events. It is the first The number of base events contained in a minimal cut set.
[0166] The probability of occurrence of each underlying event is calculated, and the partial derivative of the probability of occurrence of the top event with respect to the probability of occurrence of the underlying event is used to obtain the probability importance. The critical importance is then calculated in combination with the structural importance coefficient.
[0167] The formula for calculating probability importance is expressed as:
[0168]
[0169] In the formula, It is a probability importance. It is the probability of the top event occurring. It represents the probability of the bottom component occurring.
[0170] The formula for calculating critical importance is expressed as follows:
[0171]
[0172] In the formula, It is critical importance.
[0173] The process parameters corresponding to critical importance values greater than the preset importance threshold are taken as influencing process parameters, and the corresponding rules are extracted by combining the fault tree logic diagram to obtain the location association parameters.
[0174] The corresponding rules may include rule 1: excessive step distance → displacement execution deviation → excessive planar position deviation.
[0175] Rule 2: Excessive contact pressure + excessive adhesive viscosity → adsorption and release deviation → excessive deviation in planar position.
[0176] The steps to obtain the positioning correction control parameters include:
[0177] A parameter mapping table is established based on the positioning association parameters and positioning adjustment parameters, from which the parameter adjustment direction and adjustment magnitude threshold are determined.
[0178] Parameter mapping table:
[0179]
[0180] Adjustment direction: If the step distance is too large, causing the deviation to exceed the standard, adjust the direction by reducing the step distance. If insufficient contact pressure causes unstable adhesion, adjust the direction by increasing the pressure.
[0181] Determine the maximum adjustment threshold for a single step: Based on the parameter-deviation influence ratio in the association rules, set an upper limit for the single step adjustment, such as a single step adjustment of the movement step distance ≤ 0.002 mm and a single step adjustment of the contact pressure ≤ 0.05 N, to avoid sudden parameter changes that could cause equipment vibration or damage to the drill bit.
[0182] The adjustment weight factors corresponding to the process parameters are calculated based on their critical importance, and then sorted from high to low to determine the adjustment priority.
[0183] Based on the adjustment priority, the parameter correction amount is calculated using the parameter mapping table, and the results are summarized to obtain the preliminary adjustment parameters.
[0184] For the parameter with the highest weighting factor, the correction amount is calculated by combining the influence ratio in the association rule.
[0185] Medium and low priority parameter correction: Calculate the correction amount of other parameters sequentially according to the weighting factor, while also considering the coupling relationship between parameters. For example, after adjusting the step size, the displacement compensation pulse number needs to be corrected simultaneously.
[0186] Based on the first-level causal chain, coupling compensation coefficients are added to the parameter combinations with coupling relationships, and the positioning correction control parameters are obtained by constraint verification based on the equipment safety threshold.
[0187] In summary, the AI vision-based high-precision positioning control method for self-adhesive dot drills provided in this embodiment achieves micron-level positioning accuracy through real-time AI vision detection and displacement-pressure self-adjustment control. The fault tree analysis method can quickly identify key process parameters affecting positioning accuracy, improving the efficiency of process parameter adjustment. Through online feedback and adaptive adjustment mechanisms, it can compensate for interference factors such as mechanical clearance, temperature and humidity changes, and workpiece deformation in real time, enhancing the system's adaptability. High-precision positioning and stable control ensure consistent bonding position and pressure for each drill bit, significantly improving product appearance consistency and overall product quality consistency.
[0188] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0189] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A high-precision positioning control method for self-adhesive dot drills based on AI vision, comprising: Collect material property data of self-adhesive rhinestones and morphological parameter data of workpieces, and formulate basic positioning process parameters in combination with target point drilling pattern. Use displacement pressure self-adjustment method to calculate positioning adjustment parameters based on the displacement of adjustment mechanism and point drilling contact pressure. The positioning control parameters are obtained by adjusting the basic positioning parameters according to the positioning adjustment parameters. Alignment image data of the workpiece and the drill bit are collected during the drilling process. The alignment image data is processed using the positioning calibration method to obtain the time-series positioning coordinate data. If the time-series positioning coordinate data deviates from the preset positioning coordinates, then feature extraction is performed on the current time-series positioning coordinate data to obtain the positioning deviation feature of the ornament; The positioning deviation characteristics of the trim piece and the basic positioning process parameters are analyzed using the fault tree method to obtain positioning correlation parameters. The positioning adjustment parameters are then optimized to obtain positioning correction control parameters.
2. The high-precision positioning control method for self-adhesive dot drills based on AI vision according to claim 1, characterized in that: The steps for determining the basic positioning process parameters include: The basic properties of the diamond trim, the characteristics of the self-adhesive layer, and the adsorption properties are collected as the material property data. The overall size, surface curvature, and edge contour of the workpiece are obtained. The morphological parameter data are generated by combining the coordinate range, flatness, and spacing between adjacent areas of the target point drilling area. Extract the diamond arrangement density, arrangement pattern and diamond specification combination of the target diamond pattern, and output the coordinate-based dot matrix diagram of the pattern; The drilling cycle of a single diamond is calculated based on the coordinated dot matrix and the preset daily production capacity of the production line, and the operating speed of the mechanism is calculated in combination with the morphological parameter data. The motion zone is divided according to the material property data, and the motion mechanism's transmission ratio is combined to convert the mechanism's running speed into the pulse equivalent of the servo motor, and the segmented motion acceleration is set. The initial value of the vacuum adsorption pressure is set according to the adsorption characteristics, and the spot drilling contact pressure is determined by combining the characteristics of the self-adhesive layer and the flatness of the workpiece surface. Based on the required accuracy of the partition, the initial step distance of the motion mechanism is calculated. For different curved workpieces, the attitude angle of the drill bit is corrected by combining the normal vector angle of the drilling area.
3. The high-precision positioning control method for self-adhesive dot drills based on AI vision according to claim 2, characterized in that: The steps for calculating the positioning adjustment parameters include: Establish a correlation model between the displacement of the motion mechanism and the contact pressure of the drilling, record the optimal contact pressure value under different displacement deviations, and form a displacement-pressure correlation comparison table. Based on the target coordinates of the coordinated dot matrix and the actual stopping coordinates of the current motion mechanism, the drill displacement deviation is calculated, and then corrected by mechanical clearance and temperature and humidity to obtain the target required displacement. The target required contact pressure is obtained by matching the initial contact pressure value corresponding to the target required displacement amount with the displacement pressure correlation table and adjusting it in conjunction with the point drilling contact pressure. The target displacement is converted into motor control parameters, and the displacement adjustment parameters are obtained by adjusting them according to the motion scenario and different types of workpieces. The target required contact pressure is converted into actuator parameters, and the pressure adjustment parameters are obtained by adjusting the adhesive layer viscosity and contact time. These parameters are then integrated with the displacement adjustment parameters to obtain the positioning adjustment parameters.
4. The high-precision positioning control method for self-adhesive dot drills based on AI vision according to claim 3, characterized in that: The steps for forming the displacement-pressure correlation lookup table include: The displacement offset of the motion mechanism and the contact pressure of the spot drill are used as independent variables, and the final positioning accuracy and the good rate of the drill-decorated fit are used as dependent variables. The displacement deviation range is divided according to the target positioning accuracy threshold. The independent and dependent variables were screened using a single variable method, following the logic of fixed displacement deviation - adjusted contact pressure - data collection, to obtain valid data. For each combination of displacement deviations, the displacement pressure data with the smallest average positioning residual, the smallest residual standard deviation, and the qualified fit status are selected from the effective data. Based on the displacement pressure data, a correlation model is constructed using multiple linear regression. The optimal contact pressure value is obtained by substituting the displacement deviation combination into the correlation model, and the displacement pressure correlation table is formed by using the displacement deviation as an index.
5. The high-precision positioning control method for self-adhesive dot drills based on AI vision according to claim 1, characterized in that: The steps for adjusting the positioning control parameters include: The positioning adjustment parameters are prioritized according to their adjustment effect and response speed, and divided into first priority, second priority and third priority. For the first priority, based on the current pulse equivalent of the servo motor, the displacement compensation pulse number is converted into the actual displacement adjustment amount. It is then determined whether the ratio of the actual displacement adjustment amount to the basic motion step distance is greater than a preset threshold. If so, the basic motion step distance is corrected. For the second priority, the positioning adjustment parameters are superimposed on the basic contact pressure threshold to obtain the pressure control range, and the vacuum pressure decay time is corrected according to the viscous force deviation when the drill bit is released. For the third priority, in response to temperature and humidity fluctuations, a temperature and humidity compensation coefficient is introduced into the basic positioning parameters to correct the movement step distance and pressure parameters. In response to mechanical clearance wear, a clearance compensation coefficient is superimposed to correct the displacement compensation pulse number.
6. The high-precision positioning control method for self-adhesive dot drills based on AI vision according to claim 2, characterized in that: The steps for acquiring the alignment image data include: Zhang's calibration method was used to perform hand-eye calibration of the industrial camera and the drilling motion mechanism to establish the mapping relationship between pixel coordinates and physical coordinates; Select a light source based on the material property data, and adjust the brightness and color temperature of the light source, and match the exposure time according to the speed of the moving mechanism; Based on the coordinated dot matrix, a visual detection area is divided within the camera's field of view. The image sampling frequency is calculated based on the speed of the moving mechanism and the target positioning accuracy. The triggering method is set using an encoder. When the motion mechanism carrying the diamond ornament moves to the visual detection area and the triggering mode is reached, a process image is acquired according to the image sampling frequency. The alignment image data is obtained by performing distortion correction, noise reduction, grayscale conversion, and contrast enhancement on the process image.
7. The high-precision positioning control method for self-adhesive dot drills based on AI vision according to claim 1, characterized in that: The steps for obtaining the time-series positioning coordinate data include: Based on the material property data, feature points of different shaped diamond ornaments are extracted from the alignment image data, and pseudo feature points caused by noise and reflection are identified to obtain the diamond ornament feature points. A standard template image of the workpiece marking points is retrieved, and the marking points are matched in the alignment image data using a normalized cross-correlation algorithm to obtain the pixel coordinates of the marking points; Using the pixel coordinates of the marked points as a reference, the pixel-level relative pose of the diamond ornament to the workpiece is calculated, converted into a physical relative pose, and combined with the theoretical reference coordinates of the workpiece in the motion coordinate system, the real-time absolute positioning coordinates of the diamond ornament are calculated. Using the pulse signal of the main encoder of the production line as the time reference, timestamps are added to the real-time absolute positioning coordinates corresponding to each frame of the image for synchronous alignment to obtain the time-series positioning coordinate data.
8. The high-precision positioning control method for self-adhesive dot drills based on AI vision according to claim 6, characterized in that: The steps for extracting the positioning deviation features of the ornaments include: Retrieve the corresponding sampling time from the current time-series positioning coordinate data, and calculate the absolute and relative deviations for each sampling time to form a deviation dataset; The deviation dataset is divided into multiple regional sub-datasets based on the visual detection region, and time-series features are extracted from three dimensions: trend, volatility, and suddenness. The spatial features of each regionalized subset are extracted from three dimensions—location characteristics, morphological characteristics, and correlation characteristics—and integrated with the temporal features to obtain the positioning deviation features.
9. The high-precision positioning control method for self-adhesive dot drills based on AI vision according to claim 8, characterized in that: The steps for obtaining the location association parameters include: The analysis object is determined based on the positioning deviation characteristics. The deviation problem that needs to be solved is defined as the top event, and the events related to the basic positioning process parameters are defined as bottom events. For the aforementioned top event, it is decomposed into intermediate events from four aspects: visual detection, motion execution, stress control, and environmental interference, and a first-level causal chain is established. Each intermediate event is further decomposed down until it is associated with the underlying event, the logical gate relationships between different events are determined, and a fault tree logic diagram is drawn. Based on the fault tree logic diagram, find the combination of bottom events that led to the occurrence of the top event to form a minimal cut set, and calculate the structural importance coefficient of each bottom event; The probability of occurrence of each bottom event is calculated, and the partial derivative of the probability of occurrence of the top event with respect to the probability of occurrence of the bottom event is used to obtain the probability importance. The critical importance is then calculated in combination with the structural importance coefficient. The process parameters corresponding to critical importance values greater than a preset importance threshold are used as influencing process parameters, and the corresponding rules are extracted from the fault tree logic diagram to obtain the location association parameters.
10. The high-precision positioning control method for self-adhesive dot drills based on AI vision according to claim 9, characterized in that: The steps for obtaining the positioning correction control parameters include: A parameter mapping table is established based on the positioning association parameters and the positioning adjustment parameters, and the parameter adjustment direction and adjustment magnitude threshold are determined from it; The adjustment weight factors corresponding to the process parameters are calculated based on the critical importance, and the parameters are sorted from high to low to determine the adjustment priority. According to the adjustment priority, the parameter correction amount is calculated in conjunction with the parameter mapping table, and the preliminary adjustment parameters are obtained by summarizing. The positioning correction control parameters are obtained by adding coupling compensation coefficients to the parameter combinations with coupling relationships based on the first-level causal chain and by constraining and verifying them according to the equipment safety threshold.