A surface defect detection system for metal products

Abstract

This utility model discloses a surface defect detection system for metal products, including an operating table, an intelligent lighting unit, and an intelligent detection module. The operating table is equipped with a loading tray, a unloading tray, and a collaborative robot. The intelligent lighting unit includes a ring-shaped light source base, a multi-color light source, and a first camera, with the multi-color light source and the first camera positioned on the inner ring surface of the ring-shaped light source base. The detection model of the intelligent detection module is obtained by inputting image data of a perfect test product and a test product with multiple defects into a deep learning model for training. This detection system can improve the efficiency of metal product defect detection.

Description

Technical Field

[0001] This utility model relates to the field of defect detection technology, and in particular to a surface defect detection system for metal products. Background Technology

[0002] Surface defects are external flaws in products, differing from normal product surfaces. These differences are often discernible to the human eye under suitable lighting conditions. Therefore, in traditional industrial production, manual visual inspection is the most common method for detecting surface defects. However, due to the limitations of human attention and the unavoidable contact involved in the inspection process, manual inspection no longer meets the needs of current metal production and has even become a significant obstacle to productivity improvement. How to efficiently detect surface defects in metals online has become an urgent issue for all metal manufacturers.

[0003] With the advancement of science and technology and the further development of new industrialization, the metal production and inspection process can also be combined with technological concepts such as the Industrial Internet of Things, artificial intelligence, and computer vision.

[0004] Current machine learning-based defect detection systems collect defect data and train the system on it. However, if order volumes for a particular product are low, the precision metalworking industry may not have enough defective products to collect. To address this issue, a new solution is needed that doesn't target defects specific to any particular product. Summary of the Invention

[0005] The technical problem to be solved by this utility model is to provide a surface defect detection system for metal products. The intelligent detection module of this system uses only data collected from the "perfect sample" of the current metal product and the dataset of defective products from other products to train the detection model. For the precision metal processing industry with a wide variety of products but low output, it can perform efficient quality inspection.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by this utility model is: a metal product surface defect detection system, including an operating table, an intelligent lighting unit and an intelligent detection module. The operating table is equipped with a loading tray, a unloading tray and a collaborative robot; the intelligent lighting unit includes an annular light source base, a multi-color light source and a first camera, wherein the multi-color light source and the first camera are disposed on the inner ring surface of the annular light source base.

[0007] Furthermore, the collaborative robot has a robotic arm capable of grasping and moving objects and rotating at any angle. The multi-color light source can selectively emit monochromatic light or striped light. The operating platform also has a positioning area between the loading tray and the annular light source base. A positioning camera is installed on one side of the positioning area, and a bottom camera is installed below the positioning area for positioning the metal products in the loading tray. The bottom camera is used to capture images of the bottom of the metal products. The metal products are aluminum, copper, or steel products. The unloading tray includes a qualified product unloading tray and a defective product unloading tray. The intelligent lighting unit also includes a second camera. The detection model of the intelligent detection module is obtained by inputting image data of perfect test products and test products with multiple defects into a deep learning model for training.

[0008] This invention proposes a surface defect detection system for metal products, used to detect defects in different materials (aluminum, copper, or steel) with varying surface roughness. The collaborative robot of the detection system exposes the surface of the metal product to a camera and can easily move the product for inspection. An intelligent lighting unit can find the optimal light for different metal products and can also emit striped light for defect detection. The detection model data of this application collects defect data from perfect and imperfect products, making the detection model more comprehensive and overcoming the problem of lacking imperfect models for certain special products. For the precision metal processing industry, which produces a wide variety of products but has low production volumes, this system can improve detection efficiency. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this utility model or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings described below are only some embodiments of this utility model, wherein:

[0010] Figure 1 This is a schematic diagram of a surface defect detection system for metal products. Detailed Implementation

[0011] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention. The structures, proportions, sizes, etc., shown in the drawings are only for illustrative purposes to aid those skilled in the art and are not intended to limit the implementation conditions of the present invention. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in proportions, or adjustments to size, without affecting the effects and objectives of the present invention, should still fall within the scope of the technical content disclosed in the present invention. Furthermore, the terms such as "upper," "lower," "left," "right," "middle," and "one" used in this specification are only for clarity of description and are not intended to limit the scope of the present invention. Changes or adjustments to their relative relationships, without substantially altering the technical content, should also be considered within the scope of the present invention.

[0012] like Figure 1 As shown, a surface defect detection system for metal products includes an operating table A, an intelligent lighting unit B, and an intelligent detection module.

[0013] The operating platform is equipped with a collaborative robot 7, a loading tray 1, a defective product unloading tray 8, and a qualified product unloading tray 9. The product to be tested is placed on the loading tray 1. The collaborative robot 7 has a robotic arm 10, which can grasp the product to be tested, move it, and rotate it at any angle.

[0014] The intelligent lighting unit B includes a ring-shaped light source base 4, a multi-color light source 5, and a first camera 12. The multi-color light source 5 and the first camera 12 are disposed on the inner ring surface of the ring-shaped light source base 4.

[0015] The detection model of the intelligent detection module in this application is obtained by inputting image data of both a perfect product under test and a product under test with multiple defects into a deep learning model for training. The deep learning model is a mature model in the existing technology.

[0016] The robotic arm 10 is used to grasp the product to be tested from the feeding tray 1 and place it into the annular light source base 4. The multi-color light source 5 emits light of different colors, and the operator determines the color of light that will provide the clearest image for different metal products. Under the illumination of the optimal color light, in conjunction with the rotation of the robotic arm, the first camera takes pictures of the product to be tested from different angles.

[0017] In some embodiments, if the robotic arm of the collaborative robot cannot completely flip the product under test and one side is always unable to be photographed, the operating table also has a positioning area 10 between the loading tray and the all-around illumination area of ​​the light source. A positioning camera 2 is set on one side of the positioning area 10, and a bottom camera 3 is set below the positioning area for positioning the metal product in the loading tray. The bottom camera 3 is used to capture an image of the bottom of the metal product.

[0018] In some embodiments, the multicolor light source 5 can also emit striped light for defect detection, and the imaging deformation of the striped light can detect defects.

[0019] In some embodiments, the detection system can target aluminum, copper, or steel parts with dimensions smaller than 500mm x 500mm x 500mm. In some embodiments, the intelligent lighting unit may also include a second camera positioned opposite the first camera.

[0020] The specific detection methods of the above detection system are as follows:

[0021] S1 constructs a learning model by collecting image data of the perfect test product and other test products with various defects, performing data augmentation and deep learning on the image data, and constructing a machine learning-based detection model.

[0022] S2 Selecting an appropriate light source: The collaborative robot grasps the product to be tested and places it in the illumination area of ​​a multi-color light source. The multi-color light source illuminates the product using different colors of light, selecting the color that produces the clearest image.

[0023] S3 Image Acquisition: The product under test is exposed to appropriate colored light, the collaborative robot grasps the product under test and flips it from various angles, and the first camera captures images of different sides of the product under test;

[0024] S4 Comparison Detection: The intelligent detection module collects multiple image data of the product under test and compares the image data with the detection model to perform defect detection;

[0025] S4 Inspection Result Classification: The collaborative robot's gripping device places qualified products into the qualified product unloading tray and unqualified products into the defective product unloading tray.

[0026] This invention proposes a surface defect detection system for metal products. The detection model collects defect data from both perfect and imperfect products, making the model more comprehensive. An intelligent lighting unit can emit multiple colors of light and identify the optimal light for different products. The collaborative robot's mechanical gripper can rotate in any direction and grasp parts of different shapes. For the precision metal processing industry, which produces a wide variety of products but has low production volumes, this system and method can improve detection efficiency.

[0027] The above description is only a preferred embodiment of the present utility model and is not intended to limit the present utility model. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present utility model should be included within the protection scope of the present utility model.

Claims

1. A surface defect detection system for metal products, comprising an operating table, an intelligent lighting unit, and an intelligent detection module, characterized in that, The control panel is equipped with a loading tray, a unloading tray, and a collaborative robot. The intelligent lighting unit includes a ring-shaped light source base, a multi-color light source, and a first camera, wherein the multi-color light source and the first camera are disposed on the inner ring surface of the ring-shaped light source base.

2. A metal product surface defect detection system according to claim 1, wherein The collaborative robot has a robotic arm that can grasp and move objects and rotate at any angle.

3. The metal product surface defect detection system of claim 1, wherein The multicolor light source can selectively emit monochromatic light or striped light.

4. The metal product surface defect detection system of claim 1, wherein The operating platform also has a positioning area between the feeding tray and the annular light source base. A positioning camera is set on one side of the positioning area and a bottom camera is set below the positioning area for positioning the metal products in the feeding tray. The bottom camera is used to capture the bottom image of the metal products.

5. The metal product surface defect detection system of claim 1, wherein The metal product is made of aluminum, copper, or steel.

6. The metal product surface defect detection system of claim 1, wherein The feeding tray includes a qualified product feeding tray and a defective product feeding tray.

7. The surface defect detection system for metal products according to claim 1, characterized in that, The intelligent lighting unit also includes a second camera.

8. The surface defect detection system for metal products according to claim 1, characterized in that, The detection model of the intelligent detection module is obtained by inputting image data of perfect test products and test products with multiple defects into a deep learning model for training.

Images