Self-adaptive digital twin reference architecture to improve process quality

Abstract

The aim of software-defined manufacturing is a more adaptable production through software. This requires highly adaptive, versatile cyber-physical production systems (CPPSs), which react flexibly to changes and perform crucial adjustments. Self-adaptive digital twins (DTs) enable adaptability through monitoring the behavior of a CPPS, identifying problems, and determining adjustments for the optimization of processes. DTs work with an idealized representation of the CPPS. Changes in system behavior that negatively impact the process execution, e.g., caused by tool wear or variations of environmental influences, and, thus, the quality of the resulting product, are ignored. This ignorance leads to defective products, sub-optimal product quality, and hazardous CPPS states. To mitigate this, we present an approach to incorporate quality awareness into a model-driven reference architecture of a self-adaptive DT. The devised concept automatically analyzes and optimizes system behavior. To this end, a component for quality evaluation extends the reference architecture. The DT and CPPS are connected via OPC UA. To validate the approach, a fiber laser machine is utilized. The DT can operate both on a physical and a virtual laser machine.