Braided Wire Connection Inspection
Confirm complex braided and stranded electrical connections
Graphical programming environment for deep learning-based industrial image analysis
Some high-voltage wire connections for electric vehicle (EV) batteries and other applications are made by twisting two braided copper wires or cables together without an external connector to create a tight physical and electrical contact between the strands.
Subcomponent assembly inspection requires confirming that the wires are twisted together in a way that ensures an effective electrical connection between them. A poor connection can lead to decreased current, safety risks through current leakage and overheating, and connection failure, requiring rework of finished assemblies or leading to returns by customers.
The braided wire is complex, textured, visually unpredictable, and reflective, and appears against a complex background. Conventional machine vision is unable to reliably distinguish the braiding of the wires themselves from the pattern of the connection between the wires, and there is no way to program all of the possibilities.
Cognex Deep Learning easily inspects complex, variable braided wire for defects. It trains on sets of images that include ranges of connections that have been marked both as good and unacceptable, based on functional testing. The classification tool learns the criteria for distinguishing good connections from nonfunctional connections while disregarding the complex background. Once trained, its accuracy is higher than other inspection methods without requiring more complex and time-consuming direct testing.