Information retrieval for AI-supported product repair and re-design

Repair data of former product generations has enormous potential for supporting the detection of product weaknesses but is often not available as useful information in product design. In this paper, a novel method for intelligent information retrieval from repair data is presented. Based on unsupervised machine learning techniques, this method identifies failure-relevant patterns and dependencies of interconnected parts and assemblies. These gained insights give a valuable contribution to the systematic re-design of next-generation products. Moreover, the approach supports efficient repair and sustainable operational processes. The validation of the artificial intelligence-based assistance system is shown for welding equipment.