Automatic workpiece sorting and plating quality grading system for barrel plating production line

Abstract

The invention relates to an automatic workpiece sorting and coating quality grading system for a barrel plating production line, which relates to the technical field of industrial nondestructive testing and automatic sorting, and comprises a differential rolling transmission mechanism, a position encoder, a linear array dynamic imaging unit, a multi-frequency eddy current impedance detection unit, a pneumatic suspension sorting execution unit and a central control unit, the transmission mechanism drives the workpiece to rotate to cooperate with full-coverage scanning; the central control unit utilizes a position encoder to realize space-time alignment of visual magnetic data, extracts a real-time lift-off variable quantity based on image data, compensates and corrects an eddy current original complex impedance signal through an index model, eliminates mechanical jitter interference, and further inverts the thickness and hardness of a coating. And the system finally combines the appearance defect identification result and the inversion parameter to generate a grading instruction, and controls the pneumatic nozzle to complete lossless shunting. The problem of detection errors caused by lift-off interference in dynamic conveying is effectively solved, and high-precision comprehensive evaluation of the appearance and the internal quality of the workpiece is achieved.

Description

Technical Field

[0001] This application relates to the field of industrial non-destructive testing and automated sorting technology, and in particular to an automatic sorting and coating quality grading system for workpieces in a barrel plating production line. Background Technology

[0002] In the electronic components, magnetic materials, and precision machining industries, barrel plating is widely used for surface treatment of tiny workpieces to impart corrosion resistance, wear resistance, or specific electromagnetic properties. With the increasing demands for consistent component quality in high-end manufacturing, full inspection of the plating thickness, hardness, and appearance defects of barrel-plated workpieces has become a critical step in the production process.

[0003] Currently, online inspection of such batches of small workpieces mainly relies on either machine vision inspection or eddy current testing technology. Machine vision technology excels at identifying surface scratches, defects, and other external features, but it cannot penetrate the coating to assess its internal thickness and material hardness. While eddy current testing technology is sensitive to changes in the coating thickness and microstructure of conductive materials, it has significant limitations in dynamic online inspection scenarios. The signal amplitude of eddy current testing is extremely sensitive to the distance between the probe and the measured surface (i.e., the lift-off distance). During high-speed automated conveying, workpieces inevitably experience minor mechanical vibrations or positional shifts. The noise introduced into the eddy current signal by these geometric changes is often far greater than the signal changes caused by minor coating defects, resulting in a severe decrease in the signal-to-noise ratio and a high risk of misjudgment or missed detection.

[0004] To address the lift-off effect, existing technologies typically employ high-precision mechanical clamping devices to secure workpieces. However, this significantly reduces inspection efficiency and fails to meet the cycle time requirements of large-scale production lines. Another common conveying method is linear belt transport, but for cylindrical or annular workpieces, linear transport can only detect the side facing the sensor, failing to achieve full coverage inspection of the circumference and creating a significant blind spot. Furthermore, existing sorting mechanisms often use mechanical forks or pneumatic push rods; this contact-based sorting action can easily cause secondary impacts or scratches on the surface of precision workpieces, affecting the yield of the final product. Therefore, how to simultaneously achieve full-coverage scanning of workpieces within a high-speed, dynamic conveying system, effectively eliminate the interference of mechanical vibration on eddy current detection accuracy, and achieve non-destructive comprehensive evaluation of both appearance and internal quality is a pressing technical challenge for the industry. Summary of the Invention

[0005] The purpose of this application is to solve the technical problems existing in the quality inspection of barrel-plated workpieces, such as low detection coverage, poor signal-to-noise ratio of eddy current detection signals due to workpiece conveying vibration, and inability to simultaneously take into account the comprehensive evaluation of appearance and internal coating quality.

[0006] This invention provides an automatic workpiece sorting and coating quality grading system for a barrel plating production line, including a differential rolling transmission mechanism, a position encoder, a linear array dynamic imaging unit, a multi-frequency eddy current impedance detection unit, a pneumatic suspension sorting execution unit, and a central control unit.

[0007] The differential rolling conveyor mechanism consists of a first conveyor belt assembly and a second conveyor belt assembly arranged in parallel to form a V-shaped or U-shaped conveyor channel for carrying the workpiece to be inspected. The first drive motor and the second drive motor independently drive the two conveyor belt assemblies, and the central control unit controls them to maintain a constant linear speed difference. By using friction, the workpiece to be inspected is driven to rotate around its own axis while translating along the conveyor path, achieving a full-coverage scan of the surface of the workpiece to be inspected.

[0008] The position encoder is rigidly connected to the drive wheel shaft of the differential rolling transmission mechanism, and collects and transmits pulse signals in real time to provide a position synchronization reference.

[0009] The linear array dynamic imaging unit is positioned above the differential rolling transmission mechanism. In conjunction with the telecentric backlight, it performs continuous line scanning on the rotating workpiece to be inspected, generating two-dimensional unfolded image data. This two-dimensional unfolded image data is used for appearance defect identification and extraction of real-time lift-off changes.

[0010] The multi-frequency eddy current impedance detection unit is located on one side of the differential rolling transmission mechanism. It emits a multi-frequency excitation magnetic field containing high-frequency and low-frequency components to the workpiece to be tested and collects the original complex impedance signal reflecting the electromagnetic properties of the workpiece.

[0011] The pneumatic suspension sorting unit is located at the end of the transmission mechanism. According to the grading instructions issued by the central control unit, it uses an array of air nozzles to spray high-pressure airflow, changing the movement trajectory of the workpiece in a non-contact state and guiding it to the corresponding grading collection bin.

[0012] The central control unit is connected to each of the above-mentioned units and is configured to execute the following data processing and control logic:

[0013] Regarding the spatiotemporal mapping of visual magnetometry data, the central control unit calculates the spatiotemporal lag based on the physical installation distance between the linear array dynamic imaging unit and the multi-frequency eddy current impedance detection unit, as well as the pulse resolution of the position encoder. By acquiring the first pulse count value at the trigger acquisition time of the linear array dynamic imaging unit and the second pulse count value at the acquisition time of the multi-frequency eddy current impedance detection unit, and combining it with the pulse offset, the unfolded image data and the original complex impedance signal are indexed and aligned in a unified spatial coordinate system.

[0014] In terms of vision-based lift-off compensation, an edge detection operator is used to extract the upper edge pixel coordinate curve of the workpiece to be inspected from the unfolded image data. The vertical distance difference between this curve and the preset detection baseline is calculated to obtain the real-time lift-off change. To address the lift-off effect interference caused by workpiece vibration in eddy current testing, an exponential decay model is used to construct a compensation correction factor. The correction factor is related to the real-time lift-off change. Satisfying the exponential function relationship ,in This is the sensitivity coefficient related to the probe characteristics. The central control unit uses the formula... For the original complex impedance signal Corrections are made to eliminate the influence of lift-off distance fluctuations on impedance amplitude, obtaining a net impedance signal that characterizes only the workpiece material properties. .

[0015] In terms of multidimensional feature inversion, the central control unit performs feature demodulation on the net impedance signal. The amplitude characteristics of the net impedance signal at the high-frequency excitation frequency are extracted, and the coating thickness is calculated using a pre-calibrated thickness inversion function; the phase characteristics of the net impedance signal at the low-frequency excitation frequency are extracted, and the coating hardness is calculated using a pre-calibrated hardness inversion function.

[0016] In terms of comprehensive quality assessment, the central control unit calculates the statistical mean of the coating thickness and the coating uniformity index, which characterizes the dispersion of the thickness distribution. When the statistical mean of the coating thickness is within a preset thickness range, the coating uniformity index is less than a preset dispersion threshold, and the number of visual appearance defects identified based on the unfolded image data is zero, a grading instruction indicating a qualified product is generated; otherwise, a grading instruction indicating a rework product or a defective product is generated according to the specific defect type.

[0017] In summary, this application includes at least one of the following beneficial technical effects:

[0018] 1. The real-time lift-off variation of the workpiece is extracted using a linear array dynamic imaging unit, and the original complex impedance signal acquired by the multi-frequency eddy current impedance detection unit is compensated and corrected using an exponential decay model. This mechanism can remove the lift-off interference caused by workpiece conveying jitter from the eddy current signal, effectively solving the signal baseline drift problem caused by position fluctuations in dynamic detection environments, and significantly improving the online detection accuracy of micron-level coating thickness and hardness;

[0019] 2. This invention drives the workpiece to rotate around its own axis while translating by controlling the speed difference between two parallel conveyor belts. Combined with continuous scanning by a line scan camera and fixed-point measurement by an eddy current probe, it achieves full-coverage detection of the circumferential surface of cylindrical or annular workpieces, eliminating the detection blind spots present in traditional linear conveying methods. The system can simultaneously acquire multi-dimensional indicators such as appearance defects, coating thickness, hardness, and uniformity, improving the comprehensiveness of quality evaluation.

[0020] 3. This invention employs a pneumatic suspension sorting unit, utilizing high-pressure airflow from an air nozzle assembly to alter the workpiece's trajectory and achieve sorting. Compared to traditional mechanical forks or baffle structures, this non-contact sorting method avoids physical collisions between the actuator and the workpiece surface, preventing secondary scratches or impacts to the precision plating during sorting and ensuring the finished product's appearance quality. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall three-dimensional structure of the automatic workpiece sorting and coating quality grading system for a barrel plating production line according to an embodiment of this application;

[0022] Figure 2 This is a schematic diagram of the automatic workpiece sorting and coating quality grading method in a barrel plating production line according to an embodiment of this application;

[0023] Figure 3 This is a timing diagram illustrating the spatiotemporal mapping and synchronous acquisition of visual magneto-optical data in an embodiment of this application.

[0024] Figure 4 This is a flowchart of the data processing based on visual contour compensation eddy current signals in an embodiment of this application;

[0025] Figure 5 This is a schematic diagram comparing the lift-off compensation effect in the embodiments of this application;

[0026] Figure 6 This is a block diagram of the comprehensive quality judgment logic in the embodiments of this application. Detailed Implementation

[0027] The following is in conjunction with the appendix Figure 1 - Appendix Figure 6 This application will be described in further detail below.

[0028] See attached document Figure 1 - Appendix Figure 2 The present invention provides an automatic sorting and coating quality grading system 100 for workpieces in a barrel plating production line. The system is designed to perform continuous physical parameter detection and quality grading on workpieces after barrel plating.

[0029] The automatic workpiece sorting and coating quality grading system 100 of the barrel plating production line mainly includes: a differential rolling transmission mechanism 200, a linear array dynamic imaging unit 300, a multi-frequency eddy current impedance detection unit 400, a pneumatic suspension sorting execution unit 500, and a central control unit 600.

[0030] The differential rolling transfer mechanism 200 extends along the process transport path of the workpiece, forming the physical transport carrier of the system. The differential rolling transfer mechanism 200 includes a first conveyor belt assembly 210 and a second conveyor belt assembly 220. The first conveyor belt assembly 210 and the second conveyor belt assembly 220 are arranged parallel to each other on a horizontal plane, forming a V-shaped transport channel between them for carrying and transporting the workpiece to be tested.

[0031] The first conveyor belt assembly 210 is connected to a first servo drive motor 211, and the second conveyor belt assembly 220 is connected to a second servo drive motor 221. The first servo drive motor 211 and the second servo drive motor 221 are electrically connected to the central control unit 600. The central control unit 600 outputs different speed control commands to maintain a preset speed difference between the linear speed of the first conveyor belt assembly 210 and the linear speed of the second conveyor belt assembly 220. This speed difference drives the workpiece located in the V-shaped transmission channel to move forward along the conveying path while continuously rotating around its own axis, thereby causing the outer surface of the workpiece to unfold in a spiral shape during transmission.

[0032] A linear array dynamic imaging unit 300 is mounted above the differential rolling transmission mechanism 200. The linear array dynamic imaging unit 300 includes a linear array industrial camera 310 and a high-frequency flash source 320. The lens optical axis of the linear array industrial camera 310 is vertically downward and aligned with the center line of the V-shaped transmission channel, and its field of view covers the lateral width of the workpiece. The high-frequency flash source 320 provides illumination to the imaging area of ​​the linear array industrial camera 310. The linear array industrial camera 310 is connected to the central control unit 600 via a high-speed data bus for transmitting the acquired scanned image data of the workpiece surface to the central control unit 600.

[0033] See attached document Figure 4 A multi-frequency eddy current impedance detection unit 400 is positioned downstream of the linear array dynamic imaging unit 300 along the workpiece conveying direction. The multi-frequency eddy current impedance detection unit 400 includes a non-contact multi-frequency eddy current probe 410 and a probe adjustment bracket 420. The non-contact multi-frequency eddy current probe 410 is fixed directly above the V-shaped transmission channel via the probe adjustment bracket 420. A preset physical gap is maintained between the sensing end face of the non-contact multi-frequency eddy current probe 410 and the surface of the workpiece passing through this location. The non-contact multi-frequency eddy current probe 410 is electrically connected to the central control unit 600 and is used to transmit multi-frequency excitation signals to the workpiece and acquire the complex impedance signals fed back by the workpiece.

[0034] A pneumatic suspension sorting execution unit 500 is located at the output end of the differential rolling conveyor mechanism 200. The pneumatic suspension sorting execution unit 500 includes a position tracking sensor 510, multiple pneumatic nozzles 520, and multiple graded collection bins 530. The multiple pneumatic nozzles 520 are arranged sequentially along the conveying path, each corresponding to a different graded collection bin 530. The pneumatic nozzles 520 are connected to the central control unit 600 via high-speed solenoid valves. The graded collection bins 530 include at least a qualified product bin, a rework product bin, and a defective product bin.

[0035] The central control unit 600, as the core processing and control center of the system, integrates a synchronization triggering module 610, a lift-off compensation calculation module 620, and a quality comprehensive judgment module 630. The central control unit 600 receives the position encoding signal from the differential rolling transmission mechanism 200 and, based on the signal, synchronously triggers the image acquisition of the linear array dynamic imaging unit 300 and the signal sampling of the multi-frequency eddy current impedance detection unit 400.

[0036] The following describes the overall workflow of the automatic workpiece sorting and coating quality grading system 100 in the barrel plating production line, based on the above structure:

[0037] After being sorted by the feeding device, the workpieces to be inspected enter the V-shaped transmission channel of the differential rolling transmission mechanism 200 in a single row. Driven by the speed difference between the first conveyor belt assembly 210 and the second conveyor belt assembly 220, the workpieces establish a composite motion posture of translation and rotation.

[0038] When the workpiece passes through the field of view of the linear array dynamic imaging unit 300, the linear array industrial camera 310 performs continuous line scanning on the rotating workpiece surface, generating unfolded image data of the workpiece surface. The central control unit 600 receives the unfolded image data and performs real-time analysis on the image edges, extracting the vertical jump contour data of the workpiece during the rolling process. This data characterizes the real-time lift-off change of the workpiece surface.

[0039] Subsequently, the workpiece is transferred to the area below the multi-frequency eddy current impedance detection unit 400. The non-contact multi-frequency eddy current probe 410 acquires the original complex impedance signal of the workpiece. The central control unit 600 uses a pre-established spatiotemporal mapping relationship to align the real-time lift-off change acquired by the linear array dynamic imaging unit 300 with the original complex impedance signal acquired by the multi-frequency eddy current impedance detection unit 400 on the time axis.

[0040] The lift-off compensation calculation module 620 uses the real-time lift-off change extracted visually to correct the original complex impedance signal, removing signal fluctuations caused by workpiece mechanical runout, and obtaining a net impedance signal reflecting the material properties of the workpiece coating. The quality comprehensive judgment module 630 generates sorting instructions based on the appearance defect features of the unfolded image data and the coating thickness and hardness features obtained by inverting the net impedance signal, according to preset grading standards.

[0041] When the workpiece arrives at the pneumatic suspension sorting execution unit 500, the central control unit 600 controls the corresponding pneumatic nozzle 520 to open according to the sorting instruction, and uses airflow to push the workpiece into the corresponding grade collection box 530 to complete the detection and grading process.

[0042] In this embodiment, the differential rolling transmission mechanism 200 is the physical foundation of the entire system, responsible for realizing the combined motion of "translation" and "rotation" of the workpiece under test within the detection area. This mechanism is designed to overcome the limitation of traditional single-sided static inspection, which cannot cover all surface defects of barrel-plated parts.

[0043] The differential rolling transmission mechanism 200 mainly consists of a first conveyor belt assembly 210, a second conveyor belt assembly 220, a first servo drive motor 211, and a second servo drive motor 221.

[0044] The first conveyor belt assembly 210 and the second conveyor belt assembly 220 are arranged parallel to each other within the detection area, forming a V-shaped conveying channel between them. The width of this channel is slightly smaller than the nominal outer diameter of the workpiece being tested, thus ensuring stable support of the workpiece and guaranteeing contact between the workpiece and the transmission surfaces of both conveyor belts. The conveyor belt surfaces are made of a high-friction coefficient material to ensure that the driving force is effectively converted into the rotational torque of the workpiece, reducing slippage.

[0045] The first servo drive motor 211 drives the first conveyor belt assembly 210 at a preset linear speed via a transmission structure. Motion; the second servo drive motor 221 drives the second conveyor belt assembly 220 at a preset linear speed via another transmission structure. Movement. The central control unit 600 outputs independent control commands to the first servo drive motor 2111 and the second servo drive motor 221 to ensure... .

[0046] Within the V-shaped channel, the workpiece is subjected to frictional forces from the two conveyor belts. Since the frictional forces on both sides are equal in magnitude and opposite in direction, a couple is generated that drives the workpiece to rotate. The workpiece's motion can be decomposed into translational motion along the conveying direction and rotational motion about its own axis.

[0047] Translational speed of the workpiece along the conveying path Determined by the average linear speed of the conveyor belts on both sides:

[0048] ;

[0049] angular velocity of the workpiece rotating about its own axis It depends on the speed difference between the two conveyor belts and the diameter of the workpiece. :

[0050] ;

[0051] in,

[0052] Linear velocity of the first conveyor belt assembly 210;

[0053] Linear speed of the second conveyor belt assembly 220;

[0054] : The nominal outer diameter of the workpiece to be measured.

[0055] The central control unit 600 adjusts the input based on the diameter of the workpiece of different specifications. and the set minimum translation speed Adjust in real time and The adjustment process ensures that the workpiece passes through the effective detection length of the linear array dynamic imaging unit 300. At that time, at least one complete 360° rotation must be completed. The rotation coverage constraint is as follows:

[0056] ;

[0057] The central control unit 600 is equipped with a speed closed-loop control module. The first servo drive motor 211 and the second servo drive motor 221 integrate high-precision rotary encoders (not shown). These encoders provide real-time feedback of the motor's rotational speed to the central control unit 600. The central control unit 600 uses PID and other control algorithms to precisely correct the speed, ensuring that the deviation between the actual linear speed and the set linear speed is within the allowable range, thereby guaranteeing the workpiece's rotational angular velocity. Its stability and controllability provide stable motion parameters for subsequent image unfolding and stitching in linear array imaging.

[0058] The differential rolling transmission mechanism 200 is designed to ensure that the outer surface of the workpiece maintains a constant angular velocity. The relative top detection sensors (linear array dynamic imaging unit 300 and multi-frequency eddy current impedance detection unit 400) are continuously deployed, realizing the physical posture preparation for 360° full surface detection of the workpiece.

[0059] In this embodiment, the linear array dynamic imaging unit 300 is fixedly mounted above the V-shaped transmission channel of the differential rolling transmission mechanism 200. The linear array dynamic imaging unit 300 mainly includes a linear array industrial camera 310 and a high-frequency strobe source 320. The optical axis of the linear array industrial camera 310 is perpendicular to the conveying plane of the workpiece, and the center line of the optical axis passes through the geometric center of the V-shaped transmission channel. The linear array industrial camera 310 is equipped with a telecentric lens to eliminate perspective distortion of the workpiece within the depth of field, ensuring consistency in the imaging ratio of workpieces with different diameters. The high-frequency strobe source 320 is symmetrically arranged on both sides of the linear array industrial camera 310, with the beam converging in a linear field of view area below the camera. This high-frequency strobe source 320 uses a strip LED array, and its emission frequency is synchronized with the line scanning frequency of the linear array industrial camera 310 to provide high-brightness instantaneous illumination during high-speed workpiece rotation, freezing motion blur.

[0060] The multi-frequency eddy current impedance detection unit 400 is positioned downstream of the linear array dynamic imaging unit 300 along the workpiece conveying direction. The core component of the multi-frequency eddy current impedance detection unit 400 is a non-contact multi-frequency eddy current probe 410. This non-contact multi-frequency eddy current probe 410 is suspended above the V-shaped transmission channel via a precision probe adjustment bracket 420. The probe adjustment bracket 420 has a vertical (Z-axis) fine-tuning function, used to set the physical gap between the sensing end face of the non-contact multi-frequency eddy current probe 410 and the upper surface of the workpiece passing through that position; this gap is defined as the reference lift-off distance. Baseline lift-off distance The setting value is usually between 0.5mm and 2.0mm, and the specific value is selected according to the range characteristics of the probe.

[0061] To achieve precise spatiotemporal alignment of the visual magnetometry data, the line scan imaging center of the linear industrial camera 310 and the detection center of the non-contact multi-frequency eddy current probe 410 maintain a fixed longitudinal mounting distance in the workpiece conveying direction. The longitudinal installation spacing These are key physical parameters for the system's timing synchronization calculations; their values ​​are precisely calibrated using mechanical installation dimensions and pre-stored in the central control unit 600. (The last sentence appears to be incomplete and unrelated to the preceding text. It likely refers to a process where the workpiece moves at a translational speed.) During transmission, this spacing creates a fixed time lag, providing a physical reference for subsequent data fusion algorithms.

[0062] The pneumatic suspension sorting execution unit 500 is located at the end output area of ​​the differential rolling conveyor 200. This unit mainly consists of a position tracking sensor 510, multiple sets of high-speed pneumatic nozzles 520, and corresponding graded collection bins 530. The position tracking sensor 510 (such as a fiber optic sensor or a laser beam sensor) is installed at the end of the detection area to detect the moment when the workpiece leaves the detection area and enters the sorting area, thereby resetting the workpiece's position tracking count.

[0063] Multiple sets of high-speed pneumatic nozzles 520 are linearly arranged on one side of the conveying path, with each set of nozzles corresponding to a specific graded collection bin 530. The graded collection bins 530 typically include bins for qualified products, rework products, and defective products, used to collect workpieces that have passed quality assessment, require replating repair, and are scrapped, respectively. The high-speed pneumatic nozzles 520 are connected to the factory's compressed air source via fast-response solenoid valves. The nozzle outlets face the side of the V-shaped conveying channel and are designed with a streamlined air guiding structure.

[0064] When the central control unit 600 issues a sorting command based on the detection results, the high-speed pneumatic nozzle 520 at the corresponding position ejects a high-pressure airflow the instant the workpiece reaches the nozzle's effective area. This airflow does not directly impact the workpiece, but utilizes the Coanda Effect or lateral momentum exchange of the airflow to generate a low-pressure zone or lateral thrust on the side of the workpiece, causing the workpiece to be controlled to leave the predetermined trajectory of the V-shaped conveying channel and fall parabolically into the corresponding graded collection bin 530. This non-contact pneumatic sorting method avoids the secondary scratches or bumps that may be caused to the workpiece plating layer by traditional mechanical lever sorting, and is particularly suitable for barrel-plated parts with extremely high appearance quality requirements. The specific selection of pneumatic components and the air circuit connection method are standard technical means in the field of industrial automation and will not be detailed here.

[0065] See attached document Figure 3 , Figure 3 This is a timing diagram illustrating the spatiotemporal mapping and synchronous acquisition of magnetic resonance (MRRT) data according to an embodiment of the present invention, which includes the processing logic for spatiotemporal mapping of MRRT data. In this embodiment, the central control unit 600 executes the spatiotemporal mapping mechanism of MRRT data through an integrated synchronization trigger module 610. This mechanism aims to solve the problem of asynchronous data acquisition caused by the different physical installation positions of the linear array dynamic imaging unit 300 and the multi-frequency eddy current impedance detection unit 400, ensuring that during subsequent algorithm processing, specific pixels on the visual image and specific sampling points of the eddy current impedance signal correspond to the same physical location on the workpiece surface.

[0066] The synchronous trigger module 610 is hardwired to the rotary encoder of the servo drive motor in the differential rolling transmission mechanism 200. When the workpiece moves on the conveyor belt, the rotary encoder outputs a real-time pulse signal, which characterizes the real-time displacement of the workpiece along the conveying direction. The central control unit 600 internally maintains a global system time axis. And establish a global spatial coordinate system based on encoder pulses.

[0067] The spatiotemporal mapping and alignment process of visual magnetometry data specifically includes the following steps:

[0068] S210 establishes a global spatiotemporal reference. The synchronous trigger module 610 receives the pulse sequence from the encoder and calculates the current real-time translational speed of the workpiece based on the diameter of the drive wheel of the conveyor belt and the reduction ratio. Simultaneously, the synchronous trigger module 610 generates trigger signals at a preset spatial resolution (e.g., 0.1 mm per movement), triggering the linear scan industrial camera 310 to perform one line of image exposure acquisition and the non-contact multi-frequency eddy current probe 410 to perform one impedance data sampling. This position-based triggering mode eliminates the influence of minute fluctuations in conveying speed on the sampling density, ensuring consistency between visual and electromagnetic data in spatial sampling rate.

[0069] S220 calculates the time lag between sensors. The system has pre-stored physical installation parameters. This parameter is defined as the straight-line distance in the workpiece conveying direction from the optical axis centerline of the line scan industrial camera 310 to the detection centerline of the non-contact multi-frequency eddy current probe 410. It is based on the real-time calculated workpiece translation speed. The system calculates the time lag required for the same point on the workpiece to move from the visual inspection position to the eddy current inspection position. :

[0070] ;

[0071] in:

[0072] : Indicates the time lag in the acquisition of visual magnetometry data, in seconds (s);

[0073] : Indicates the physical longitudinal distance between the visual inspection center and the eddy current inspection center, in millimeters (mm).

[0074] : Indicates the real-time translational speed of the workpiece along the conveying direction, in millimeters per second (mm / s).

[0075] S230, constructing a synchronous fusion dataset. The central control unit 600 allocates a first-in, first-out (FIFO) data buffer in memory. Image data acquired by the linear scan industrial camera 310 is stored in the image buffer in chronological order, and complex impedance signals acquired by the non-contact multi-frequency eddy current probe 410 are stored in the impedance buffer. When the system is at the current moment... The latest eddy current impedance signal was acquired. At the same time, the synchronization trigger module 610 does not directly use the image data at the current moment, but instead uses the calculated time lag. The index is traced back through the historical data in the image buffer to extract the time. Corresponding image frame data .

[0076] The system binds the back-extracted image data with the impedance data at the current moment to generate a visual-magnetic fusion dataset. :

[0077] ;

[0078] in:

[0079] : Indicates time The generated aligned and merged data packets;

[0080] : Indicates a historical moment The collected visual image data of the workpiece surface contains the outline and appearance information of the workpiece at that physical location;

[0081] : Indicates the current moment The original complex impedance signal of the workpiece at the same physical location was collected;

[0082] : Represents the pixel coordinate index of the line scan camera in the horizontal scanning direction.

[0083] Through the aforementioned spatiotemporal mapping mechanism, the system achieves physical location normalization for heterogeneous data. This enables the system to accurately utilize the contour jitter extracted from the visual image (i.e., the physical location within the image) in subsequent processing steps. (Geometric shape at time) to correct the eddy current probe at The impedance signal, distorted by the lift-off effect, is constantly acquired, thus providing a reliable data foundation for high-precision coating quality inversion. The specific software implementation of data buffer overflow management and memory addressing is a standard technique in the field of computer data structures, and those skilled in the art can configure it according to actual hardware resources.

[0084] See attached document Figure 3 , Figure 3 This is a schematic diagram of a real-time lift-off extraction process based on visual contours according to an embodiment of the present invention.

[0085] In this embodiment, the central control unit 600 executes a specific image processing and geometric calculation logic designed to accurately quantify the minute vertical displacement of the workpiece from two-dimensional image data. This displacement is typically caused by mechanical vibrations during differential rolling, the workpiece's own geometric edges (such as the edges of a hexagonal nut), or minor unevenness in the conveyor belt. For eddy current testing, minute changes in the distance between the probe and the measured surface (i.e., the lift-off distance) cause significant nonlinear drift in the impedance signal; therefore, accurately extracting this distance change is a prerequisite for subsequent signal compensation.

[0086] The real-time lift-off extraction process based on visual contours specifically includes the following steps:

[0087] S310 performs sub-pixel level edge contour extraction. The central control unit 600 retrieves the spatiotemporally aligned image frame data. ,in The system first defines a region of interest (ROI) for the image frame, covering the upper edge of the workpiece. Within this ROI, edge detection operators (such as the Canny or Sobel operators) are applied for coarse edge localization to obtain pixel-level edge coordinates. To meet the micrometer-level accuracy requirements of eddy current compensation, the system further performs Gaussian surface fitting or polynomial interpolation on the gray-level gradient distribution near the coarse edge to calculate the vertical coordinates of the workpiece's upper edge with sub-pixel accuracy. This coordinate represents the height position of the workpiece's upper surface in the image coordinate system at the moment of visual acquisition.

[0088] S320, construct the mapping transformation from image coordinates to physical space. The system pre-stores the optical calibration parameters of the linear scan industrial camera 310, including the camera's horizontal and vertical magnification. In this embodiment, the vertical pixel equivalent is defined. (Unit: mm / pixel), representing the actual physical space dimension corresponding to the height of one pixel in the image. The instantaneous physical height of the workpiece's upper surface is calculated based on the conveyor reference plane (i.e., the lowest point or virtual center line of the V-shaped conveyor belt). :

[0089] ;

[0090] in:

[0091] : Indicates time The instantaneous physical height of the upper surface of the workpiece, in millimeters (mm);

[0092] : Represents the vertical coordinates of the sub-pixel edge extracted from the image frame, in pixels.

[0093] : Indicates the vertical pixel equivalent calibration coefficient of the camera, in millimeters per pixel (mm / pixel).

[0094] : Represents the constant of fixed height deviation of the origin of the image coordinate system relative to the transport reference plane, in millimeters (mm).

[0095] S330, calculate the instantaneous lift-off distance. The non-contact multi-frequency eddy current probe 410 is known to be installed at a fixed height above the transport reference plane. Since the visual magnetosurgical data has been spatiotemporally aligned, the time... Workpiece geometric height That is, time. The height of the workpiece located below the eddy current probe. The system calculates the absolute lift-off distance between the sensing end face of the eddy current probe and the workpiece surface at this moment. :

[0096]

[0097] in:

[0098] : Indicates time Real-time absolute distance between the eddy current probe and the workpiece surface;

[0099] : Indicates the physical installation height of the eddy current probe sensing end face from the transmission reference surface. This value is a fixed constant.

[0100] S340, Extract relative liftoff fluctuation. To adapt to the subsequent impedance compensation model, the system needs to obtain the liftoff change relative to the standard operating point. The optimal coupling distance during system calibration is set as the reference liftoff value. The system calculates the real-time lift-off fluctuation. :

[0101] ;

[0102] in:

[0103] : Indicates time The lifting fluctuation value is positive, indicating an increase in distance (workpiece sinking), and negative, indicating a decrease in distance (workpiece jumping upward).

[0104] : Indicates the system's preset standard lift-off distance, which is usually set to the center value of the linear range of the eddy current probe.

[0105] Through the above steps, the system transforms visual image data, originally used only for detecting appearance defects, into high-precision geometric data characterizing the dynamic position of the workpiece. The data stream is transmitted in real time to the lift-off compensation calculation module 620 as an external reference variable to correct the eddy current impedance signal, thereby realizing cross-modal data coupling between optical and electromagnetic measurements at the physical level. The specific code implementation of the edge detection operator and the camera calibration process are standard techniques in the field of computer vision. Those skilled in the art can select appropriate algorithm libraries for deployment based on actual working conditions, and will not be detailed here.

[0106] See attached document Figure 5 , Figure 5 This is a schematic diagram comparing the lift-off compensation effect in an embodiment of this application. In this embodiment, the central control unit 600 uses the lift-off compensation calculation module 620 to perform the core signal cleaning task. The algorithm aims to establish a mathematical relationship between mechanical geometric changes and electromagnetic impedance changes, and through mathematical decoupling, separate the true impedance component determined only by the properties of the coating material from the mixed signal.

[0107] The dynamic compensation process for multi-frequency eddy current impedance signals specifically includes the following steps:

[0108] S410, construct the original impedance signal vector. The central control unit 600 reads the original complex impedance signal that has been spatiotemporally aligned with the visual data. For any specific frequency point in multi-frequency eddy current detection... The signal output by the eddy current probe coil is represented as a two-dimensional vector in the complex plane. System definition time. The original impedance vector for:

[0109] ;

[0110] in:

[0111] : Represents the real part of the measured impedance, i.e. the resistive component, which is mainly affected by eddy current heat loss and material conductivity;

[0112] : Represents the imaginary part of the measured impedance, i.e., the reactance component, which is mainly affected by the coil inductance and the permeability of the material.

[0113] S420, establish the compensation transfer function based on lift-off feedback. According to electromagnetic field theory, when the distance between the probe and the conductor surface changes slightly, the trajectory of the impedance signal on the complex plane presents a nonlinear "lift-off curve". Within a small range of fluctuations (e.g., ±0.5mm ± 0.5mm), this change can be approximately decomposed into an exponential decay of the amplitude and a rotational shift of the phase. The lift-off compensation calculation module 620 calculates the real-time lift-off fluctuation amount input in step S340. Construct a dynamic compensation matrix :

[0114] ;

[0115] in:

[0116] : Indicates the amplitude gain compensation term, used to offset the signal attenuation caused by increased lift-off;

[0117] : Represents the amplitude attenuation sensitivity coefficient, in mm −1 ;

[0118] : Represents the phase rotation sensitivity coefficient, with units of rad / mm;

[0119] Matrix part This represents a two-dimensional rotation matrix used to rotate the deflected impedance vector back to the reference phase angle.

[0120] The above coefficients and It is related to the excitation frequency Strongly correlated functions. In this embodiment, the system pre-stores functions for high frequencies. and low frequency The different coefficient tables, i.e. and To correspond to the electromagnetic field penetration depth characteristics at different frequencies.

[0121] S430 performs impedance vector correction calculation. The lift-off compensation calculation module 620 will use the dynamic compensation matrix... Acting on the original impedance vector The corrected net impedance vector is calculated using matrix multiplication. :

[0122] ;

[0123] Expanding the calculation, we get:

[0124] ;

[0125] in:

[0126] : Represents the compensated resistance component;

[0127] : Represents the compensated reactance component.

[0128] S440, output net signal sequence. The system repeats the above calculation for each sampling point to generate a continuous net impedance signal sequence. This net impedance signal sequence is physically equivalent to the signal generated at an ideal constant lift-off distance (i.e., ...). The signal is measured under these conditions. Through this step, the mechanical runout interference caused by workpiece rolling is removed from the signal, retaining only the electromagnetic response caused by changes in the coating thickness on the workpiece surface, differences in substrate hardness, and changes in the material's microstructure. This significantly improves the signal-to-noise ratio and accuracy of subsequent physical parameter inversion. The specific software implementation of matrix operations can be accomplished using an FPGA hardware accelerator or the floating-point unit in a DSP digital signal processor, which is a conventional implementation method in the field of digital signal processing.

[0129] See attached document Figure 6 , Figure 6 This is a logic block diagram for comprehensive quality assessment according to an embodiment of the present invention. In this embodiment, the central control unit 600 uses the comprehensive quality assessment module 630 to calculate parameters of the compensated net signal and introduces statistical indicators to quantitatively evaluate the distribution characteristics unique to the barrel plating process.

[0130] The multidimensional feature inversion and coating uniformity evaluation process specifically includes the following steps:

[0131] S510 performs physical parameter inversion. Based on a pre-established "impedance-physical property" calibration database, the system performs the corrected net impedance signal... The mapping is to specific physical parameters.

[0132] Regarding the selection of excitation frequency, a high-frequency excitation frequency is set. The skin depth is less than the preset minimum coating thickness to ensure that the eddy currents are mainly distributed within the coating; the low-frequency excitation frequency is set. The skin depth is greater than the maximum coating thickness to ensure that its magnetic field can penetrate into the substrate material.

[0133] The system extracts the net impedance amplitude under high-frequency excitation. The polynomial fitting function generated by the least squares method Calculate instantaneous coating thickness :

[0134] ;

[0135] Simultaneously, the system extracts the net impedance phase angle under low-frequency excitation. Through a pre-calibrated linear regression or nonlinear mapping function Estimating the equivalent hardness value of the workpiece :

[0136] ;

[0137] in, The thickness fitting coefficient is obtained by measuring a set of standard samples with different known thicknesses and performing curve fitting. Since the workpiece is rotating, the above calculation process generates a thickness distribution sequence unfolding along the circumference of the workpiece. and hardness distribution sequence .

[0138] S520, Calculate the coating uniformity index (CUI). To address the coating distribution differences caused by uneven tumbling in the barrel plating process, this embodiment calculates the standard deviation of the thickness distribution sequence as the coating uniformity index. The system selects the position corresponding to the complete rotation of the workpiece. There are 1 sampling points, among which The number of pulses per revolution fed back by the encoder is determined by the following formula:

[0139] ;

[0140] in, The coating uniformity index For the first The instantaneous thickness value at each sampling point. This represents the average coating thickness during one rotation of the workpiece. Additionally, the system simultaneously calculates the average hardness value during one rotation. .

[0141] S530 performs a comprehensive grading assessment. The quality comprehensive assessment module 630 combines the visual appearance inspection results provided by the linear array imaging unit. (set up This indicates that the appearance is acceptable. Indicates a minor defect. (Indicating a serious defect), generating hierarchical instructions based on preset logic. :

[0142] ;

[0143] in, The thickness is the acceptable threshold. The minimum thickness threshold for allowing rework. This is the lower limit threshold for hardness. This is the uniformity threshold.

[0144] S540 outputs control signals. Judgment result. The signal is converted into a switching signal and sent to the pneumatic suspension sorting execution unit 500. If the workpiece is determined to be a qualified product, it continues to be conveyed to the end along the transport channel; if it is determined to be a rework or defective product, the system triggers the nozzle to activate when the workpiece reaches the corresponding position. Simultaneously, the system records... Data and defect types are used to monitor the stability of the barrel plating process.

[0145] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Identical components are represented by the same reference numerals. Therefore, all equivalent changes made to the structure, shape, and principle of this application should be included within the scope of protection of this application.

Claims

1. An automatic workpiece sorting and coating quality grading system for a barrel plating production line, characterized in that, include: A differential rolling transmission mechanism is used to carry the workpiece to be inspected and drive the workpiece to be inspected to translate along the conveying path and rotate around its own axis by setting a conveying speed difference. A position encoder is mounted on the differential rolling transmission mechanism to collect and transmit pulse signals in real time to establish a position coordinate reference. A linear array dynamic imaging unit is disposed above the differential rolling transmission mechanism and is used to continuously scan the rotating workpiece to be inspected and generate unfolded image data. A multi-frequency eddy current impedance detection unit is disposed on one side of the differential rolling transmission mechanism and is used to emit a multi-frequency excitation magnetic field to the workpiece to be tested and acquire the original complex impedance signal. A pneumatic suspension sorting execution unit is located at the end of the differential rolling transmission mechanism and is used to sort the workpieces to be inspected according to the grading instructions. as well as The central control unit is communicatively connected to the position encoder, the linear array dynamic imaging unit, the multi-frequency eddy current impedance detection unit, and the pneumatic suspension sorting execution unit, respectively. The central control unit is configured to: extract the real-time lift-off change of the workpiece to be inspected relative to the inspection reference surface based on the unfolded image data; use the real-time lift-off change to compensate and correct the original complex impedance signal to obtain a net impedance signal; invert physical parameters based on the net impedance signal; and generate the grading instruction based on the physical parameters.

2. The automatic workpiece sorting and coating quality grading system for a barrel plating production line according to claim 1, characterized in that, The differential rolling transmission mechanism includes: The first conveyor belt assembly and the second conveyor belt assembly are arranged in parallel to form a V-shaped or U-shaped transmission channel for carrying the workpiece to be inspected; The first drive motor and the second drive motor independently drive the first conveyor belt assembly and the second conveyor belt assembly, respectively. The central control unit controls the rotational speeds of the first drive motor and the second drive motor, ensuring a constant speed difference between the first linear speed of the first conveyor belt assembly and the second linear speed of the second conveyor belt assembly. This allows the workpiece to be inspected to rotate while moving forward with the conveyor belt, driven by friction.

3. The automatic workpiece sorting and coating quality grading system for a barrel plating production line according to claim 1, characterized in that, The central control unit includes: The visual magneto-magnetic data spatiotemporal mapping module is used to receive the pulse signal fed back by the position encoder, calculate the spatiotemporal lag based on the physical installation distance between the linear array dynamic imaging unit and the multi-frequency eddy current impedance detection unit, and align the real-time lift-off change with the original complex impedance signal on the time axis. The lift-off compensation calculation module is used to construct a compensation correction factor based on the real-time lift-off change, and use the compensation correction factor to remove the lift-off effect interference from the original complex impedance signal, and output the net impedance signal. The multidimensional feature inversion module is used to demodulate the net impedance signal, extract the high-frequency and low-frequency components respectively, and map them to the physical parameters.

4. The automatic workpiece sorting and coating quality grading system for a barrel plating production line according to claim 3, characterized in that, The linear array dynamic imaging unit includes a high frame rate linear array camera and a telecentric backlight. The central control unit is configured to perform the following steps to extract the real-time lift-off change: The linear array dynamic imaging unit is controlled to acquire the edge contour image of the workpiece to be detected. The upper edge pixel coordinate curve of the edge contour image is extracted using an edge detection operator; Calculate the vertical distance difference between the upper edge pixel coordinate curve and the preset detection baseline, and use the vertical distance difference as the real-time lift-off change.

5. The automatic workpiece sorting and coating quality grading system for a barrel plating production line according to claim 3, characterized in that, The lift-off compensation calculation module is configured to construct the compensation correction factor using an exponential decay model. The compensation correction factor and the real-time lift-off change satisfy an exponential function relationship. The lift-off compensation calculation module divides the original complex impedance signal by the compensation correction factor, or multiplies the original complex impedance signal by the reciprocal of the compensation correction factor, to obtain the net impedance signal, such that the amplitude of the net impedance signal is independent of the spatial position fluctuation of the workpiece to be tested.

6. The automatic workpiece sorting and coating quality grading system for a barrel plating production line according to claim 3, characterized in that, The multi-frequency eddy current impedance detection unit is equipped with a high-frequency excitation frequency and a low-frequency excitation frequency. The multidimensional feature inversion module is specifically configured as follows: Extract the amplitude characteristics of the net impedance signal at the high-frequency excitation frequency, and calculate the coating thickness of the workpiece to be tested using a preset thickness inversion function; The phase characteristics of the net impedance signal at the low-frequency excitation frequency are extracted, and the coating hardness of the workpiece to be tested is calculated using a preset hardness inversion function. The coating thickness and the coating hardness are used as the physical parameters.

7. The automatic workpiece sorting and coating quality grading system for a barrel plating production line according to claim 6, characterized in that, The central control unit also includes a comprehensive quality assessment module; The comprehensive quality assessment module is configured as follows: Receive the unfolded image data generated by the linear array dynamic imaging unit and identify visual appearance defects on the surface of the workpiece to be inspected; Calculate the statistical mean of the coating thickness and the coating uniformity index; When the statistical mean of the coating thickness is within a preset thickness range, the coating uniformity index is less than a preset discrete threshold, and the number of visual defects is zero, a grading instruction indicating that the product is qualified is generated.

8. The automatic workpiece sorting and coating quality grading system for a barrel plating production line according to claim 1, characterized in that, The pneumatic suspension sorting execution unit includes: The sorting channel is connected to the output end of the differential rolling transmission mechanism; An array of air nozzles is arranged along the sidewall of the sorting channel, and each air nozzle is connected to an independent solenoid valve. According to the grading instructions and the signals from the position encoder, when the workpiece to be inspected reaches the designated position, the central control unit controls the corresponding solenoid valve to open, and sprays out a high-pressure airflow to blow the workpiece to be inspected away from the original trajectory of the sorting channel.

9. The automatic workpiece sorting and coating quality grading system for a barrel plating production line according to claim 8, characterized in that, The system also includes multiple graded collection bins; The graded collection bins are respectively positioned below the different material drop trajectories of the pneumatic suspension sorting execution unit; The graded collection bins include at least: a defective bin for collecting workpieces whose coating thickness or hardness does not meet the standards, a rework bin for collecting workpieces with visual defects, and a qualified bin for collecting workpieces whose various indicators meet the standards.

10. The automatic workpiece sorting and coating quality grading system for a barrel plating production line according to claim 3, characterized in that, The position encoder is an incremental rotary encoder, which is rigidly connected to the drive wheel shaft of the differential rolling transmission mechanism; The spatiotemporal mapping module for visual magneto data is specifically configured as follows: Obtain the first pulse count value at the moment the linear array dynamic imaging unit triggers acquisition; Obtain the pulse offset corresponding to the physical installation spacing; At the moment when the multi-frequency eddy current impedance detection unit acquires data, the current second pulse count value is read; Based on the first pulse count value, the second pulse count value, and the pulse offset, a one-to-one correspondence index relationship is established between the unfolded image data and the original complex impedance signal.

Images