Time to embrace MODERN digital twins!

When we deployed the first industrial scale real-time “IoT" and “Digital Twin” systems in USA in 2000, our digital twin methods were elementary. (BTW, IoT or Digital Twin was not a thing then! We called our digital twin “Enlive” – as the customer could monitor their machines live).

Turns out things have not changed a lot in the 2+ decades since! Most “machine learning” (ML) in digital twin (DTs) involves thresholding of amplitudes and assessment of FFT peaks; we call these “brittle” estimates because they produce a lot of false alarms and missed alarms . . .

ML has come a long way since the year 2000. Modern ML methods are much more stable, reliable and useful. The table below is a simple comparison of then and now . . .

* List of ML methods is not exhaustive; these ones can be found in TwinARC digital twin

Data has been compared to oil: “Data is the new Oil”! You can’t use oil without refining and extracting useful “fractions”. Similarly, we need Modern DT to extract information and knowledge from data to provide useful outputs for IIoT-related decision making.

At the top of the DT “ladder”, we have Causal ML-based digital twin – this is a game changer! Causality IS scientific knowledge, here abstracted as a “mechanism” with quantified cause-effect relationships among parts – they can be experimented with and simulated in software to perform Root-cause and What-if analysis without interfering with plant operations and measuring new data!

This 2-min video will give you glimpse of how this complex (but high business value!) technology can be applied to your IoT use case. Knowing the cause-effect relationships among parts of your system is the knowledge needed to truly understand and improve performance . . .

PG

?

#twinarc #digitaltwin #iot #knowledgedigitaltwin