Digital Twin-Driven Technological Process Monitoring for Edge Computing and Cloud Manufacturing Applications

Abstract

AbstractThe use of the cloud-trained digital twin at the edge opens up new possibilities for autonomous systems, such as novel real-time artificial intelligence applications. The edge-deployed digital twin can continually improve through self-learning, while ensuring that its cloud counterpart is synchronized and up-to-date. When running a digital twin on the edge, applications, for example, device protection features, including shutoff, can be used instantly if digital twin analysis detects or predicts a hazard. Analytics created by the digital twin can help local control and vice versa. This leads to proactive control applications that ensure autonomous operation. Edge computing offers clear advantages if there are problems with low latency, connectivity, security, privacy, and data transfer. The trade-offs are relatively high initial costs and maintenance costs. On the other side of the scale is cloud manufacturing, where the initial cost and maintenance overhead are average, but latency, connectivity, and the amount of data transferred can be a problem with the associated costs. With IoT, manufacturing companies will increasingly manage data, and cloud and edge computing will provide a high level of “smartness” to their operating environment. Using techniques such as online machine learning on data streaming and real-time learning, the digital twin can continuously self-learn and evolve.