Decision support in manufacturing processes using ML-driven knowledge repositories for process parameter initiation

The problem of losing institutional knowledge after the departure of experienced employees is prevalent in many companies. To tackle this challenge, this paper introduces Design of Experiments (DoE) as an established industrial practice for identifying process-inherent relationships. Considering the limitations of DoE in terms of mapping non-linear behaviors, machine learning (ML) models are presented as powerful repositories of process knowledge. The paper demonstrates the suitability of ML for parameter recommendations and accurate target projections and emphasizes its role in guiding decision-making processes, by utilizing ML-driven knowledge repositories for intuitive visual support in quality management.