Battery Grid Defect Detection
Inspect separating grids for defects before installation
Graphical programming environment for deep learning-based industrial image analysis
Large-format NiCd rechargeable batteries used in electric vehicles (EVs), grid storage, and industrial applications are made of blocks of multiple battery cells. Each battery cell consists of anode and cathode plates separated from each other by grids or meshes that allow the free circulation of electrolyte throughout the cell. Once sealed, such batteries can work without maintenance for years.
The separating grids are essential to maintaining battery life. Bends, gaps, and other defects in a grid will result in decreased electrical separation between the plates, electrical leakage, and reduced battery life. The surfaces of the grid or mesh are complex and fragile. Flaws resulting from manufacturing and handling can be small, have a variety of appearances, and be randomly located anywhere across the grid’s complex surface. Conventional machine vision has trouble reliably detecting this type of randomly located defect.
Once an individual cell is sealed before installation in the battery, it can no longer be inspected. If a decreased ability to hold a charge is only detected at final testing, the entire battery will need to be discarded.
Cognex Deep Learning ensures battery grids are defect-free before installation. The defect detection tool learns the appearance of a defect-free battery grid from a small set of images of acceptable separating grids. After that, the tool identifies even small defects in a battery grid, regardless of size, appearance, or location, and rejects grids with any anomaly.
If the design or pattern of the battery grid changes, Cognex Deep Learning retrains on images of the new design in a matter of minutes and be back online for inspection of the new design without any need for programming.