Project background

Cylindrical lithium batteries are a type of lithium battery with mature technology, high degree of automation, stable product quality, and relatively low cost. Therefore, they have a very high market share and a wide range of applications. During the production process, various defects inevitably appear on the surface of the battery, affecting the quality of the product and the assembly of the battery pack.

At present, the surface defects of batteries mainly rely on manual inspection, which has low efficiency and high missed detection rate, making it difficult to meet the needs of automated production. Therefore, in order to improve the speed and accuracy of cylindrical battery detection, Huahan Weiye has flexibly customized a machine vision defect detection scheme for cylindrical lithium battery end face defects, targeting the detection of appearance defects of cylindrical batteries.

Appearance defect detection of cylindrical batteries

1. Difficulties in detection

Before wrapping the metal shell of a cylindrical battery in a vinyl coating, surface defects need to be detected. Not only does it require the detection system to detect normal changes and minor abnormalities in the battery casing, but it also needs to identify all severe scratches.

Because the size and shape of each defect are slightly different, using traditional rule-based visual algorithms for detecting such objects has a relatively high rate of missed and false detections, and low detection efficiency. In addition, due to the reflective nature of the object surface, the cylindrical shape of the battery can sometimes be rendered as a blurry and unfocused image, making detection more complex.

（Original image display）

2. Testing items

On the side of the cylindrical battery shell: large pits, linear defects, scratches, hidden blocks, rust, water stains, dirt, etc.

3. Solution

Huahan AIDI Suite Deep Learning Intelligent Detection System

（Lighting diagram）

4. Test results

1) Huahan Weiye's flexible customization solution can accurately locate surface defects and anomalies on the side, top, and bottom of cylindrical batteries, while ignoring irrelevant changes. The 2.5D+AI software can accurately identify damaged battery casings, improve detection accuracy, and save detection costs.

2) Can stably detect defects above 0.4mm, with a zero miss rate and an error rate of ≤ 0.1%.

（Presentation of renderings）

Project Summary

The customized solution from Huahan Weiye can detect and identify surface defects such as dents or deformations on the metal surface, scratches on the metal surface, damage or wrinkles on the coating, abnormal coating size, oxidation of the metal surface, leakage of the positive electrode, welding penetration at the bottom of the negative electrode shell, and protruding welding points, greatly improving the efficiency and accuracy of inspection, and helping enterprises reduce costs and increase efficiency.

The solution for the appearance defects of cylindrical batteries can be applied in scenarios such as surface defects on the positive and negative electrodes of lithium batteries, surface defects on white glass sheets, surface defect detection after the blue film of battery packs, and surface defect detection of soft pack batteries. It is also widely applied and can meet the strict project delivery time and testing quality requirements of manufacturing enterprises. The effect has been widely recognized by customers.