How Faithful are my Digital and Physical Twins?
Paula Muñoz
University of Malaga
Digital Twins design and development is becoming an increasingly complex
engineering task, and in this process, one of the main challenges is often overlooked—
how can you really prove they are twins? In this talk, we focus on measuring the fidelity
of the two twins, to ensure that both the digital and the physical twin accurately
reproduces the behavior of its counterpart. Specifically, we will explore the term
‘fidelity’, what it means in the context of digital twins and how it can be assessed.
Our approach to measuring fidelity relies on comparing the behavioral traces of two
twins using a trace alignment algorithm. We define a method to determine if two
behaviors are sufficiently similar, align their equivalent states, and measure how closely
they match. This approach allows us to compare models and identify which parts of
their behavior fail to faithfully emulate their real-world counterparts. We demonstrate
the potential of this approach by applying it to three real systems: a lift, an incubator
and a robotic arm.
Speaker bio: Paula Muñoz is a PhD candidate at the University of Malaga (Spain), specializing in Digital Twin technology. Her research began with the conceptualization and
architecture of Digital Twins, and later evolved into the field of Digital Twin validation,
where she focuses on defining fidelity and developing an approach to measure it. In
addition to her work on Digital Twins, Paula has also explored areas like uncertainty
and model-based software engineering.