Search

CN-122016805-A - Ultrasonic welding false welding detection system and method

CN122016805ACN 122016805 ACN122016805 ACN 122016805ACN-122016805-A

Abstract

The invention discloses an ultrasonic welding false welding detection system and method, and relates to the technical field of ultrasonic welding detection, wherein the system comprises a hardware component and a visual software module, the hardware component comprises an upper camera, a lens, an annular light source, a PLC lower computer and the like, and the visual software module is configured with program instructions such as initialization, image acquisition, feature detection and the like; the detection method sequentially completes the detection of the expansion size, the grid edge width, the grid area and the peripheral flash area by collecting images through an upper camera, and automatically judges the cold joint by combining a visual algorithm. The invention realizes nondestructive online full detection, the detection precision reaches +/-0.01 mm, the single-station detection beat is less than or equal to 1 second, the problems of large waste and high erroneous judgment risk of the traditional detection products are solved, the invention is suitable for the copper-aluminum-bus metal ultrasonic welding products of the new energy power battery, and the detection efficiency and the quality control level are improved.

Inventors

  • LI SANYOU
  • ZHENG DONGHUI
  • DOU PENG
  • LI QINGPING

Assignees

  • 东莞市达瑞新能源科技有限公司

Dates

Publication Date
20260512
Application Date
20260127

Claims (10)

  1. 1. The utility model provides an ultrasonic welding rosin joint detecting system which characterized in that, includes hardware component and vision software module, the hardware component contains fixed last camera, camera lens, annular light source, light source controller, communication connecting wire and the dongle that sets up, the vision software module is furnished with system start-up, hardware initialization, image acquisition, feature detection, data processing and the interactive program instruction of communication, hardware component and vision software module realize ultrasonic welding product's rosin joint detection in coordination.
  2. 2. The ultrasonic welding cold joint detection system according to claim 1, wherein the upper camera is an industrial camera with 1200 ten thousand pixels and resolution of 4024 x 3036, the installation height is not lower than 400mm so as to avoid a material taking shaft, the lens is 25mm in specification, the long side of the visual field of the annular light source is 180+/-10 mm, and the wiring space reservation of the hardware component is not smaller than 60mm.
  3. 3. The ultrasonic welding cold joint detection system according to claim 1, further comprising a PLC lower computer, wherein the vision software module performs data interaction with the PLC through a TCP/IP protocol, and waits for an ACK acknowledgement signal of the PLC after data transmission, and the structural accuracy of the system is 0.1mm in both XY direction and Z direction.
  4. 4. The ultrasonic welding false welding detection method is characterized by comprising the following steps of: Step 1, placing a product to be detected in an acupoint of a carrier, and collecting images of the welded product from top to bottom through a fixedly arranged upper camera; Step 2, detecting the size of the product expansion material through a visual algorithm, grabbing edge protruding points of the contact position of the aluminum product, calculating the distance D between the farthest protruding points and the datum line, judging as OK when D is greater than a standard set value, and judging as NG when D is less than the standard set value; Step 3, detecting the edge width W of 6*4 grids rectangular grids formed by welding spots of the product, judging as OK when W is smaller than a standard set value, and judging as NG when W is larger than the standard set value; Step 4, calculating the area value of each internal target area of the 24 lattices, and judging whether an abnormality with too small area or multiple lattices connected exists or not according to the preset area boundary value; and 5, detecting the flash area of the rectangular periphery formed by 24 grids through a gray level comparison algorithm, and judging whether the flash is abnormal or not.
  5. 5. The ultrasonic welding cold joint detection method according to claim 4, wherein in step 3, the uniformity of the grid edge is determined by grasping the area of the grid edge with the gray level similar to that of the product, and when the black area ratio exceeds the threshold value, the product is determined to have cold joint.
  6. 6. The method for detecting the cold joint of the ultrasonic welding according to claim 4, wherein the beat of the image acquisition in the step 1 is satisfied that the total time of photographing and operation of a single station is not more than 1 second, and the detection precision of a vision system is +/-0.01 mm.
  7. 7. The ultrasonic welding cold joint detection method according to claim 4, wherein in step 4, a black area inside each lattice is identified by a visual algorithm, and whether the lattice has a break or a missing defect is judged by calculating whether the area of the black area is within a preset range.
  8. 8. The ultrasonic welding cold joint detection method according to claim 4, wherein in step 5, whether the flash area exceeds the allowable range is identified by comparing the gray value difference between the grid rectangular peripheral area and the product surface with a gray tool.
  9. 9. The method of detecting an ultrasonic welding cold joint according to claim 4, further comprising a risk coping step of judging whether to give way to reception or not according to consistency of a detection result when a lattice is connected with an unwelded portion due to irregular welding laces or a gray blind area is formed by light blocking on a vertical face of a product, and recording abnormal data for subsequent parameter optimization.
  10. 10. The ultrasonic welding cold joint detection method according to claim 4, wherein the vision software module carries an AI operation model system, the AI operation model system including an operation server configured with a training module and an operation module; The training module is used for extracting ultrasonic welding product image data, historical detection result data and defect labeling data stored in the storage server and carrying out iterative training on the operation model; The operation module is used for calculating the acquired welding image data of the product through the trained operation model, extracting characteristic information of the expansion edge, the lattice structure and the flash area in the image, and further obtaining the relevant defect data of the cold joint corresponding to the image data, wherein the defect data is sent to the storage server and the PLC lower computer through the Ethernet; The operation model in the operation server comprises: The convolutional neural network extracts depth features in the image data through a convolutional kernel and generates a feature map containing features of the expansion material size, the grid edge, the grid area and the flash region; the region proposal network is used for generating candidate frames aiming at the expansion material protruding points, the grid edge regions, the grid inner regions and the peripheral flash regions in the feature map, and primarily judging whether corresponding defects exist in the image; the interest pool is used for carrying out normalization processing on candidate frames with different sizes, converting the candidate frames into a feature map with fixed size and ensuring the consistency of subsequent calculation; The soft K mean value clustering is used for carrying out cluster analysis on defect characteristics in the candidate frames and identifying defect types, wherein the defect types comprise unqualified expansion material sizes, abnormal grid edge widths, out-of-standard grid edge uniformity, abnormal grid area and out-of-standard peripheral flash areas; after training and operation of the images of the ultrasonic welding products with the preset quantity, the soft K-means clustering dynamically adjusts the clustering quantity based on the elbow principle, so that the clustering quantity is kept at the optimal quantity, and a new false welding related defect type can be identified according to the change of the clustering quantity in the training process.

Description

Ultrasonic welding false welding detection system and method Technical Field The invention relates to the technical field of ultrasonic welding detection, in particular to an ultrasonic welding false welding detection system and method. Background The ultrasonic welding of copper-aluminum-bus metal in the new energy power battery belongs to a solid phase welding process, and an oxide film on the surface of the metal is removed through ultrasonic vibration, so that the metal forms solid phase bonding by means of interatomic attraction, and compared with the traditional fusion welding, the heat influence can be reduced. However, the solid phase bonding property of the welded joint results in that the traditional nondestructive testing methods such as X-ray detection, ultrasonic detection and the like are not applicable. At present, two detection modes are mainly adopted in the industry, namely, manual destructive spot-check testing, namely, destructive welding is identified by destroying product observation residues, so that a large amount of products are wasted, mass production is not facilitated, nondestructive detection spot-check is carried out, manual evaluation judgment is needed for X-ray detection, medium scanning is needed for ultrasonic detection, and therefore, the post-process processing is affected, and high misjudgment risk exists. The existing detection method cannot realize online full detection of products, and is difficult to meet the quality control requirement of automatic production, so that a nondestructive, high-precision and high-efficiency ultrasonic welding cold joint detection technology is needed. Disclosure of Invention In order to solve the technical problems in the background technology, the invention provides an ultrasonic welding false welding detection system and method. The invention provides an ultrasonic welding false welding detection method, which comprises a hardware component and a visual software module, wherein the hardware component comprises an upper camera, a lens, an annular light source, a light source controller, a communication connecting line and a dongle which are fixedly arranged, the visual software module is configured with program instructions of system starting, hardware initialization, image acquisition, feature detection, data processing and communication interaction, and the hardware component and the visual software module are used for realizing the false welding detection of an ultrasonic welding product in a cooperative manner. Further, the upper camera is an industrial camera with 1200 ten thousand pixels and resolution of 4024 x 3036, the installation height is not lower than 400mm so as to avoid a material taking shaft, the lens is 25mm in specification, the long side of the visual field of the annular light source is 180+/-10 mm, and the wiring space reservation of the hardware component is not less than 60mm. Further, the ultrasonic welding false welding detection system further comprises a PLC lower computer, the visual software module performs data interaction with the PLC through a TCP/IP protocol, an ACK confirmation signal of the PLC is required to be waited after data transmission, and the structural accuracy of the system is 0.1mm in both the XY direction and the Z direction. Further, an ultrasonic welding cold joint detection method comprises the following steps: Step 1, placing a product to be detected in an acupoint of a carrier, and collecting images of the welded product from top to bottom through a fixedly arranged upper camera; Step 2, detecting the size of the product expansion material through a visual algorithm, grabbing edge protruding points of the contact position of the aluminum product, calculating the distance D between the farthest protruding points and the datum line, judging as OK when D is greater than a standard set value, and judging as NG when D is less than the standard set value; Step 3, detecting the edge width W of 6*4 grids rectangular grids formed by welding spots of the product, judging as OK when W is smaller than a standard set value, and judging as NG when W is larger than the standard set value; Step 4, calculating the area value of each internal target area of the 24 lattices, and judging whether an abnormality with too small area or multiple lattices connected exists or not according to the preset area boundary value; and 5, detecting the flash area of the rectangular periphery formed by 24 grids through a gray level comparison algorithm, and judging whether the flash is abnormal or not. Further, in the step 3, the uniformity of the grid edge is judged by grabbing the area of the grid edge, which is similar to the gray level of the product, and when the black area ratio exceeds the threshold value, the product is judged to have the cold joint. Furthermore, the beat of the image acquisition in the step 1 meets the condition that the total time of photographing and operation of a single station is not