Chapter 11 - Adaptive Complex Systems: Digital Twins

Abstract

The design, control, and maintenance of complex systems are a challenge. Often it is difficult to understand the whole-system behavior because the knowledge of component behavior and interaction is uncertain. Such systems are often deployed into dynamic environments whose behavior is liable to change. This chapter reviews the features of complex systems and proposes an approach based on creating digital twins of systems that are capable of adaptation. We discuss technologies for digital twins and propose that the adaptation should be based on machine learning. We provide a simple tutorial example of agents with machine learning using our proposed technology and describe how we have used the technology to build a digital twin for supply chain networks.