How Democratized Artificial Intelligence Can Move Manufacturing to a New Evolution Pace
Then you will be able to adjust your
production planning and slow down or
accelerate for the next production
period based on your targets of overall
efficiency, costs, quality and demand
satisfaction indicators. In manufacturing
some tools are using, for instance,
simple linear regression 13 and ARIMA 14 .
identify abnormal behaviors and
patterns in a set of data qualified as
normal data thru mostly two
approaches,
supervised
and
15
unsupervised models . The purpose of
these approaches is to detect a
repeatable
and
non-repeatable,
unexpected,
undesirable
behavior
hidden in an expected, normal behavior.
This AI domain is now also included in
the Digital twins 16 approach for Assets
and Process.
2. Machine learning (ML) is the second AI
area also embedded in manufacturing
business
applications
(predictive
maintenance, production monitoring,
human resources, finance, supply chain,
marketing and sales). This technology is
using a large set of historical data to
Let’s go back to our predictive
maintenance case; you already know and
set up your monitoring thresholds for a
Figure 2: Example of now + 10 minutes data prediction approach
13 https://en.wikipedia.org/wiki/Linear_regression
14 https://en.wikipedia.org/wiki/Autoregressive_integrated_moving_average
15 https://blog.clearbrain.com/posts/the-two-types-of-machine-learning
16
Digital twin refers to a digital replica of physical assets, processes and systems that can be used for various purposes. The
digital representation provides both the elements and the dynamics of how an Internet of Things device operates and lives
throughout its life cycle.
http://innovate.fit.edu/plm/documents/doc_mgr/912/1411.0_Digital_Twin_White_Paper_Dr_Grieves.pdf;
http://research3.fit.edu/camid/documents/doc_mgr/1221/Origin%20and%20Types%20of%20the%20Digital%20Twin.pdf
- 10 -
November 2017