Search

CN-122016802-A - System and method for detecting all-dimensional intelligent defects on surface of workpiece

CN122016802ACN 122016802 ACN122016802 ACN 122016802ACN-122016802-A

Abstract

The invention discloses a workpiece surface omnibearing intelligent defect detection system and method in the technical field of workpiece detection, and aims to solve the technical problems of single illumination direction, limited shooting visual angle, unstable image quality, feedback lag of detection results and scattered data management of the existing detection equipment. The system comprises a detection table module, a synchronous belt sliding table vision module, a workpiece side detection module, a side light source, a six-axis mechanical arm, an annular light source, a control computer, a three-color lamp, a main frame, a baffle and a control button, wherein the automatic positioning, uniform illumination, full-surface imaging, real-time defect identification and normalized data archiving of the workpiece are realized through the cooperation of a plurality of modules. The detection coverage rate is more than or equal to 99%, the defect identification accuracy is more than or equal to 98%, the single workpiece detection period is less than or equal to 30 seconds, and the detection method is suitable for surface defect detection of workpieces such as metal, castings and the like, and meets the high-beat and high-precision detection requirements of modern manufacturing industry.

Inventors

  • WANG KAIQIANG
  • SHI JINLONG
  • Ye Diaotao
  • CHEN JIWEI
  • Wang Yongchuo
  • HE XIUJUAN
  • ZHANG DONGMING

Assignees

  • 东莞市泰锐琦五金电子有限公司

Dates

Publication Date
20260512
Application Date
20260116

Claims (10)

  1. 1. The all-dimensional intelligent defect detection system for the surface of the workpiece is characterized by comprising a main frame (1), a detection table module (4), a synchronous belt sliding table visual module (2), a workpiece side detection module (3), a side light source (5) and a six-axis mechanical arm (6), and further comprising a control computer (8) and a tri-color lamp (9) which are matched with the main frame (1); a baffle (10) is arranged on the inner side of the main frame (1), and a control button (11) is arranged on the surface of the baffle; The detection platform module (4) comprises an electric rotating platform (401), a clamp (402), a workpiece guide block (403), a rotating platform driving motor (404) and a power interface (405), wherein the rotating platform driving motor (404) is arranged below the electric rotating platform (401), and the clamp (402) and the workpiece guide block (403) are arranged above the electric rotating platform (401); The synchronous belt sliding table visual module (2) comprises a first driving motor (201), a second driving motor (202), a foot margin (203), a first industrial camera (204), a synchronous belt guide rail (205), an XY adapter (206), a coupler (207), a motor seat (208) and a third driving motor (209), wherein the first driving motor (201), the second driving motor (202) and the third driving motor (209) are fixed through the motor seat (208), and the first industrial camera (204) is installed on the synchronous belt guide rail (205) through the XY adapter (206); The workpiece side surface detection module (3) comprises a screw guide rail driving motor (301), an industrial camera II (302), a Z-axis dovetail groove sliding table (303), a camera fixing piece (304), an X-axis dovetail groove sliding table (305), a sliding table (306), a screw guide rail (307) and an alloy plating hand wheel (308), wherein the industrial camera II (302) is installed on the Z-axis dovetail groove sliding table (303) through the camera fixing piece (304), and the Z-axis dovetail groove sliding table (303) is in sliding connection with the X-axis dovetail groove sliding table (305); The six-axis mechanical arm (6) is provided with an annular light source (7) at the tail end, the lateral light source (5) is arranged on the inner side of the main frame (1), and the control computer (8) is electrically connected with the detection table module (4), the synchronous belt sliding table visual module (2), the workpiece lateral detection module (3), the lateral light source (5), the six-axis mechanical arm (6), the annular light source (7) and the three-color lamp (9) respectively.
  2. 2. The system for detecting the all-round intelligent defects on the surface of the workpiece according to claim 1, wherein an electric rotating table (401) of the detecting table module (4) can rotate 360 degrees along a Z axis, the rotating speed ranges from 0rpm to 10rpm, a driving motor (404) of the rotating table is a servo motor, the power is 400W, and a power interface (405) is a 220V alternating current interface.
  3. 3. The all-around intelligent defect detection system for the surface of the workpiece according to claim 1, wherein a driving motor I (201) and a driving motor II (202) of the synchronous belt sliding table vision module (2) are connected with a synchronous belt guide rail (205) through a coupler (207) to drive an industrial camera I (204) to move along an X axis (stroke 0-800 mm) and a Y axis (stroke 0-600 mm), and a driving motor III (209) to drive the industrial camera I (204) to move along a Z axis (stroke 0-300 mm), wherein the resolution of the industrial camera I (204) is 1600 ten thousand pixels.
  4. 4. The all-around intelligent defect detection system for the workpiece surface according to claim 1, wherein a screw guide rail driving motor (301) of the workpiece side detection module (3) drives an X-axis dovetail groove sliding table (305) to move through a screw guide rail (307), so as to drive an industrial camera II (302) to move along an X-axis (stroke 0-500 mm), and an alloy plating hand wheel (308) is linked with the X-axis dovetail groove sliding table (305) to realize fine adjustment of the Y-axis direction of the industrial camera II (302), the fine adjustment precision is 0.01mm, and the resolution of the industrial camera II (302) is 2000 ten thousand pixels.
  5. 5. The all-round intelligent defect detection system for the workpiece surface according to claim 1 is characterized in that the side light source (5) is an LED surface light source, the illumination intensity is 0-5000lux and is adjustable, the annular light source (7) is a flexible LED light source, the color temperature is 3000-6500K and is adjustable, the repeated positioning precision of the six-axis mechanical arm (6) is less than or equal to 0.05mm, and the annular light source (7) can be driven to adjust the illumination angle by 0-360 degrees.
  6. 6. The system for detecting the all-round intelligent defects on the surface of the workpiece according to claim 1, wherein a defect recognition algorithm based on deep learning is built in the control computer (8), and can recognize defects of corrosion spots, cracks, indentations, air holes, foreign matters and scratches through training of 10 ten thousand+ defect samples, and the system further comprises a data storage module, wherein a folder can be automatically created according to workpiece number/detection date, and an original image, a processed image and a detection result acquired by the industrial camera I (204) and the industrial camera II (302) are stored.
  7. 7. The all-round intelligent defect detection system for the surface of the workpiece according to claim 1, wherein the three-color lamp (9) has an audible and visual alarm function, namely, a green lamp is normally on when the detection is qualified, a red lamp is blinking and buzzing (volume is 60-80 dB) when the detection is unqualified, a yellow lamp is blinking when the equipment is abnormal, and the control button (11) comprises a start key, a pause key and a scram key.
  8. 8. The system for detecting the all-round intelligent defects on the surface of the workpiece according to claim 1, wherein the clamping mode of the clamp (402) is one of mechanical clamping, pneumatic clamping, vacuum adsorption or electromagnetic clamping, the clamping force is adjustable by 50-500N, and the positioning deviation of the workpiece guide block (403) is less than or equal to 0.1mm, so that the placing positions of the workpieces are consistent.
  9. 9. A method for detecting an all-round intelligent defect on a surface of a workpiece based on the system of any one of claims 1 to 8, comprising the steps of: Step S1, feeding and positioning, namely placing a workpiece on an electric rotating table (401) manually or by a mechanical arm, centering and positioning a workpiece guide block (403), and clamping the workpiece by a clamp (402) under the instruction of a control computer (8), wherein a rotating platform driving motor (404) drives the electric rotating table (401) to rotate to an initial angle of 0 DEG, and the control computer (8) detects a positioning state and triggers a yellow lamp of a tri-color lamp (9) to give an alarm when the workpiece is not clamped or not clamped; Step S2, illumination adjustment, namely controlling a computer (8) to turn on a side light source (5), adjusting illumination intensity to 3000lux, driving an annular light source (7) to move to a position 100mm above a workpiece by a six-axis mechanical arm (6), and adjusting the angle (30-45 ℃) and brightness (3500-4000 lux) of the annular light source (7) according to an initial image acquired by a second industrial camera (302), so as to eliminate shadows and reflections; Step S3, image acquisition, namely driving a first driving motor (201) and a second driving motor (202) of a synchronous belt sliding table vision module (2) to drive an industrial camera I (204) to move along an X axis and a Y axis, acquiring 1 piece of top surface image of a workpiece every 20mm, and 30-40 pieces of top surface image, wherein a screw rod guide rail driving motor (301) of a workpiece side surface detection module (3) drives an industrial camera II (302) to move along the X axis every 90 degrees of rotation of an electric rotating table (401), and acquiring 20 pieces of side surface images; step S4, defect identification, namely controlling a computer (8) to perform Gaussian filtering noise reduction and histogram equalization enhancement on an image, dividing a workpiece area through a U-Net network, extracting defect characteristics based on a ResNet network, identifying defect types and judging whether the workpiece area is qualified or not (crack length is more than 0.5mm, pore diameter is more than or equal to 0.3mm and judging whether the workpiece area is unqualified); step S5, result feedback and archiving, namely transmitting the detection result to a display module by a control computer (8), triggering a red light alarm of a tri-color light (9) when the detection result is unqualified, creating a folder of 'workpiece number_detection date', and archiving all images and detection results by a data storage module; and S6, resetting the system, namely loosening the clamp (402), resetting the electric rotating table (401) to 0 degrees, returning the industrial camera I (204) and the industrial camera II (302) to initial positions, closing the side light source (5) and the annular light source (7), and enabling the system to enter a standby state.
  10. 10. The method for detecting the omnibearing intelligent defects on the surface of the workpiece according to claim 9, wherein in the step S3, the acquisition frequency of an industrial camera I (204) and an industrial camera II (302) is 10 frames/second, the rotating speed of an electric rotating table (401) is 5rpm, 1 image is ensured to be acquired every 20 degrees of rotation of the workpiece, in the step S4, the defect identification accuracy is more than or equal to 98%, the detection period of a single workpiece is less than or equal to 30 seconds, in the step S2, a control computer (8) monitors the brightness of the image in real time, when the brightness deviation exceeds +/-5%, the brightness of a side light source (5) and an annular light source (7) is automatically adjusted, in the step S5, the detection result is synchronously uploaded to a cloud backup, and the searching response time is less than or equal to 1 second through the searching of the number of the workpiece.

Description

System and method for detecting all-dimensional intelligent defects on surface of workpiece Technical Field The invention relates to the field of workpiece surface detection, in particular to an omnibearing intelligent defect detection system and method for a workpiece surface. Background In the production process of modern manufacturing industry, workpiece surface defect detection is an important link for ensuring product quality. Common defects include corrosion spots, cracks, indentations, pores, foreign matters, scratches and the like, and if the defects are not recognized in time, the defects can directly influence the performance and the service life of a product, and even cause potential safety hazards. The traditional detection mode relies on manual visual or handheld cameras to shoot one by one, and then whether defects exist or not is judged manually. The method has the obvious defects of low detection efficiency, single workpiece detection time of usually more than 2 minutes, poor stability, high misjudgment rate up to more than 5 percent under the influence of experience and fatigue degree of detection personnel, and large repeated labor capacity, and is difficult to adapt to the high-beat and automatic detection requirements of 'more than or equal to 2 workpieces per minute' on a modern production line. In order to improve the automation level, the existing machine vision detection equipment gradually replaces manual work, but a plurality of technical bottlenecks still exist: 1. The illumination direction is single, the imaging quality is poor, most devices adopt top fixed light sources, the illumination direction is not adjustable, and complex curved surfaces or reflective surfaces of workpieces are easy to form shadows and highlight areas, so that the image characteristics are incomplete. For example, dark areas often appear on the arc-shaped surface of the metal casting, tiny defects such as air holes, cracks and the like are shielded, the identification precision of a detection algorithm is low, and the accuracy is generally less than or equal to 90%; 2. The shooting visual angle is limited, and the detection has dead angles, namely a single camera fixed mounting structure is adopted, and a workpiece is driven to rotate by manpower or a motor to acquire multi-angle images. The operation is complicated, positioning deviation is easy to generate, the detection coverage rate is only less than or equal to 85%, defects of the side face, the inclined face and the groove area of the workpiece are easy to miss, and full-surface detection cannot be realized; 3. The real-time performance is insufficient due to the fact that after the image acquisition, the image is manually imported into a computer for offline analysis, the delay from acquisition to output of the detection result exceeds 5 minutes, the instant feedback of a production line cannot be realized, unqualified workpieces easily flow into a next process, and the reworking cost is increased; 4. And the data management is decentralized, the traceability is poor, the detection images and the results are stored in a decentralized folder form, and a unified archiving and classifying mechanism is not formed. When the repeated history detection data is needed, the repeated history detection data is needed to be searched one by one manually, the searching efficiency is low, and the standardization and traceability requirements of production management are difficult to meet. In the prior art, although the metal surface defect detection device disclosed in the patent CN223006101U (2025, 6, 20) realizes automatic feeding and discharging through belt body conveying, a detection component is fixed, only the upper surface of a workpiece can be detected, defects of the side face and the inclined plane cannot be covered, a light source is fixed, a complex curved surface is easy to generate shadow, the image definition is insufficient, manual data transmission is still needed after detection, and a real-time identification function is not realized. The metal substrate detection equipment disclosed in the patent CN116519714B (2023, 8 and 29 of the grant bulletin date) realizes double-sided detection through a turnover mechanism, but the light source and the camera are fixed, the side surface and the complex curved surface cannot be imaged, the illumination angle and the brightness are not adjustable, the reflective surface is easy to generate a detection blind area, the detection result needs to be manually imported and analyzed, the data storage is scattered, and the standardized management is lacking. In summary, the existing detection equipment still has obvious defects in the aspects of illumination uniformity, detection comprehensiveness, imaging definition, detection instantaneity and the like. Therefore, a person skilled in the art provides a system and a method for detecting an omnibearing intelligent defect on a workpiece surface, so as to solve the