An image recognition-based FPC flexible circuit board intelligent production management system

Abstract

The present application relates to production management technical field, especially to a kind of FPC flexible circuit board intelligent production management system based on image recognition.The method includes the following steps: by positioning acquisition module, FPC is fixed and original detection image is obtained;By curvature calibration module, edge offset is identified and board body pose calibration instruction is generated to complete rectification;Subsequently, regional resampling module carries out local high-precision collection to pad, circuit and mounting area to form structured detection data;Finally, quality evaluation module extracts detection field to generate production quality result and record to storage system, realize automatic detection calibration and quality tracking process.The present application drives automatic positioning calibration by image recognition, regional high-precision detection and quality data closed-loop management, realizes the high-precision alignment of flexible circuit board production process, adaptive detection and intelligent quality control, to significantly improve production efficiency and yield.

Description

Technical Field

[0001] This invention relates to the field of production management technology, and in particular to an intelligent production management system for FPC flexible circuit boards based on image recognition. Background Technology

[0002] In the current production process of flexible printed circuit boards (FPCs), the high flexibility and susceptibility to deformation of the boards, as well as their significant susceptibility to temperature and mechanical stress, easily lead to quality problems such as positional misalignment, trace breakage, pad deformation, and misalignment during the mounting, soldering, transfer, and inspection stages. Traditional production methods mainly rely on manual resetting, low-precision global imaging, or semi-automatic inspection equipment for calibration and quality inspection. However, the inspection accuracy is greatly affected by the operator's experience, and it is impossible to achieve real-time identification and closed-loop correction of dynamic positional misalignment, which easily leads to problems such as misjudgment, missed detection, and high rework rates.

[0003] Meanwhile, existing processes mostly involve whole-board inspection, failing to segment and differentiate inspections of pad areas, trace paths, and component mounting areas. This results in low identification rates for minute defects and an inability to quickly locate and analyze high-density circuit boards and tiny pads. Furthermore, inspection data is not structured and linked to the production process, lacking traceability and dynamic quality monitoring capabilities, making it difficult to meet the requirements of intelligent manufacturing for high-precision inspection, real-time calibration, and closed-loop management of production data. Summary of the Invention

[0004] Therefore, it is necessary to provide an image recognition-based intelligent production management system for FPC flexible circuit boards to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, an intelligent production management system for FPC flexible circuit boards based on image recognition includes the following modules: The positioning and acquisition module is used to fix the FPC flexible circuit board to be produced in the adjustable clamping structure of the positioning fixture, and to trigger the optical acquisition module to start the initial image acquisition to obtain the original detection image. The curvature calibration module is used to identify the edge offset magnitude of the original detection image and use the edge offset magnitude to generate plate pose calibration commands for manipulating mechanical transmission execution components. The region resampling module is used to perform local image resampling on the pad area, trace path area and component mounting area of ​​the FPC flexible circuit board in response to the completion of the board pose calibration command, so as to obtain structured data for quality inspection. The quality assessment module is used to confirm the regional production status data of the flexible circuit board at the production stage based on the structured data of quality inspection, and to synchronously record the production quality assessment results of the regional production status data to the storage management subsystem.

[0006] The present invention has the following beneficial effects: I. By deploying image recognition-driven positioning acquisition, curvature calibration, and region resampling mechanisms in the automated flexible circuit board production line, automatic board alignment and defect-level precision image acquisition are achieved during the production process. The system can automatically identify FPC edge offset, geometric deformation, and installation errors, and generate precise posture calibration commands for mechanical transmission actuators. This avoids the problems of low efficiency in traditional manual adjustments and fixture error compensation, thereby significantly improving the positioning accuracy and automated adjustment capabilities of FPCs during the transmission, mounting, and soldering stages.

[0007] Second, by employing a regional intelligent resampling method, differentiated inspection is performed on pad areas, trace paths, and component mounting areas. Combined with multiple types of structured quality inspection indicators, real-time quality perception and data tracking of the production stage are achieved. Compared to existing one-time low-precision whole-board imaging solutions, this system achieves on-demand high-resolution local acquisition, real-time re-image verification, and error closed-loop correction, enabling the inspection and production processes to be linked and driven, reducing false positives and false negatives, and improving the system's sensitivity and controllability in identifying micro-defects such as process deviations, oxidized pads, and circuit breaks.

[0008] Third, by establishing a quality scoring mechanism based on inspection data, production status data can be quantified, traceable, and retrospectively tracked. Production quality assessment results can be automatically recorded in the storage management subsystem to support MES / ERP process linkage, making the quality status of the production process transparent, anomalies locatable, and deviations predictable. After system implementation, it can significantly improve the consistency of flexible circuit board production, engineering closed-loop correction capabilities, and production yield, achieving an upgrade from traditional manual inspection methods to intelligent, highly sensitive positioning, and high-precision manufacturing methods, possessing high industrial application value and promotional significance. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of a module in an intelligent production management system for FPC flexible circuit boards based on image recognition. Figure 2 for Figure 1 A flowchart illustrating the quality assessment module in the middle section; Figure 3 This is a schematic diagram of a scenario for an image recognition-based intelligent production management system for FPC flexible circuit boards according to this application; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0010] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0011] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0012] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0013] To achieve the above objectives, please refer to Figures 1 to 3 A smart production management system for flexible printed circuit boards (FPCs) based on image recognition is applied to an automated production line for FPCs equipped with an optical acquisition module, positioning fixtures, mechanical transmission and execution components, and a storage management subsystem. The system includes: S1: Fix the FPC flexible circuit board to be produced in the adjustable clamping structure of the positioning fixture, and trigger the optical acquisition module to start the initial image acquisition to obtain the original detection image; In one embodiment, reference Figure 3 The operator places the FPC flexible circuit board to be produced into the adjustable clamping structure of the positioning fixture. The adjustable clamping structure includes lateral clamping arms for limiting the edge of the FPC and a fine-tunable clamping mechanism. The clamping mechanism can achieve light pressure control on the FPC flexible circuit board by adjusting the helical micro-displacement component, so as to ensure that the FPC flexible circuit board is flat and attached to the reference surface of the fixture, and avoid FPC deformation or damage caused by over-clamping.

[0014] During clamping, the pressure feedback unit built into the positioning fixture monitors the clamping pressure in real time. When the clamping pressure reaches a preset range (preferably 0.15N-0.35N), it confirms that the FPC flexible circuit board has entered a stable and fixed state. Subsequently, the fixture's posture adjustment module automatically corrects the planar angle of the FPC, aligning the longitudinal conductor direction of the FPC with the optical axis direction of the optical acquisition module, thereby reducing subsequent image recognition errors.

[0015] After confirming clamping and posture positioning, the optical acquisition module automatically starts the initial image acquisition process. The optical acquisition module includes an adjustable focal length industrial lens, a structured light projection unit, and a multispectral supplementary lighting component. The multispectral supplementary lighting component can adjust the proportion of red, blue, and white light according to the reflectivity of the FPC surface material to reduce the brightness difference between the metal lines and the substrate, thereby improving the contrast of subsequent edge detection and defect identification.

[0016] The optical acquisition module performs scanning imaging at a preset frame rate and dynamically adjusts the focal length based on the acquired optical focusing feedback information to ensure that the main line areas on the FPC surface are within the optimal imaging range. The final result is a raw detection image data set including at least one global field-of-view image and several high-precision magnified local images.

[0017] In another embodiment, to reduce the impact of micro-vibrations that may occur during the shooting process, the positioning fixture can also be equipped with a vibration damping structure. When a slight shift caused by external mechanical interference is detected in the acquisition environment, the vibration damping structure automatically absorbs vibrations and compensates for the fixture, keeping the original detection image clear and in focus.

[0018] S2: Identify the edge offset magnitude of the original detection image and use the edge offset magnitude to generate plate pose calibration commands for manipulating mechanical transmission execution components; In one embodiment, edge feature analysis is performed on the original detection image obtained in step S1. Specifically, firstly, a preset edge detection operator is called to extract the edges of the board contour in the original detection image to obtain the pixel trajectory of the board boundary; then, using the standard reference position of the target edge of the board in the image coordinate system, the pixel offset difference between the actual edge and the reference edge is calculated to obtain the edge offset amplitude data of the board in the horizontal, vertical and rotational directions.

[0019] In a preferred embodiment, the edge offset magnitude data includes: the offset of the left edge of the plate, the offset of the right edge of the plate, the offset of the upper edge of the plate, the offset of the lower edge of the plate, and the rotation offset of the corner points of the plate. The above offset data is converted to the physical space coordinate system corresponding to the mechanical transmission execution component for subsequent pose adjustment calculations.

[0020] In another embodiment, to improve calibration accuracy, pixel noise filtering and error compensation are performed on the edge offset amplitude data. The noise filtering includes moving average processing of the edge position data of continuous detection frames. The error compensation includes correcting the displacement amplification, rotation error trend and offset error caused by lens distortion based on the mechanical structure error model, so as to ensure that the offset data can truly reflect the actual position of the plate.

[0021] Based on the corrected edge offset amplitude data, a plate pose calibration command is generated and sent to the mechanical transmission execution component. The plate pose calibration command includes: horizontal displacement calibration amount, vertical displacement calibration amount, rotation calibration amount, and corresponding action execution speed and action priority, so that the mechanical execution component can perform pose correction operation according to the optimal motion path.

[0022] In another optional embodiment, a forced calibration mode is triggered when the edge offset in any direction exceeds a preset threshold range, such as an offset exceeding ±0.3mm or a rotation offset exceeding ±0.5°. In this mode, the mechanical actuator adopts a segmented fine-tuning strategy, that is, first performs rotational direction calibration, and then performs lateral and longitudinal position calibration, in order to avoid secondary offset caused by multi-axis linkage.

[0023] Finally, by executing the plate posture calibration command, the plate position is restored to a standard posture that conforms to the target installation or processing reference.

[0024] S3: In response to the completion of the board pose calibration command, local image resampling is performed on the pad area, trace path area and component mounting area in the FPC flexible circuit board to obtain structured data for quality inspection. In one embodiment, after the mechanical transmission execution component completes the board pose calibration and returns a pose calibration completion command, it automatically enters the local image resampling stage. This stage is completed collaboratively by the optical acquisition module and the image processing unit, and is used to perform high-precision resampling of key areas in the FPC flexible circuit board and extract structured inspection data for subsequent quality assessment.

[0025] More specifically, before performing this step, the target sampling grid is first regenerated based on the final calibration posture of the board, and the sampling priorities of the pad area, trace path area, and component mounting area are determined according to the process reference points on the FPC flexible circuit board. Among them, the pad area has the highest priority to ensure traceability of solder joint quality; the trace path area is used to determine trace breaks, corrosion, or width deviations; and the component mounting area is used to detect mounting accuracy and misalignment.

[0026] During the sampling process, the optical acquisition module automatically adjusts the focal length, exposure parameters, and sampling magnification according to the regional accuracy requirements to ensure consistent image quality across different areas. For example, the pad area uses higher magnification optical resampling to reveal the microstructure at the pad edges, while the trace path area uses medium magnification sampling to maintain a consistent visual recognition range.

[0027] After optical acquisition, the image processing unit preprocesses the acquired image, including illumination equalization, noise suppression, edge sharpening, and local distortion correction, to ensure clear image features and stable boundaries. Subsequently, based on the characteristics of each region, parameters such as geometric feature points, edge contours, pixel grayscale distribution, linewidth variation, component positioning deviation, and pad integrity are automatically extracted and recorded in a structured format.

[0028] Finally, the data structuring module converts the above-mentioned detection data into a unified format of quality inspection structured data, including region number, sampling timestamp, sampling coordinates, detection feature data and deviation level identifier, for use in subsequent judgment logic, visualization display and production system closed-loop feedback.

[0029] In another embodiment, to reduce image changes caused by the deformation of flexible materials, an auxiliary reference laser scanning device can be triggered synchronously before sampling to model and compensate for the micro-deformation of the sampling area, thereby further improving the accuracy and consistency of local image resampling.

[0030] S4: Based on the structured data of quality inspection, confirm the regional production status data of the production stage of the flexible circuit board, and synchronously record the production quality assessment results of the regional production status data to the storage management subsystem.

[0031] In one embodiment, after acquiring structured quality inspection data, the process proceeds to the flexible circuit board production stage confirmation and quality assessment process. This process is executed collaboratively by a quality analysis module, a production stage identification module, and a data synchronization module, which automatically identifies the current process stage of the flexible circuit board and generates production status data for the corresponding area.

[0032] More specifically, the region number, sampling type label, and characteristic deviation level in the structured data of quality inspection are first analyzed and matched with a preset stage classification table corresponding to the flexible circuit board production process. For example, when the data includes pad deformation deviation or solder wettability score, the flexible circuit board is determined to be in the post-soldering inspection stage, while when the data mainly reflects trace width, etching uniformity, and surface displacement error, it is determined to be in the etching or pre-mounting stage. Through the above matching process, regional production status data for the flexible circuit board is formed.

[0033] Subsequently, the quality analysis module performs a comprehensive quality assessment based on deviation levels, defect types, inspection area characteristics, and production stage attributes in the structured data. The assessment includes, but is not limited to, circuit integrity level, pad processing consistency score, component placement accuracy level, and surface process consistency index, and the assessment results are written into the regional production status data in a standardized quantitative form.

[0034] After generating regional production status data, the system further determines whether there are quality risks affecting the process flow based on the level information in the data structure. For example, if a high-deviation-level defect occurs consecutively in a certain inspection area, a rework flag can be automatically triggered; if all inspection data are within the normal range, the system is marked as qualified.

[0035] After data confirmation, the regional production status data and corresponding production quality assessment results are written into the storage management subsystem through the data synchronization module and archived according to the unique identification code of the flexible circuit board. The corresponding module data is stored using wireless or wired network. The storage format includes production stage identifier, inspection timestamp, regional quality score, risk level classification and historical data association index, so as to be used for subsequent tracking, production line scheduling decisions and production closed-loop optimization.

[0036] In another implementation, to improve operational efficiency, the regional production status data can be adaptively adjusted based on historical detection trends, thereby reducing the false judgment rate and improving the dynamic adaptability of flexible circuit board production management.

[0037] Preferably, the curvature calibration module further includes, before generating the plate pose calibration command for manipulating the mechanical transmission actuator using edge curvature, the following: Several edge curvature change line segments are extracted from the original detection image to characterize the shape of the flexible circuit board, and the position with the most obvious curvature change is marked as the initial reference point; The initial reference points are matched sequentially with the corresponding reference points in the preset standard contour template according to their order of appearance, forming a chain of positional correspondence. Based on the offset between the initial reference point and the corresponding benchmark point in the position correspondence chain, the initial reference point is divided into misalignment points, and the remaining initial reference points are marked as alignment points.

[0038] In one embodiment, before the curvature calibration module generates board pose calibration instructions for manipulating mechanical transmission actuators, curvature features are first extracted and reference points are matched on the flexible circuit board outline boundary in the original detection image to ensure the accuracy of subsequent pose calibration.

[0039] More specifically, an edge detection algorithm is used to scan the contour region of the flexible circuit board in the original detection image pixel by pixel, and extract several curvature change line segments that can reflect the geometric edge change trend of the circuit board. Subsequently, the local curvature extreme points, curvature direction abrupt change points, and curvature positive and negative change inflection points of each curvature change line segment are determined, and the position with the most obvious curvature change is identified and marked as the initial reference point for subsequent comparison with the benchmark analysis.

[0040] After marking the initial reference points, a standard contour template matching the current test board model is retrieved from the flexible circuit board standard structure database. The initial reference points are sequentially matched point-to-point with the corresponding reference points in the template according to the contour direction, and a positional correspondence chain is generated based on spatial geometric distance, shape continuity, and curvature trend consistency. If multiple possible positions exist for a candidate point during the matching process, the reference point with the consistent line direction between upstream and downstream points and the closest curvature change trend is selected as the final matching target.

[0041] After establishing a complete chain of positional correspondences, the offset data between the initial reference point and the corresponding benchmark point is further calculated, including horizontal offset, vertical offset, and angular offset. If the offset of an initial reference point exceeds a preset threshold range, the reference point is marked as an out-of-alignment point; if the offset is within an acceptable range, the reference point is marked as an aligned point.

[0042] In another embodiment, to avoid overall identification deviation caused by local reference point offset, a neighborhood consistency check can be performed on the position correspondence chain. When the misaligned point appears in a local continuous region, a local fitting optimization task is triggered, which corrects the curvature direction through interpolation and recalculates the offset to improve the accuracy of the reference point marking.

[0043] Preferably, the specific steps for generating the plate pose calibration command for controlling the mechanical transmission actuator are as follows: Using the location of the misalignment point as the basis for adjustment, the area where the FPC flexible circuit board is located is divided into several independent adjustment segments in the original detection image, and each adjustment segment corresponds to one or more consecutive misalignment points; Calculate the offset direction and offset magnitude of each adjustment segment, and retain segments with offset magnitudes higher than the execution threshold as valid adjustment segments; For segments with insufficient offset, automatic buffering correction is performed based on adjacent alignment points. If the correction is still insufficient after three consecutive corrections, the segment is marked as not requiring adjustment. The effective adjustment segments are arranged in their original order, and the offset direction, offset magnitude, and reference point position of each segment are recorded to form a plate pose calibration command for controlling the mechanical transmission execution components.

[0044] In one embodiment, after the curvature calibration module distinguishes between aligned and misaligned points of the initial reference point, it generates board pose calibration instructions for controlling the mechanical transmission execution components based on the distribution information of the misaligned points, so as to realize the automated and accurate calibration of the flexible circuit board.

[0045] Specifically, the location corresponding to the misalignment point is used as the basis for calibration and adjustment. The flexible circuit board area in the original detection image is geometrically partitioned, and the area is divided into several independent adjustment segments according to the contour direction and the density of misalignment points. Each adjustment segment contains one or more adjacent or continuously distributed misalignment points. If the distance between two misalignment points is less than a preset continuous threshold, they are considered to belong to the same adjustment segment to avoid unstable adjustment execution due to overly fine segmentation.

[0046] Subsequently, for each adjustment segment, the offset direction and offset magnitude are calculated. The offset direction is determined based on the offset trend of the misalignment point, and can be one or a combination of left, right, up, or down offset; the offset magnitude is obtained based on the cumulative spatial deviation between the misalignment point and the corresponding reference point. A threshold is applied to the offset magnitudes of all segments, and only segments with offset magnitudes exceeding the execution threshold are retained as valid adjustment segments, while the remaining segments are not immediately entered into the execution queue.

[0047] For segments whose offset magnitude is insufficient to meet the execution threshold, the adjacent point buffer correction algorithm is triggered. This algorithm uses adjacent aligned points or misaligned points with small offsets as constraint benchmarks to automatically micro-compensate the segment position and update the segment offset magnitude. If the segment still fails to meet the execution threshold after three consecutive buffer corrections, the system marks the segment as one that does not require adjustment and stops further correction calculations.

[0048] After determining the final valid adjustment segments, the valid segments are arranged according to the original edge detection order, and the offset direction, offset magnitude, and corresponding reference point position are recorded for each segment. Simultaneously, they are encoded into a standardized coordinate motion command format, enabling them to be recognized and executed by the mechanical transmission execution component.

[0049] Furthermore, in another implementation, the temporal sequencing of actions between multiple effective segments can be optimized to avoid reverse repetitive actions by the transmission execution component due to conflicting adjustment directions of different segments. Through this mechanism, the final generated board pose calibration command not only includes the adjustment target but also the execution order, execution level, and expected action range, effectively improving the stability and efficiency of flexible circuit board precision positioning calibration.

[0050] Preferably, the offset direction, offset magnitude, and reference point position corresponding to each segment are recorded to form a plate pose calibration command for controlling the mechanical transmission execution component, including: The initial execution priority of each segment is set based on the original order of the effectively adjusted segments; The initial execution priority is adjusted by the offset direction, offset magnitude, and reference point position corresponding to each segment to obtain the sorted adjusted segments; The motion continuity of the sorted adjustment segments is checked to obtain the check results. The motion continuity check is used to confirm whether there is a conflict or overtravel risk in the movement direction between consecutive segments. Based on the verification results, the adjustment segment is converted into an execution action instruction that can be recognized by the mechanical transmission execution component, thus obtaining the plate pose calibration instruction.

[0051] In one embodiment, after recording the offset direction, offset magnitude, and reference point position of the edge segments of the flexible circuit board, the system converts the pose deviation information of the segments into action commands executable by the mechanical control layer. The specific steps are as follows: Based on the original geometric arrangement order of the identified valid adjustment segments, an initial execution priority setting is performed on all segments. This priority is ordered according to the local structural logic of the flexible circuit board, for example, starting from the main positioning edge and numbered sequentially along the edge contour of the board, so that the segment adjustment order conforms to the natural geometric law of flexible board posture correction.

[0052] Next, the offset direction, offset magnitude, and precise coordinates of the reference point for each segment are calculated using 3D calibration weights, and the initial execution priority is dynamically adjusted accordingly. If a segment has a large offset magnitude or its offset direction involves the main attitude adjustment axis, the execution priority of that segment will be increased; conversely, if the adjustment amount of that segment is small or it belongs to a detail compensation segment, its execution priority will be delayed, thus obtaining a sorted list of adjusted segments.

[0053] Subsequently, the sequential adjustment segments are checked for motion continuity to confirm whether there are any conflicts between mechanical actions. For example, if the offset directions of consecutive segments tend to cancel each other out, there is a risk of intersection between diagonal execution paths, or the displacement of a segment may cause the mechanical transmission execution component to overtravel, dynamic correction will be performed at this stage. If correction is triggered, the segment will be reordered and the reference offset will be automatically adjusted to obtain a verification result that meets the mechanical execution characteristics.

[0054] After verification, the offset direction, offset magnitude, and reference point position of the sorted and verified adjustment segments are converted into execution commands recognizable by the mechanical transmission execution components. This conversion includes mapping the segment adjustment parameters to mechanical drive command formats, such as linear movement commands, angle rotation commands, easing and limiting control commands, and safety redundancy compensation commands. Simultaneously, all action commands are serialized and accompanied by execution condition parameters to ensure continuous and safe execution without reverse overshoot, ultimately forming the plate pose calibration command.

[0055] It is worth noting that, in another embodiment, the generated plate pose calibration command can also be simulated and deduced, and the command can be judged based on the real-time displacement error feedback generated in the deduction to ensure that the actual execution is consistent with the target calibration posture.

[0056] Preferably, the motion continuity check for the sorted adjusted segments includes: Perform path analysis on the sorted adjustment segments to confirm the movement direction and distance of each adjustment segment; Based on the movement direction of adjacent segments in each adjustment segment, confirm whether there are execution situations with opposite directions or repeated movements; Based on the execution status, travel limit detection is performed on each adjustment segment to confirm whether the movement distance of the segment exceeds the execution range allowed by the mechanical transmission execution component; The confirmation results of the integrated implementation status and scope constitute the verification results.

[0057] In one embodiment, path analysis is performed on the sorted adjustment segments. Path analysis includes calculating the corresponding offset direction vector, target movement direction trend, and movement distance range for each segment, and generating path analysis data for evaluating the rationality of the execution order. During the path analysis, based on the geometry of the flexible circuit board edge and the coordinates of the positioning reference point, the motion attributes of each segment in three-dimensional space are determined, including linear movement, angular displacement, or combined movement types, and the segment motion attribute labels are recorded.

[0058] Based on path analysis data, the directional relationship between adjacent segments is determined for the sorted adjustment segments. If the movement directions of adjacent segments are consistent or there is a transitional directional continuity, they are marked as continuous actions; if there are opposite directions, repeated cancellations, or reversed execution order, they are judged as action conflict states. For example, if the current segment requires a slight upward adjustment of the plate, but the next segment needs to perform the same downward compensation, this situation will be judged as opposite directions. For the discovered opposite directions or repeated movements, a risk label will be recorded to trigger subsequent execution order correction.

[0059] Each adjustment segment undergoes travel limit detection to confirm whether the segment's movement distance meets the travel limit characteristics of the mechanical transmission execution component. The travel limit detection process includes comparing the segment offset distance with the single-step executable range, cumulative travel allowable range, and safety buffer range of the mechanical execution component. If the movement distance of a segment exceeds the allowable range, distance scaling, action splitting, or path replanning will be automatically triggered, and a travel processing mark will be added to that segment.

[0060] The results of the motion direction relationship determination and the travel limit detection are integrated to generate the final motion continuity verification result. If all segments are determined to be continuous, executable and do not trigger risk flags, the verification result is marked as passed; if there are motion conflicts, duplicate cancellations or overtravel risks, the verification result is marked as needing adjustment and is automatically fed back to the instruction optimization module for reordering or motion splitting.

[0061] Alternatively, in another embodiment, a virtual execution simulation can be added during the verification process to verify in advance whether there is any potential mechanical interference or execution sequence lock-up risk in the actual physical environment, so as to further improve the reliability of the verification results.

[0062] Preferably, before performing local image resampling on the pad area, trace path area, and component mounting area of ​​the FPC flexible circuit board in the region resampling module, the following steps are also included: After the plate pose calibration command is completed, an action confirmation flag is set for each adjustment segment to trigger optical detection for positioning verification of the plate pose calibration.

[0063] In one embodiment, after the plate pose calibration command is executed by the mechanical transmission execution component, a corresponding action confirmation flag is set for each adjustment segment involved in the calibration execution process. The action confirmation flag is used to record the execution status of each adjustment segment and serves as the basis for triggering subsequent optical detection actions.

[0064] The positioning verification process is initiated. This process uses the optical acquisition module to detect the positioning accuracy of the local area where adjustments were performed, confirming whether the positional offset of the adjusted segment has met the preset accuracy requirements. During the positioning verification, the deviation between the actual pose change and the expected change in the adjusted area is recorded and analyzed, and the corresponding action confirmation markers are updated based on the deviation. For example, if a positional error is detected in a certain adjustment segment, the action confirmation marker for that segment is marked as "reset required," prompting the mechanical transmission execution component to re-perform the fine-tuning action in that area; conversely, if the actual detection result meets the positioning accuracy range, the action confirmation marker for that adjustment segment is marked as "completed."

[0065] After performing positioning verification and updating the markers for all adjusted segments, the action confirmation markers are summarized and analyzed. When all adjusted segments are in the "completed" state, it is confirmed that the board has completed the standard posture lock. At this point, the area resampling module enters the next stage of operation, performing local image resampling on the pad area, trace path area, and component mounting area of ​​the flexible circuit board to obtain a high pixel ratio and clearer detection image.

[0066] Preferably, the positioning verification for triggering optical detection to perform plate pose calibration includes: Based on the action confirmation mark, the optical acquisition module is triggered again to reshoot the adjusted segment at a fixed point, the shooting focal length of the optical acquisition module is adjusted to the local detection mode, and the light source brightness is switched to the high illumination constant mode. Call the positioning command of the optical acquisition module to align the acquisition lens with the reference point position corresponding to the adjustment segment; Set the reshoot area range so that the reshoot area covers the edge line segment of the adjustment segment and the adjacent pad area; The optical acquisition module is activated to acquire images of the adjusted segment, resulting in a local detection image. The plate pose calibration and verification are performed on the adjusted segment based on the local detection image.

[0067] In one embodiment, based on the action confirmation marker generated in the previous step, the optical acquisition module is invoked again to initiate the fixed-point reshoot process. Upon initiation of the reshoot, the optical acquisition module switches the shooting focus from global detection mode to local detection mode, enabling the lens to precisely focus on the corresponding area of ​​the adjusted segment to be reviewed. Simultaneously, the dynamic compensation light source mode originally used for global detection is switched to a constant high-light mode to ensure higher brightness uniformity and stable edge contrast in the local area of ​​the acquired image, thereby facilitating subsequent feature recognition.

[0068] The positioning command in the optical acquisition module is invoked, causing the acquisition lens to automatically align itself based on the position of the reference point corresponding to the adjustment segment. After the lens performs autofocus, minor position adjustments, and edge recognition confirmation, it enters the reshoot area setting stage. At this time, based on the contour range of the adjustment segment, the reshoot area is set to cover the edge line segment of the adjustment segment and its adjacent pad area, ensuring that the acquired image data includes not only the adjustment position itself but also the calibration reference area used for position comparison.

[0069] After the area is defined, the optical acquisition module is activated to acquire images, obtaining a local detection image corresponding to the current pose of the plate. Once acquisition is complete, the local detection image is input into the image recognition component for depth analysis.

[0070] Based on local detection images, the system verifies whether the adjusted segment has reached the expected execution path. It also compares the edge positioning results before and after adjustment, the offset of the pad center position, and the stability characteristics of the detection frame to confirm whether the board pose calibration has been accurately performed. When the verification result shows that the adjusted segment meets the preset positioning accuracy requirements, the action confirmation flag for the corresponding adjusted segment is updated to "Verification Passed." If there is insufficient offset or over-adjustment, the action confirmation flag is updated to "Secondary Calibration Required," triggering subsequent correction execution logic.

[0071] Preferably, the positioning command for the optical acquisition module is as follows: Send a camera movement command to move the camera along the X-axis to the horizontal coordinate corresponding to the position of the adjusted segment; Send a lens fine-tuning command to move the lens along the Y-axis to get closer to the center area of ​​the reference point; Send a lens focus command to adjust the lens focal length to an imaging focus that matches the height of the FPC flexible circuit board surface.

[0072] In one embodiment, based on the adjustment segment position information recorded in the preceding steps, a lens lateral movement command is sent to the optical acquisition module, causing the lens to move along the X-axis to the lateral coordinate position corresponding to the adjustment segment. During the movement, real-time position feedback monitoring of the lens trajectory is performed simultaneously to avoid trajectory deviation or exceeding the travel limit due to excessive movement.

[0073] Subsequently, once the lens reaches the lateral target position, fine-tuning commands are sent to the lens, causing it to move precisely along the Y-axis over short distances. This fine-tuning process is achieved through step-driven operation, with the lens displacement typically controlled within the range of 0.05 mm to 0.2 mm. Visual feature comparison is used to gradually bring the lens closer to the center area of ​​the reference point in the adjusted segment, ensuring that the final alignment error of the positioning process is controlled within ±0.1 mm.

[0074] After the lens position is locked, the system sends a lens focusing command, triggering the focusing mechanism via the autofocus module within the optical acquisition module. This focusing mechanism automatically adjusts the lens focal length using a dynamic focusing algorithm, based on the actual thickness and surface height differences of the FPC flexible circuit board, and controls the depth of focus to the optimal imaging position that matches the fine texture features of the FPC flexible circuit board surface. Once focusing is complete, the system confirms that image sharpness, edge sharpness, and illumination contrast all meet preset imaging threshold conditions, ultimately completing the positioning operation.

[0075] As an example of the present invention, reference is made to... Figure 2 As shown, the quality assessment module in this example includes: S41: Determine the production stage of the flexible circuit board based on the pad area, trace path area and component mounting area in the FPC flexible circuit board. S42: Read the inspection fields corresponding to the production stage of the flexible circuit board through the structured data of quality inspection; S43: Obtain regional production status data of the production stage of the flexible circuit board based on the detection field; S44: Conduct a quality assessment of regional production status data to obtain production quality assessment results; S45: Synchronously record the production quality assessment results to the storage management subsystem.

[0076] In one embodiment, based on the local image resampling results of the FPC flexible circuit board, region identification and parsing are performed on the pad area, trace path area, and component mounting area. Using preset region production stage identification rules, the three types of regions are mapped and matched with production process stages: the pad area corresponds to the pre-soldering stage, the trace path area corresponds to the power-on continuity detection stage, and the component mounting area corresponds to the mounting and packaging stage. Through this mapping relationship, the current production stage identifier of the flexible circuit board is finally confirmed.

[0077] The quality inspection system reads structured data to identify the inspection fields corresponding to the current production stage. These fields include, but are not limited to, pad integrity, trace continuity, mount offset, optical defects, and surface contamination. Different combinations of inspection fields are automatically invoked for different production stages to improve inspection accuracy and adaptability to different stages.

[0078] Based on the read detection fields, further production status data for the corresponding areas of the flexible circuit board is extracted. The extracted data typically includes information such as area defect type, defect location coordinates, defect area, offset, abnormal surface reflectivity, and structural integrity level. Simultaneously, a timestamp and area index number are appended to each status data entry, forming a traceable and retrievable production status label.

[0079] A quality assessment is conducted on the regional production status data. The assessment process includes multiple criteria, such as defect quantity thresholds, allowable offset ranges, trace breakpoint feature morphology identification results, and component mounting firmness scores. Based on the assessment rules, quality grade labels are generated for the flexible circuit boards, such as "Pass," "Pending Repair," and "Rejected," to obtain the final production quality assessment result.

[0080] The production quality assessment results are synchronously recorded in the storage management subsystem. The records include not only the final assessment results, but also the corresponding testing time, the number of the testing device, the operator's identity information, regional production status data, and corresponding testing fields. The storage management subsystem supports subsequent production tracking, batch quality statistical analysis, and the generation of process optimization strategies.

[0081] Preferably, quality assessment of regional production status data includes: Extract the corresponding position offset, pad integrity data, and circuit continuity test results from the regional production status data, and organize them according to the regional number; Perform a comparative check on the positional offset. If the offset is within the range of ±0.05 mm to ±0.25 mm, it is recorded as a slight offset area; if the offset exceeds ±0.25 mm, it is recorded as a severe offset area. The integrity of the pads is checked for flatness and adhesion rate. Pads with flatness below 85% or surface oxidation area exceeding 0.3 mm² are recorded as abnormal pad areas. Line continuity is checked for breaks. If the length of a line break is greater than 0.10 mm or the exposed area of ​​copper foil exceeds 0.20 mm², it is marked as an abnormal line area. Integrate the results of inspections and tests in different areas to obtain production quality assessment results.

[0082] In one embodiment, regional production status data is read, and the included position offset, pad integrity data, and circuit continuity detection results are organized according to region number. The organized data is stored in an internal cache by region for subsequent judgment and result labeling.

[0083] Secondly, regarding the positional offset, a comparison is performed according to a preset standard. When an offset in a certain area is detected to be within the range of ±0.05 mm to ±0.25 mm, the area is marked as "slight offset," and the offset direction and trend are recorded. If the offset exceeds ±0.25 mm, the area is marked as "severe offset," and a rework prompt for that area is triggered.

[0084] The integrity of the solder pads is inspected. The inspection process includes checking the flatness of the solder pads and judging the surface adhesion. When the flatness of the solder pad is found to be less than 85% or the surface oxidation area exceeds 0.3 mm², the area where the solder pad is located is recorded as "solder pad abnormality". At the same time, the abnormal shape (such as depression, delamination or oxidation) is recorded as a basis for subsequent traceability.

[0085] Next, a line continuity check is performed, and the length of the break and the exposure of the copper foil surface are identified. When a line break length exceeding 0.10 mm or a copper foil exposed area exceeding 0.20 mm² is detected, the detected area is marked as "line abnormality," and the corresponding process information is automatically associated for process cause location.

[0086] The results of position offset level, pad quality status and circuit continuity test are summarized, and the judgment label corresponding to each area is stored in the quality assessment output table. At the same time, an overall production quality assessment result is generated for use in the production management system or for manual confirmation.

[0087] Of particular importance, the curvature calibration module includes: Extract edge segments of the flexible circuit board from the original detection image; The extracted flexible circuit board edge segments are compared to confirm the offset distance and offset direction of the edge segments relative to the standard positioning boundary. The offset distance is divided into corresponding movement levels, and a corresponding adjustment method is assigned to each movement level; Generate clamping fine-tuning commands, linear movement commands, or rotation compensation commands based on the offset direction and the magnitude of movement; The generated clamping fine-tuning instructions, linear movement instructions, or rotation compensation instructions are combined and output in the order of execution to form plate posture calibration instructions for controlling mechanical transmission execution components.

[0088] In one embodiment, the curvature calibration module is used to automatically analyze and accurately correct the posture deviation of the flexible circuit board (FPC) after clamping and fixing, so as to ensure that the board is in a standardized spatial position during subsequent visual inspection, mounting positioning and transfer processes.

[0089] First, edge extraction processing is performed based on the acquired original detection images. This embodiment employs a sub-pixel edge tracing algorithm to detect the edges of the flexible circuit board and extract continuous edge segments. These edge segments include a left edge segment, a right edge segment, and optional upper or lower edge auxiliary segments, providing complete edge geometric reference data.

[0090] Subsequently, a position comparison operation is performed on the extracted edge segments. The comparison process uses a preset standard positioning boundary as a reference frame, and calculates the offset distance and offset direction between different edge segments and the standard positioning boundary. For example, when the left edge segment offsets outward, the offset direction is recorded as positive offset; while when the plate body rotates and tilts, the offset gradient difference between the right and left edges is recorded and determined as a rotational deviation type.

[0091] After the offset calculation is completed, the offset distance is divided into orders of magnitude. In this embodiment, the offset orders of magnitude can be referred to as Table 1: Table 1. Correspondence between offset magnitude and calibration strategy By using the above division method, we can ensure that the most suitable mechanical correction scheme is executed according to the offset intensity, thereby improving correction efficiency and avoiding overcorrection.

[0092] Subsequently, corresponding operation commands are generated based on the determined offset direction and movement magnitude. If the offset is minor, the system generates a clamping fine-tuning command to adjust the position of the clamping mechanism; when the offset is medium, a linear movement command is automatically generated to drive the platform to move along the X or Y coordinate axis; if the offset is tilted or reaches the rotation calibration threshold, a rotation compensation command is generated to adjust the plate's attitude angle.

[0093] Finally, based on offset priority, adjustment order, and mechanical response characteristics, the generated clamping fine-tuning commands, linear movement commands, or rotation compensation commands are arranged into an execution sequence to form the final board pose calibration command. This calibration command is transmitted to the mechanical transmission execution component for execution, causing the flexible circuit board to gradually return to the standard positioning posture, achieving automatic closed-loop calibration.

[0094] Of particular importance, extracting the edge segments of the flexible circuit board from the original detection image also includes: Determine the region where the flexible circuit board is located in the original detection image; Adjust the field of view of the optical acquisition module so that the edge of the flexible circuit board is within the effective recognition range; The outer contour of the flexible circuit board is scanned segment by segment based on the effective recognition range; The continuous edge points obtained from the scan are connected in sequence to form the edge segments of the flexible circuit board.

[0095] In one embodiment, in order to accurately identify the edge shape of the flexible circuit board, this step extracts the edge line segments of the flexible circuit board from the original detection image and uses these edge line segments as the basis data for subsequent board pose calibration and region resampling.

[0096] Specifically, based on the original detection image acquired by the optical acquisition module, the region containing the flexible circuit board in the original detection image is located through color feature threshold recognition, structured texture feature analysis, and image region contour detection algorithms, and a flexible circuit board region localization box is generated. The localization box is used to limit the processing range of subsequent edge recognition to avoid misidentification caused by background interference.

[0097] Subsequently, the field of view of the optical acquisition module is adjusted to ensure that the edge of the flexible circuit board is in a clearly visible area, and to ensure that the pixel resolution of the edge area meets the edge detection accuracy requirements. For example, when the edge thickness is less than 0.4 mm, the system automatically adjusts the acquisition focal length and enables the local exposure compensation function to improve the imaging contrast of the flexible circuit board edge to a recognizable threshold range.

[0098] After optimizing the acquisition field of view, the outer contour of the flexible circuit board is scanned segment by segment based on the effective recognition range. The scanning process adopts an envelope contour extraction algorithm, combined with Canny operator edge detection and local binarization threshold control, to perform multi-frame averaging fitting on continuous edge information, thereby reducing edge detection errors caused by board surface reflection, local bending, or positional deviation.

[0099] Finally, the edge point data obtained from segment-by-segment scanning are organized in spatial coordinate order, and continuous edge segments are generated using a curve fitting algorithm. When the flexible circuit board exhibits local bending or nonlinear boundaries, a segmented fitting strategy is automatically adopted to ensure that the output edge segments are consistent with the actual deformation of the flexible circuit board, thus ensuring the stability and accuracy of subsequent board posture calibration calculations.

[0100] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0101] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. An intelligent production management system for FPC flexible circuit boards based on image recognition, characterized in that, An automated production line for flexible circuit boards, equipped with an optical acquisition module, positioning fixture, mechanical transmission and execution components, and a storage management subsystem, is described in the system as follows: The positioning and acquisition module is used to fix the FPC flexible circuit board to be produced in the adjustable clamping structure of the positioning fixture, and to trigger the optical acquisition module to start the initial image acquisition to obtain the original detection image. A curvature calibration module is used to identify the edge offset magnitude of the original detection image and generate plate pose calibration commands for manipulating the mechanical transmission actuator component using the edge offset magnitude; wherein, before generating the plate pose calibration commands for manipulating the mechanical transmission actuator component using the edge offset magnitude, the curvature calibration module also includes: Several edge curvature change line segments are extracted from the original detection image to characterize the shape of the flexible circuit board, and the position with the most obvious curvature change is marked as the initial reference point; The initial reference points are matched sequentially with the corresponding reference points in the preset standard contour template according to their order of appearance, forming a chain of positional correspondence. Based on the offset between the initial reference point and the corresponding reference point in the position correspondence chain, the initial reference point is divided into misalignment points, and the remaining initial reference points are marked as alignment points; specifically, the plate pose calibration command for controlling the mechanical transmission execution component is generated as follows: Using the location of the misalignment point as the basis for adjustment, the area where the FPC flexible circuit board is located is divided into several independent adjustment segments in the original detection image, and each adjustment segment corresponds to one or more consecutive misalignment points; Calculate the offset direction and offset magnitude of each adjustment segment, and retain segments with offset magnitudes higher than the execution threshold as valid adjustment segments; For segments with insufficient offset, automatic buffering correction is performed based on adjacent alignment points. If the correction is still insufficient after three consecutive corrections, the segment is marked as not needing adjustment. For segments with insufficient offset, the adjacent point buffering correction algorithm is triggered. The adjacent point buffering correction algorithm uses adjacent alignment points or misaligned points with small offsets as constraint references to automatically micro-compensate the segment position and update the segment offset. The effective adjustment segments are arranged in the original order, and the offset direction, offset magnitude and reference point position of each segment are recorded to form a plate pose calibration command for controlling the mechanical transmission execution component. The region resampling module is used to perform local image resampling on the pad area, trace path area and component mounting area of ​​the FPC flexible circuit board in response to the completion of the board pose calibration command, so as to obtain structured data for quality inspection. The quality assessment module is used to confirm the regional production status data of the flexible circuit board at the production stage based on the structured data of quality inspection, and to synchronously record the production quality assessment results of the regional production status data to the storage management subsystem.

2. The image recognition-based intelligent production management system for FPC flexible circuit boards according to claim 1, characterized in that, Record the offset direction, offset magnitude, and reference point position for each segment to form the plate pose calibration command used to control the mechanical transmission execution component, including: The initial execution priority of each segment is set based on the original order of the effectively adjusted segments; The initial execution priority is adjusted by the offset direction, offset magnitude, and reference point position corresponding to each segment to obtain the sorted adjusted segments; The motion continuity of the sorted adjustment segments is checked to obtain the check results. The motion continuity check is used to confirm whether there is a conflict or overtravel risk in the movement direction between consecutive segments. Based on the verification results, the adjustment segment is converted into an execution action instruction that can be recognized by the mechanical transmission execution component, thus obtaining the plate pose calibration instruction.

3. The image recognition-based intelligent production management system for FPC flexible circuit boards according to claim 2, characterized in that, The motion continuity check for the sorted and adjusted segments includes: Perform path analysis on the sorted adjustment segments to confirm the movement direction and distance of each adjustment segment; Based on the movement direction of adjacent segments in each adjustment segment, confirm whether there are execution situations with opposite directions or repeated movements; Based on the execution status, travel limit detection is performed on each adjustment segment to confirm whether the movement distance of the segment exceeds the execution range allowed by the mechanical transmission execution component; The confirmation results of the integrated implementation status and scope constitute the verification results.

4. The image recognition-based intelligent production management system for FPC flexible circuit boards according to claim 1, characterized in that, Before performing local image resampling on the pad area, trace path area, and component mounting area of ​​the FPC flexible circuit board in the region resampling module, the following steps are also included: After the plate pose calibration command is completed, an action confirmation flag is set for each adjustment segment to trigger optical detection for positioning verification of the plate pose calibration.

5. The image recognition-based intelligent production management system for FPC flexible circuit boards according to claim 4, characterized in that, The positioning verification for triggering optical detection to calibrate the plate's pose includes: Based on the action confirmation mark, the optical acquisition module is triggered again to reshoot the adjusted segment at a fixed point, the shooting focal length of the optical acquisition module is adjusted to the local detection mode, and the light source brightness is switched to the high illumination constant mode. Call the positioning command of the optical acquisition module to align the acquisition lens with the reference point position corresponding to the adjustment segment; Set the reshoot area range so that the reshoot area covers the edge line segment of the adjustment segment and the adjacent pad area; The optical acquisition module is activated to acquire images of the adjusted segment, resulting in a local detection image. The plate pose calibration and verification are performed on the adjusted segment based on the local detection image.

6. The image recognition-based intelligent production management system for FPC flexible circuit boards according to claim 5, characterized in that, The specific positioning commands for the optical acquisition module are as follows: Send a camera movement command to move the camera along the X-axis to the horizontal coordinate corresponding to the position of the adjusted segment; Send a lens fine-tuning command to move the lens along the Y-axis to get closer to the center area of ​​the reference point; Send a lens focus command to adjust the lens focal length to an imaging focus that matches the height of the FPC flexible circuit board surface.

7. The image recognition-based intelligent production management system for FPC flexible circuit boards according to claim 1, characterized in that, The quality assessment module includes: The production stage of the flexible circuit board can be determined by the pad area, trace path area and component mounting area in the FPC flexible circuit board. The inspection fields for the corresponding production stage of the flexible circuit board are read from the structured data of quality inspection. Based on the detection fields, obtain the regional production status data of the production stage of the flexible circuit board; To obtain production quality assessment results, the regional production status data is evaluated. The production quality assessment results are recorded synchronously in the storage management subsystem.

8. The image recognition-based intelligent production management system for FPC flexible circuit boards according to claim 1, characterized in that, Quality assessment of regional production status data includes: Extract the corresponding position offset, pad integrity data, and circuit continuity test results from the regional production status data, and organize them according to the regional number; Perform a comparative check on the positional offset. If the offset is within the range of ±0.05 mm to ±0.25 mm, it is recorded as a slight offset area; if the offset exceeds ±0.25 mm, it is recorded as a severe offset area. The integrity of the pads is checked for flatness and adhesion rate. Pads with flatness below 85% or surface oxidation area exceeding 0.3 mm² are recorded as abnormal pad areas. The continuity of the line is checked for breakpoints. If the length of the breakpoint is greater than 0.10 mm or the exposed area of ​​copper foil exceeds 0.20 mm², it is marked as an abnormal area of ​​the line. Integrate the results of inspections and tests in different areas to obtain production quality assessment results.

Images