Stator Assembly Weld Inspection
Assess hairpin and leadpin welds with deep learning solutions
Graphical programming environment for deep learning-based industrial image analysis
Bar-wound copper pins, called hairpins due to their shape, have replaced wire windings in the stator of many electric motors. They are more rigid than wire so their orientation can be controlled more precisely, leading to higher and more predictable efficiency. Hairpins, or leadpins, are loaded into slots on the stator and welded together to become one single twisted conductor. Welding can introduce inclusions and porosity, increasing electrical resistance as well as reducing mechanical strength. More significant defects can break the circuit and make the entire stator nonfunctional.
Welds have a large amount of variation, both for cosmetic defects that do not affect performance, and for performance-reducing defects that show few overt signs. The weld can have too much or too little volume, be inadequately fused, or show signs of cracking. Hairpin welding inspection must detect all possible defects.
A range of cameras can be used to image the weld for analysis. Although a 3D camera may be required to measure weld volume, a 2D camera can supply the images for all other defect detection and ensure a proper positioning of the spot welder before the process starts.
Cognex Deep Learning’s defect detection and classification tools are trained using a small image set of good welds and a wide variety of defective welds to classify and detect defects.