AI analysis of tester failures

This PoC aimed to detect when a tester is a faulty one and not the product it is testing. The predictability of a tester equipment would avoid unplanned production downtime and increases productivity as unfaulty products are not returned to the fixing. The proof-of-concept approaches the challenge from analysing the data testers are producing and applying machine learning methods to find anomalies. First stage was to understand why things failed and the second was to explore if predictability could be implemented on top of the failure detection. A thesis work has been also done in this PoC.

Partners: Nokia, GE Healthcare, VTT
Keywords: AI, data analysis, IoT
Contact: Mikko Lindholm /VTT
Digital Production