The Path from Data to Actionable Information as a Driver for the Industrial Ecosystem
as
the
longest
linear
dimension
perpendicular to the Z axis), that in turn
rotates around the primary spindle of the
machine.
the data can be interpreted without complex
statistical processes.
AI and learning models will be developed at
the predictive stages of analysis with the
deployed models at the semantic stage to
classify the data earlier in the pipeline. This
approach is a typical feedback mechanism
since the full history, and the higher-level
context is unknown at this stage, and to do
so would impede the performance, function
and utility of the systems because of the
increased overhead.
When providing semantics, it is necessary to
be specific enough that one can make sense
of the data; this is where the metamodel
comes in. Each piece of data is associated
with a logical metamodel of the device or
thing that describes the components, their
relationships, constraints, and the data they
can provide. This is what is meant by device
context.
Using standards for semantics is essential.
One can sometimes make it to semantics
with proprietary data, but if one does not
use standardized semantic models at this
stage, the value of the data will be limited to
a single solution and most of the potential
will be lost. Examples of semantic standards
are MTConnect for manufacturing or
CityGML 11 for smart cities. 12
Using
standards will have a multiplicative effect on
the value of the information since there is no
way to predict the eventual use of the data
and having an open and common meaning
will protect the data collection investment.
The complexity of the analytics will
determine the complexity of the
metamodels. With a complete digital
surrogate or twin of the device, it is
necessary to have a more complex
metamodel that may refer back to the
geometry of the parts of an assembly and
various systems engineering models, given
in standards such as SysML, that provide the
first principles expectations of their
behavior.
There are many ways to create semantics;
these range from explicit – identifying the
meaning of the data based on an
understanding of the “thing” and its
function; or implicate – using AI to perform
feature recognition and classification of the
data to determine the meaning. The
selection of technology and methodology
will depend on the nature of the data and if
11
With semantics, there is still little actionable
insight since the data lacks the context of
additional business systems and information
sources. The next stage interprets the data
within the context of the use of the
equipment.
https://www.citygml.org/
12
R. Kaden *, T. H. Kolbe. 2013. "City-Wide Total Energy Demand Estimation of Buildings Using Semantic 3D City
Models and Statistical Data." ISPRS 8th 3DGeoInfo Conference & WG II/2 Workshop. Istanbul, Turkey: ISPRS Annals
of the Photogrammetry, Remote Sensing and Spatial Information Sciences.
IIC Journal of Innovation
- 27 -