Multi-degree-of-freedom reconfigurable stamping part online visual detection system and method

Abstract

The application discloses a multi-degree-of-freedom reconfigurable stamping part online visual detection system and method, relates to the technical field of visual detection, and comprises a basic support module, a motion execution module, an optical detection module, a control module, and a risk point database and a detection strategy template library associated with the control module. The application proposes to combine risk point data, drive the omnidirectional movement of a camera, eliminate blind areas such as edges and curved surfaces, reduce the use amount of the camera and cost, match exclusive solutions for risk points, cooperate with optical compensation, improve defect recognition, balance detection accuracy and efficiency, automatically retrieve parameters when changing dies, do not need manual operation, shorten the adaptation time, adapt to multi-variety production, use shockproof, dynamic light adjustment and automatic calibration design, guarantee detection accuracy stability, reduce operation and maintenance frequency and faults, support parameter updating and program reconstruction, do not need to change hardware to adapt to multiple stamping parts, improve equipment reuse rate, and reduce enterprise investment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of visual inspection, in particular to a multi-degree-of-freedom reconfigurable online visual inspection system for stamping parts. BACKGROUND

[0002] As the core components in the manufacturing field of automobiles, home appliances, etc., the surface quality of stamping parts directly determines the structural safety and service life of products. With the development of manufacturing towards high precision and high rhythm, three core requirements are put forward for online visual inspection of stamping parts: full-area coverage, adaptive precision, and rapid mold switching adaptation. However, the existing detection technology has the following key pain points, which cannot meet the needs of industrial production: the existing stamping part detection mostly adopts the mode of fixed camera array + single light source, the camera pose and light source parameters are fixed, and only specific planes or simple curved surface areas can be covered. For the high-risk areas of stamping parts such as edges, concave-convex corners and complex curved surfaces, fixed viewing angles cannot capture defect features, and missed detection is easy to occur; some detection equipment uses single-axis or dual-axis motion mechanisms, but due to the lack of degrees of freedom, it is still difficult to realize the direct observation of risk points, resulting in a decrease in image clarity and defect recognition, and the risk level, defect type and surface characteristics of different areas of stamping parts are significantly different, but the existing system uses a unified parameter detection mode: regardless of the characteristics of risk points, the same light source mode and camera parameters are used, which leads to insufficient detection precision of high-risk points, and there is no risk point exclusive database and strategy template, which cannot realize fine detection according to needs, and the mold switching adaptation efficiency is low.

[0003] In summary, the limitations of the existing online visual inspection technology for stamping parts in terms of coverage range, adaptive ability, mold switching efficiency and stability have become a key bottleneck restricting high-precision production in the manufacturing industry, and a new detection system and method that can realize multi-degree-of-freedom reconfiguration, risk point adaptive detection and rapid mold switching adaptation is urgently needed. SUMMARY

[0004] To solve the above technical problems, a multi-degree-of-freedom reconfigurable online visual inspection system for stamping parts is provided, which solves the above problems.

[0005] To achieve the above purposes, the technical scheme adopted by the present application is:

[0006] The multi-degree-of-freedom reconfigurable online visual inspection system for stamping parts comprises:

[0007] a basic support module, a motion execution module, an optical detection module, a control module, and a risk point database and a detection strategy template library associated with the control module;

[0008] The basic support module is used to provide stable support and shock protection, and comprises a concrete base, a gantry and a shockproof component, which provides a rigid foundation for system operation;

[0009] The motion execution module is electrically connected to the basic support module and is used to drive the detection component to achieve multi-degree-of-freedom motion, including a four-track motion system and a multi-axis gimbal array. The multi-axis gimbal array is equipped with an industrial camera to achieve full coverage of the detection area and can fine-tune the gimbal attitude according to the surface normal vector of the risk point.

[0010] The optical detection module is electrically connected to the motion execution module and is used to acquire clear images of the stamped parts. It includes an industrial camera and a programmable multi-channel LED strip light source, which can switch the lighting mode and parameters according to the surface features and defect types of the stamped parts.

[0011] The control module is electrically connected to the motion execution module and the optical detection module respectively, and is used to coordinate system actions and data processing. It includes a mold feature parameter library, a PLC motion control unit, a host computer control system, and the risk point database and detection strategy template library.

[0012] The risk point database stores the three-dimensional coordinates, surface normal vectors, cracking mechanism information, geometric features, defect types, and surface material reflectivity of each risk point in the stamped part.

[0013] The detection strategy template library associates a predefined detection strategy template with each risk point. The template contains the risk point coordinates and normal vector, risk level, defect type, and detection parameters corresponding to the material's reflectivity, enabling adaptive detection.

[0014] Preferably, the basic support module specifically includes:

[0015] Concrete base unit: Used to support the overall weight of the system, prevent foundation deformation during system operation, and ensure the initial positioning accuracy of the detection components;

[0016] Gantry unit: Made of high-strength steel, with built-in prestressed tie rods, and installed with precise calibration to form a stable support frame, providing a foundation for the horizontal movement of the motion execution module;

[0017] Vibration damping unit: Used to suppress vibration interference, including dampers, vibration-damping camera brackets and motor vibration isolation pads, to prevent vibration from affecting the image acquisition accuracy of industrial cameras.

[0018] Preferably, the motion execution module specifically includes:

[0019] Four-track motion unit: High-strength steel rails with hardened surface are used to drive the multi-axis gimbal array to move horizontally, providing basic displacement support for risk point coverage;

[0020] Multi-axis gimbal unit: A single gimbal supports horizontal and vertical rotation. The body adopts a lightweight structure and is equipped with dynamic balance correction components. It can calculate the optimal observation posture based on the surface normal vector of the risk point and drive the gimbal servo motor to fine-tune it to the minimum angle between the camera optical axis and the normal.

[0021] Servo drive unit: including servo motor and reducer, adopts full closed-loop control, drives the four-rail motion unit and multi-axis gimbal unit to move precisely, ensuring the accuracy of risk point positioning.

[0022] Preferably, the optical detection module specifically includes:

[0023] Industrial camera unit: used to identify microcracks, for selecting industrial cameras with suitable resolution and field of view;

[0024] LED light source unit: contains multiple strip light sources, arranged differently according to the inspection area of ​​the stamped parts, and switches the light source form according to the type of defect at the risk point. When detecting microcracks, the double-sided strip light is turned on, and when detecting highly reflective curved surface defects, the high-angle light source is turned on.

[0025] Light source control unit: Supports multi-channel dimming, adjusts the light source intensity according to the reflective characteristics of materials at risk points, and works with industrial camera exposure parameter control to avoid overexposure and underexposure of images.

[0026] Preferably, the mold feature parameter library unit of the control module specifically includes:

[0027] Parameter storage sub-unit: Parameters are stored in categories according to mold ID. Each mold parameter contains multiple sets of camera pose schemes, and is also associated with the risk point database and detection strategy template library data corresponding to the mold.

[0028] Parameter call subunit: After receiving the mold change signal, read the current mold ID and quickly call up the corresponding parameter group and associated risk point detection strategy template without manual intervention;

[0029] Parameter update subunit: Supports adding and modifying mold parameters and risk point strategy templates via the host computer. After the update, it is automatically synchronized to the PLC motion control unit to adapt to the detection of new types of stamping parts;

[0030] Dynamic optical compensation subunit: During the camera movement, the angle between the camera and the surface of the risk point of the stamping part is calculated in real time, and the LED light source illumination angle is adjusted synchronously to ensure uniform illumination in the detection area and adapt to the geometric features of the risk point.

[0031] Multiple exposure control subunit: Images of high reflectivity risk areas are acquired multiple times with different exposure levels. Reflectivity artifacts are eliminated by image fusion. At the same time, the corresponding image processing algorithm is matched according to the type of defect at the risk point. Microcracks are treated with edge enhancement algorithm and wrinkles are treated with grayscale contrast algorithm.

[0032] Defect Output Unit: Feeds the detection results back to the production line MES system in real time, while storing image data and detection logs, supporting historical data traceability, and marking defect priorities according to risk level;

[0033] Cooperative motion control unit: Adopting a track-gimbal cooperative motion model, and through an automatic pose mapping algorithm, combined with risk point coordinates and normal vector data, the industrial camera can cover the detection area of ​​traditional multiple fixed cameras, improving workspace utilization.

[0034] Precision calibration subunit: Periodically triggers precision calibration, detects motion errors through the laser tracker, automatically compensates the servo drive unit, and simultaneously calibrates the angle accuracy between the camera optical axis and the normal of the risk point to ensure long-term stability of positioning and observation attitude.

[0035] Preferably, the predefined logic of the detection strategy template library is as follows:

[0036] Based on the cracking mechanism analysis, geometric features, defect types, and surface material reflectivity of the risk points, a template containing the following parameters is generated:

[0037] Observation attitude parameters: the target angle between the camera optical axis and the normal to the risk point, and the set values ​​of the gimbal horizontal angle and pitch angle;

[0038] Light source parameters: light source type, light source intensity level, illumination angle;

[0039] Camera parameters: exposure time, gain, resolution, focus distance;

[0040] Priority parameters: detection order and number of images captured.

[0041] A multi-degree-of-freedom reconfigurable online visual inspection method for stamped parts includes the following steps:

[0042] After the die change signal is triggered, the control module reads the current stamping die ID, retrieves the corresponding parameter group from the die feature parameter library unit, and simultaneously retrieves the coordinates, normal vectors and associated detection strategy templates of all risk points of the die from the risk point database and detection strategy template library.

[0043] The PLC motion control unit drives the four-track motion unit to move the multi-axis gimbal array, completing the X / Y / Z axis positioning. At the same time, according to the observation attitude parameters in the detection strategy template, it drives the multi-axis gimbal to fine-tune to the optimal attitude where the angle between the camera optical axis and the normal of the risk point meets the requirements.

[0044] The light source control unit switches the illumination mode according to the light source parameters in the detection strategy template. When detecting microcracks, it turns on the double-sided strip light and when detecting highly reflective curved surfaces, it turns on the high-angle light. It adjusts the intensity and angle of the LED light source to ensure uniform illumination in the detection area.

[0045] The industrial camera performs automatic focusing, sets exposure time and gain according to the camera parameters in the template, and acquires images of stamped parts. During the acquisition process, the host computer analysis unit adjusts the light source angle in real time through the dynamic optical compensation subunit and performs multiple exposures on high reflective risk areas.

[0046] The host computer analysis unit calls the image processing algorithm that matches the type of risk point defect to process the acquired image, identify defects such as cracks and wrinkles, and generate an inspection report with risk point priority.

[0047] If a mold needs to be switched, repeat steps S1-S5; if the same mold is being inspected continuously, after inspecting a preset number of workpieces, the control module will automatically trigger a precision calibration, using a laser tracker to calibrate motion errors and camera posture to ensure inspection stability.

[0048] Preferably, the PLC motion control unit drives the four-rail motion unit to move the multi-axis gimbal array, completing X / Y / Z axis positioning. Simultaneously, based on the observation attitude parameters in the detection strategy template, it drives the multi-axis gimbal to fine-tune to the optimal attitude where the angle between the camera optical axis and the risk point normal meets the requirements. Specifically, this includes:

[0049] The four-track motion unit first drives the multi-axis gimbal array to move to the mold detection starting position, and then performs error compensation after it is in place.

[0050] According to the pan-tilt-zoom array's horizontal and vertical angle settings in the detection strategy template, the horizontal angle is first adjusted to the target value, and then the vertical angle is adjusted.

[0051] The servo drive unit feeds back the motion positioning signal, and the PLC motion control unit compares the actual position and pose of the camera with the template setting value. If the deviation exceeds the allowable range, it triggers a secondary adjustment until the requirements are met.

[0052] After the movement is completed, the control module sends a ready signal to the optical detection module. At the same time, it sorts the risk points to be detected according to their risk level, with high-risk points entering the detection queue first and waiting for image acquisition instructions.

[0053] Preferably, the light source control unit switches the illumination mode according to the light source parameters in the detection strategy template, turns on the double-sided strip light when detecting microcracks, turns on the high-angle light when detecting highly reflective curved surfaces, and adjusts the intensity and angle of the LED light source to ensure uniform illumination in the detection area. The light source adjustment logic is as follows:

[0054] When the defect type of the risk point is a microcrack, the light source control unit turns off the default light source, turns on the dual-side strip light, and adjusts the light source intensity to make the crack produce a shadow effect.

[0055] When the risk point defect type is wrinkling and necking, turn on the high-angle light source, adjust the intensity, and adjust the lighting angle in combination with the curved surface geometry to highlight the wrinkles and thickness changes;

[0056] When the surface material of the risk point is a bare metal with a highly reflective coating, reduce the light source intensity and extend the camera exposure time to avoid reflections from covering up the defects;

[0057] When the risk point is a sharp or concave corner, turn on the dual-angle strip light to eliminate the blind spot of the corner shadow.

[0058] Preferably, the defect detection logic in the step where the host computer analysis unit calls an image processing algorithm that matches the type of defect at the risk point to process the acquired image, identify defects such as cracks and wrinkles, and generate a detection report containing risk point priorities is as follows:

[0059] The host computer analysis unit first performs fusion processing on the multi-exposure images, and then calls the corresponding algorithm according to the type of risk point defect:

[0060] When detecting microcracks, an edge enhancement-based algorithm is used to extract linear features, which are then compared with a microcrack size threshold to determine whether they are defects.

[0061] When detecting macroscopic cracking and wrinkling, an image segmentation algorithm based on region growing is used to identify abnormal areas and determine the defect type by combining geometric morphology.

[0062] After the inspection is completed, the defects are marked according to the risk level, which includes high risk, medium risk and low risk. High-risk defects are pushed to the production line early warning terminal in real time, while medium and low-risk defects are summarized in the periodic report.

[0063] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0064] This invention proposes to combine risk point data to drive the camera to move in all directions, eliminating blind spots such as edges and curved surfaces, reducing the number of cameras used and lowering costs. It matches a dedicated solution for each risk point, and with optical compensation, improves defect identification, balances detection accuracy and efficiency, automatically retrieves parameters during mold changes without manual operation, shortens adaptation time, and is suitable for multi-variety production. It uses a shockproof, dynamic dimming, and automatic calibration design to ensure stable detection accuracy, reduce maintenance frequency and failures, and supports parameter updates and program refactoring. It adapts to multiple types of stamping parts without hardware replacement, improves equipment reuse rate, and reduces enterprise investment. Attached Figure Description

[0065] Figure 1 This is a system framework diagram of the present invention;

[0066] Figure 2 This is a flowchart illustrating the steps of the present invention. Detailed Implementation

[0067] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0068] Reference Figure 1 As shown, the multi-degree-of-freedom reconfigurable online visual inspection system for stamped parts includes:

[0069] The system includes a basic support module, a motion execution module, an optical detection module, a control module, and a risk point database and detection strategy template library associated with the control module.

[0070] The basic support module is used to provide stable support and seismic protection, including a concrete base, gantry frame and seismic components, to provide a rigid foundation for system operation;

[0071] The motion execution module is electrically connected to the basic support module and is used to drive the detection component to achieve multi-degree-of-freedom motion, including a four-track motion system and a multi-axis gimbal array. The multi-axis gimbal array is equipped with an industrial camera to achieve full coverage of the detection area and can fine-tune the gimbal attitude according to the surface normal vector of the risk point.

[0072] The optical detection module is electrically connected to the motion execution module and is used to acquire clear images of the stamped parts. It includes an industrial camera and a programmable multi-channel LED strip light source, which can switch the lighting mode and parameters according to the surface features and defect types of the stamped parts.

[0073] The control module is electrically connected to the motion execution module and the optical detection module respectively, and is used to coordinate system actions and data processing. It includes a mold feature parameter library, a PLC motion control unit, a host computer control system, and the risk point database and detection strategy template library.

[0074] The risk point database stores the three-dimensional coordinates, surface normal vectors, cracking mechanism information, geometric features, defect types, and surface material reflectivity of each risk point in the stamped part.

[0075] The detection strategy template library associates a predefined detection strategy template with each risk point. The template contains the risk point coordinates and normal vector, risk level, defect type, and detection parameters corresponding to the material's reflectivity, enabling adaptive detection.

[0076] The basic support module specifically includes:

[0077] Concrete base unit: Used to support the overall weight of the system, prevent foundation deformation during system operation, and ensure the initial positioning accuracy of the detection components;

[0078] Gantry unit: Made of high-strength steel, with built-in prestressed tie rods, and installed with precise calibration to form a stable support frame, providing a foundation for the horizontal movement of the motion execution module;

[0079] Vibration damping unit: Used to suppress vibration interference, including dampers, vibration-damping camera brackets and motor vibration isolation pads, to prevent vibration from affecting the image acquisition accuracy of industrial cameras;

[0080] The system adopts a multi-dimensional stability design consisting of a concrete base, a prestressed gantry frame, and multi-component vibration damping. This design avoids foundation deformation through a rigid structure and specifically suppresses production line vibration through dampers and vibration-damping supports. This solves the problems of image blurring and pose shift caused by vibration in existing systems, ensuring the stability of detection accuracy.

[0081] The motion execution module specifically includes:

[0082] Four-track motion unit: High-strength steel rails with hardened surface are used to drive the multi-axis gimbal array to move horizontally, providing basic displacement support for risk point coverage;

[0083] Multi-axis gimbal unit: A single gimbal supports horizontal and vertical rotation. The body adopts a lightweight structure and is equipped with dynamic balance correction components. It can calculate the optimal observation posture based on the surface normal vector of the risk point and drive the gimbal servo motor to fine-tune it to the minimum angle between the camera optical axis and the normal.

[0084] Servo drive unit: including servo motor and reducer, adopts full closed-loop control, drives four-rail motion unit and multi-axis gimbal unit to move precisely, and ensures the accuracy of risk point positioning;

[0085] The innovative design of the "four-track motion + multi-axis gimbal" multi-degree-of-freedom collaborative structure breaks through the limitations of traditional fixed or low-degree-of-freedom motion; by calculating the optimal attitude through normal vectors, the camera can achieve "direct observation" risk points, solve the problem of blind spots in the field of view in areas such as edges and curved surfaces, and at the same time, the full closed-loop control ensures positioning accuracy.

[0086] The optical detection module specifically includes:

[0087] Industrial camera unit: used to identify microcracks, for selecting industrial cameras with suitable resolution and field of view;

[0088] LED light source unit: contains multiple strip light sources, arranged differently according to the inspection area of ​​the stamped parts, and switches the light source form according to the type of defect at the risk point. When detecting microcracks, the double-sided strip light is turned on, and when detecting highly reflective curved surface defects, the high-angle light source is turned on.

[0089] Light source control unit: Supports multi-channel dimming, adjusts the light source intensity according to the reflective characteristics of materials at risk points, and works with industrial camera exposure parameter control to avoid overexposure and underexposure of images;

[0090] It achieves precise matching of "light source form - defect type - reflective characteristics", breaking through the limitations of traditional single light source; it switches the light source form for different scenarios such as micro-cracks and highly reflective curved surfaces, and combines multi-channel dimming and camera parameter coordination to solve the problems of overexposure, underexposure and failure to highlight defect features.

[0091] The mold feature parameter library unit of the control module specifically includes:

[0092] Parameter storage sub-unit: Parameters are stored in categories according to mold ID. Each mold parameter contains multiple sets of camera pose schemes, and is also associated with the risk point database and detection strategy template library data corresponding to the mold.

[0093] Parameter call subunit: After receiving the mold change signal, read the current mold ID and quickly call up the corresponding parameter group and associated risk point detection strategy template without manual intervention;

[0094] Parameter update subunit: Supports adding and modifying mold parameters and risk point strategy templates via the host computer. After the update, it is automatically synchronized to the PLC motion control unit to adapt to the detection of new types of stamping parts;

[0095] Dynamic optical compensation subunit: During the camera movement, the angle between the camera and the surface of the risk point of the stamping part is calculated in real time, and the LED light source illumination angle is adjusted synchronously to ensure uniform illumination in the detection area and adapt to the geometric features of the risk point.

[0096] Multiple exposure control subunit: Images of high reflectivity risk areas are acquired multiple times with different exposure levels. Reflectivity artifacts are eliminated by image fusion. At the same time, the corresponding image processing algorithm is matched according to the type of defect at the risk point. Microcracks are treated with edge enhancement algorithm and wrinkles are treated with grayscale contrast algorithm.

[0097] Defect Output Unit: Feeds the detection results back to the production line MES system in real time, while storing image data and detection logs, supporting historical data traceability, and marking defect priorities according to risk level;

[0098] Cooperative motion control unit: Adopting a track-gimbal cooperative motion model, and through an automatic pose mapping algorithm, combined with risk point coordinates and normal vector data, the industrial camera can cover the detection area of ​​traditional multiple fixed cameras, improving workspace utilization.

[0099] Precision calibration subunit: Periodically triggers precision calibration, detects motion errors through the laser tracker, automatically compensates to the servo drive unit, and calibrates the angle accuracy between the camera optical axis and the normal of the risk point to ensure long-term stability of positioning and observation attitude;

[0100] It integrates multiple core functions such as dynamic optical compensation, multiple exposure, and automatic parameter calling and updating to solve problems such as excessive manual intervention during mold changes, optical interference, and wasted workspace. Through algorithm matching with risk point defect types and priority labeling of detection results, it realizes intelligent and efficient detection process, while regular automatic calibration ensures long-term accuracy.

[0101] The predefined logic of the detection strategy template library is as follows:

[0102] Based on the cracking mechanism analysis, geometric features, defect types, and surface material reflectivity of the risk points, a template containing the following parameters is generated:

[0103] Observation attitude parameters: the target angle between the camera optical axis and the normal to the risk point, and the set values ​​of the gimbal horizontal angle and pitch angle;

[0104] Light source parameters: light source type, light source intensity level, illumination angle;

[0105] Camera parameters: exposure time, gain, resolution, focus distance;

[0106] Priority parameters: detection order and number of images captured.

[0107] A multi-degree-of-freedom reconfigurable online visual inspection method for stamped parts includes the following steps:

[0108] S1: After the die change signal is triggered, the control module reads the current stamping die ID, retrieves the corresponding parameter group from the die feature parameter library unit, and simultaneously retrieves the coordinates, normal vectors and associated detection strategy templates of all risk points of the die from the risk point database and detection strategy template library.

[0109] S2: The PLC motion control unit drives the four-rail motion unit to move the multi-axis gimbal array, completes the X / Y / Z axis positioning, and at the same time drives the multi-axis gimbal to fine-tune to the optimal posture where the angle between the camera optical axis and the normal of the risk point meets the requirements according to the observation posture parameters in the detection strategy template.

[0110] S3: The light source control unit switches the illumination mode according to the light source parameters in the detection strategy template. When detecting microcracks, it turns on the double-sided strip light, and when detecting highly reflective curved surfaces, it turns on the high-angle light. It adjusts the intensity and angle of the LED light source to ensure uniform illumination in the detection area.

[0111] S4: The industrial camera performs automatic focusing, sets the exposure time and gain according to the camera parameters in the template, and acquires images of the stamped parts. During the acquisition process, the host computer analysis unit adjusts the light source angle in real time through the dynamic optical compensation subunit and performs multiple exposures on high reflective risk areas.

[0112] S5: The host computer analysis unit calls the image processing algorithm that matches the type of defect at the risk point, processes the acquired image, identifies defects such as cracks and wrinkles, and generates an inspection report with risk point priority.

[0113] S6: If it is necessary to switch molds, repeat steps S1-S5; if the same mold is continuously inspected, after inspecting a preset number of workpieces, the control module will automatically trigger precision calibration, and the motion error and camera posture will be calibrated by the laser tracker to ensure inspection stability.

[0114] S2 specifically includes:

[0115] S201: The four-track motion unit first drives the multi-axis gimbal array to move to the mold detection starting position, and then performs error compensation after it is in place;

[0116] S202: The multi-axis gimbal array adjusts the horizontal angle to the target value first, and then adjusts the pitch angle according to the gimbal horizontal angle and pitch angle settings in the detection strategy template.

[0117] S203: The servo drive unit feeds back the motion positioning signal. The PLC motion control unit compares the actual position of the camera with the template setting value. If the deviation exceeds the allowable range, it triggers a secondary adjustment until the requirements are met.

[0118] S204: After the motion is completed, the control module sends a ready signal to the optical detection module. At the same time, it sorts the risk points to be detected according to their risk level. High-risk points are given priority to enter the detection queue and wait for the image acquisition command.

[0119] The light source adjustment logic of S3 is as follows:

[0120] When the defect type of the risk point is a microcrack, the light source control unit turns off the default light source, turns on the dual-side strip light, and adjusts the light source intensity to make the crack produce a shadow effect.

[0121] When the risk point defect type is wrinkling and necking, turn on the high-angle light source, adjust the intensity, and adjust the lighting angle in combination with the curved surface geometry to highlight the wrinkles and thickness changes;

[0122] When the surface material of the risk point is a bare metal with a highly reflective coating, reduce the light source intensity and extend the camera exposure time to avoid reflections from covering up the defects;

[0123] When the risk point is a sharp or concave corner, turn on the dual-angle strip light to eliminate the blind spot of the corner shadow.

[0124] The defect detection logic in step S5 is as follows:

[0125] The host computer analysis unit first performs fusion processing on the multi-exposure images, and then calls the corresponding algorithm according to the type of risk point defect:

[0126] When detecting microcracks, an edge enhancement-based algorithm is used to extract linear features, which are then compared with a microcrack size threshold to determine whether they are defects.

[0127] When detecting macroscopic cracking and wrinkling, an image segmentation algorithm based on region growing is used to identify abnormal areas and determine the defect type by combining geometric morphology.

[0128] After the inspection is completed, the defects are marked according to the risk level, which includes high risk, medium risk and low risk. High-risk defects are pushed to the production line early warning terminal in real time, while medium and low-risk defects are summarized in the periodic report.

[0129] In summary, the advantages of this invention are as follows:

[0130] This invention utilizes a "four-track motion system + multi-axis gimbal array" to construct a multi-degree-of-freedom motion execution module. Combined with the three-dimensional coordinates of risk points and surface normal vectors stored in the risk point database, it can drive industrial cameras to achieve horizontal movement and omnidirectional rotation. This ensures that high-risk areas such as edges, concave and convex corners, and complex curved surfaces of stamped parts can obtain the optimal observation posture of "camera optical axis facing the surface". At the same time, with the workspace utilization algorithm of the collaborative motion control subunit, it can cover the detection range of traditional fixed camera arrays with fewer industrial cameras. This significantly reduces equipment investment costs and completely solves the detection blind spot problem caused by fixed viewing angles or low-degree-of-freedom motion mechanisms, greatly improving the defect identification coverage.

[0131] This invention innovatively designs a "risk point database + detection strategy template library," matching a dedicated detection scheme for each risk point: based on the cracking mechanism analysis, geometric features, expected defect type, and surface material reflectivity of the risk point, it predefines the camera observation posture, light source working mode, camera operating parameters, and detection priority. For example, when detecting microcracks, it automatically activates dual-sided strip lighting; when detecting highly reflective areas, it automatically performs multiple exposures; and high-risk points automatically prioritize entering the detection process, truly achieving "one risk point, one solution." Simultaneously, in conjunction with the light source angle adjustment algorithm of the dynamic optical compensation subunit, it can eliminate illumination blind spots on different geometric surfaces, effectively improving the identification of defect features in the image. This ensures the detection accuracy of high-risk points while avoiding the waste of detection resources for low-risk points, achieving an optimal balance between detection accuracy and efficiency.

[0132] The mold feature parameter library of the control module of this invention can associate corresponding risk point data and detection strategy templates according to mold ID. When the production line triggers a mold change signal, the system can automatically retrieve the detection parameter group of the current mold without manual operation, drive the motion execution module to complete the camera pose adjustment, and simultaneously switch the light source mode and camera parameters to achieve automated connection of "start detection upon mold change". Compared with the existing technology that requires manual parameter adjustment and secondary calibration, this invention significantly shortens the mold change adaptation time, reduces manual intervention, and lowers labor costs, especially suitable for multi-variety, small-batch stamping parts production scenarios.

[0133] The basic support module of this invention, through a combination design of "concrete base + gantry frame + shock-absorbing components," can effectively suppress mechanical vibration interference in the stamping production line, avoiding camera pose shift and image blurring caused by vibration. The light source control unit and dynamic optical compensation subunit of the optical detection module work together to adjust the light source intensity and angle in real time according to the surface material characteristics of the stamped parts, eliminating high-reflectivity artifacts or underexposure problems. Simultaneously, the precision calibration subunit can automatically trigger the laser calibration process periodically, compensating for motion errors in real time and ensuring long-term stable detection accuracy. Compared to existing technologies that require frequent manual calibration and have high maintenance costs, this invention significantly reduces the frequency of maintenance operations and the probability of equipment failure, reducing the investment of maintenance resources.

[0134] This invention supports adding or modifying mold parameters and risk point detection strategy templates via a host computer through the parameter update subunit of the control module. It can adapt to stamping parts of different sizes and structural types without replacing hardware equipment. At the same time, the multi-axis gimbal and four-track motion system of the motion execution module can adjust the motion trajectory through program reconstruction. The programmable LED light source of the optical detection module supports switching between multiple lighting modes, realizing the flexible detection capability of "one set of detection equipment adapting to multiple types of products", significantly improving the equipment reuse rate and reducing the equipment investment cost of enterprises with multiple production lines.

[0135] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A multi-degree-of-freedom reconfigurable online visual inspection system for stamped parts, characterized in that, include: The system includes a basic support module, a motion execution module, an optical detection module, a control module, and a risk point database and detection strategy template library associated with the control module. The basic support module is used to provide stable support and shock protection; The motion execution module is electrically connected to the basic support module and is used to drive the detection component to achieve multi-degree-of-freedom motion, including a four-track motion system and a multi-axis gimbal array. The multi-axis gimbal array is equipped with an industrial camera to achieve full coverage of the detection area and can fine-tune the gimbal attitude according to the surface normal vector of the risk point. The optical detection module is electrically connected to the motion execution module and is used to acquire clear images of the stamped parts. It includes an industrial camera and a programmable multi-channel LED strip light source, which can switch the lighting mode and parameters according to the surface features and defect types of the stamped parts. The control module is electrically connected to the motion execution module and the optical detection module respectively, and is used to coordinate system actions and data processing. It includes a mold feature parameter library, a PLC motion control unit, a host computer control system, and the risk point database and detection strategy template library. The risk point database stores the three-dimensional coordinates, surface normal vectors, cracking mechanism information, geometric features, defect types, and surface material reflectivity of each risk point in the stamped part. The detection strategy template library associates a predefined detection strategy template with each risk point. The template contains the risk point coordinates and normal vector, risk level, defect type, and detection parameters corresponding to the material reflectivity characteristics, enabling adaptive detection. The basic support module specifically includes: Base unit: Used to support the overall weight of the system, prevent foundation deformation during system operation, and ensure the initial positioning accuracy of the detection components; Gantry unit: Used to stabilize and support the frame, providing a foundation for the horizontal movement of the motion execution module; Vibration damping unit: Used to suppress vibration interference, including dampers, vibration-damping camera brackets and motor vibration isolation pads, to prevent vibration from affecting the image acquisition accuracy of industrial cameras; The motion execution module specifically includes: Four-track motion unit: used to drive the multi-axis gimbal array to move horizontally, providing basic displacement support for risk point coverage; Multi-axis gimbal unit: A single gimbal supports horizontal and vertical rotation. The body adopts a lightweight structure and is equipped with dynamic balance correction components. It can calculate the optimal observation posture based on the surface normal vector of the risk point and drive the gimbal servo motor to fine-tune it to the minimum angle between the camera optical axis and the normal. Servo drive unit: including servo motor and reducer, adopts full closed-loop control, drives the four-rail motion unit and multi-axis gimbal unit to move precisely, ensuring the accuracy of risk point positioning.

2. The multi-degree-of-freedom reconfigurable online visual inspection system for stamped parts according to claim 1, characterized in that, The optical detection module specifically includes: Industrial camera unit: used to identify microcracks, for selecting industrial cameras with suitable resolution and field of view; LED light source unit: contains multiple strip light sources, arranged differently according to the inspection area of ​​the stamped parts, and switches the light source form according to the type of defect at the risk point. When detecting microcracks, the double-sided strip light is turned on, and when detecting highly reflective curved surface defects, the high-angle light source is turned on. Light source control unit: Supports multi-channel dimming, adjusts the light source intensity according to the reflective characteristics of materials at risk points, and works with industrial camera exposure parameter control to avoid overexposure and underexposure of images.

3. The multi-degree-of-freedom reconfigurable online visual inspection system for stamped parts according to claim 1, characterized in that, The mold feature parameter library unit of the control module specifically includes: Parameter storage sub-unit: Parameters are stored in categories according to mold ID. Each mold parameter contains multiple sets of camera pose schemes, and is also associated with the risk point database and detection strategy template library data corresponding to the mold. Parameter call subunit: After receiving the mold change signal, read the current mold ID and quickly call up the corresponding parameter group and associated risk point detection strategy template without manual intervention; Parameter update subunit: Supports adding and modifying mold parameters and risk point strategy templates via the host computer. After the update, it is automatically synchronized to the PLC motion control unit to adapt to the detection of new types of stamping parts; Dynamic optical compensation subunit: During the camera movement, the angle between the camera and the surface of the risk point of the stamping part is calculated in real time, and the LED light source illumination angle is adjusted synchronously to ensure uniform illumination in the detection area and adapt to the geometric features of the risk point. Multiple exposure control subunit: Images of high reflectivity risk areas are acquired multiple times with different exposure levels. Reflectivity artifacts are eliminated by image fusion. At the same time, the corresponding image processing algorithm is matched according to the type of defect at the risk point. Microcracks are treated with edge enhancement algorithm and wrinkles are treated with grayscale contrast algorithm. Defect Output Unit: Feeds the detection results back to the production line MES system in real time, while storing image data and detection logs, supporting historical data traceability, and marking defect priorities according to risk level; Cooperative motion control unit: It adopts a track-gimbal cooperative motion model and uses an automatic pose mapping algorithm to combine risk point coordinates and normal vector data to enable the industrial camera to cover the detection area of ​​traditional multiple fixed cameras. Precision calibration subunit: Periodically triggers precision calibration, detects motion errors through a laser tracker, automatically compensates the servo drive unit, and simultaneously calibrates the angle accuracy between the camera optical axis and the normal of the risk point.

4. The multi-degree-of-freedom reconfigurable online visual inspection system for stamped parts according to claim 1, characterized in that, The predefined logic of the detection strategy template library is as follows: Based on the cracking mechanism analysis, geometric features, defect types, and surface material reflectivity of the risk points, a template containing the following parameters is generated: Observation attitude parameters: the target angle between the camera optical axis and the normal to the risk point, and the set values ​​of the gimbal horizontal angle and pitch angle; Light source parameters: light source type, light source intensity level, illumination angle; Camera parameters: exposure time, gain, resolution, focus distance; Priority parameters: detection order and number of images captured.

5. A method for online visual inspection of multi-degree-of-freedom reconfigurable stamped parts, applicable to the online visual inspection system for multi-degree-of-freedom reconfigurable stamped parts as described in any one of claims 1-4, characterized in that, Includes the following steps: S1: After the die change signal is triggered, the control module reads the current stamping die ID, retrieves the corresponding parameter group from the die feature parameter library unit, and simultaneously retrieves the coordinates, normal vectors and associated detection strategy templates of all risk points of the die from the risk point database and detection strategy template library. S2: The PLC motion control unit drives the four-rail motion unit to move the multi-axis gimbal array, completes the X / Y / Z axis positioning, and at the same time drives the multi-axis gimbal to fine-tune to the optimal posture where the angle between the camera optical axis and the normal of the risk point meets the requirements according to the observation posture parameters in the detection strategy template. S3: The light source control unit switches the illumination mode according to the light source parameters in the detection strategy template. When detecting microcracks, it turns on the double-sided strip light, and when detecting highly reflective curved surfaces, it turns on the high-angle light. It adjusts the intensity and angle of the LED light source to ensure uniform illumination in the detection area. S4: The industrial camera performs automatic focusing, sets the exposure time and gain according to the camera parameters in the template, and acquires images of the stamped parts. During the acquisition process, the host computer analysis unit adjusts the light source angle in real time through the dynamic optical compensation subunit and performs multiple exposures on high reflective risk areas. S5: The host computer analysis unit calls the image processing algorithm that matches the type of defect at the risk point, processes the acquired image, identifies defects such as cracks and wrinkles, and generates an inspection report with risk point priority. S6: If it is necessary to switch molds, repeat steps S1-S5; if the same mold is continuously inspected, after inspecting a preset number of workpieces, the control module will automatically trigger precision calibration, and the motion error and camera posture will be calibrated by the laser tracker to ensure inspection stability.

6. The online visual inspection method for multi-degree-of-freedom reconfigurable stamped parts according to claim 5, characterized in that, S2 specifically includes: S201: The four-track motion unit first drives the multi-axis gimbal array to move to the mold detection starting position, and then performs error compensation after it is in place; S202: The multi-axis gimbal array adjusts the horizontal angle to the target value first, and then adjusts the pitch angle according to the gimbal horizontal angle and pitch angle settings in the detection strategy template. S203: The servo drive unit feeds back the motion positioning signal. The PLC motion control unit compares the actual position and posture of the camera with the template setting value. If the deviation exceeds the allowable range, it triggers a secondary adjustment until the requirements are met. S204: After the motion is completed, the control module sends a ready signal to the optical detection module. At the same time, it sorts the risk points to be detected according to their risk level. High-risk points are given priority to enter the detection queue and wait for the image acquisition command.

7. The online visual inspection method for multi-degree-of-freedom reconfigurable stamped parts according to claim 5, characterized in that, The light source adjustment logic of S3 is as follows: When the defect type of the risk point is a microcrack, the light source control unit turns off the default light source, turns on the dual-side strip light, and adjusts the light source intensity to make the crack produce a shadow effect. When the risk point defect type is wrinkling and necking, turn on the high-angle light source, adjust the intensity, and adjust the lighting angle in combination with the curved surface geometry to highlight the wrinkles and thickness changes; When the surface material of the risk point is a bare metal with a highly reflective coating, reduce the light source intensity and extend the camera exposure time to avoid reflections from covering up the defects; When the risk point is a sharp or concave corner, turn on the dual-angle strip light to eliminate the blind spot of the corner shadow.

8. The online visual inspection method for multi-degree-of-freedom reconfigurable stamped parts according to claim 5, characterized in that, The defect detection logic in step S5 is as follows: The host computer analysis unit first performs fusion processing on the multi-exposure images, and then calls the corresponding algorithm according to the type of risk point defect: When detecting microcracks, an edge enhancement-based algorithm is used to extract linear features, which are then compared with a microcrack size threshold to determine whether they are defects. When detecting macroscopic cracking and wrinkling, an image segmentation algorithm based on region growing is used to identify abnormal areas and determine the defect type by combining geometric morphology. After inspection, defects are labeled according to their risk level, which includes high risk, medium risk, and low risk. High-risk defects are pushed to the production line early warning terminal in real time, while medium and low-risk defects are pushed to the production line early warning terminal in a total. Periodic reports.

Images