Autonomous Data-Driven Quality Control in Self-learning Production Systems

Shorter product lifecycles, increased individualization and disruptive technological change is said to closely correspond with the worldwide increase in production of electric vehicles and their components. Nascent production technologies, such as additive manufacturing, enable the industrial production of customized products but are often accompanied by fluctuations in product quality, as well as low process stability. This paper describes how self-learning production systems may be enabled to efficiently adapt to these disturbances through autonomous data-driven quality control. Moreover, this paper presents how the overall latency between the occurrence of an event, which directly or indirectly influences quality, and the completed implementation of process adaptions may be reduced. The core element of the presented approach is the creation of a predictive quality model from which an inverted process model and thus process adjustments can be derived. To demonstrate the proposed concept, the presented approach is applied to a Fused Deposition Modeling production system in form of model-based parameter optimization.