How Democratized Artificial Intelligence Can Move Manufacturing to a New Evolution Pace
final product quality issue detection,
production
performance
fluctuation
detection, safety and security issue
detections and help you to react as soon as
possible before having a worse situation to
manage. For those two detections of AI
technologies, it is important to understand
that we must change our usual way to
capture and record data from real-life. The
usual way, currently implemented in
Manufacturing Execution Systems, is to filter
the raw data to record only the abnormal
behaviors
detected
by
predefined
engineering models and set up thresholds on
filtered indicators that you want to monitor.
This approach was implemented to reduce
the cost of data communication and storage
and emphasized the limited computing
power which was not able to process large
amounts of data to provide useful insights.
capability to process and analyze the data, it
makes the historic data management
approach
obsolete
and
dangerous.
Dangerous, because it is not giving the new
Predictive and Machine Learning AI the
opportunity to provide their power and
valuable outcomes. And dangerous because
the new manufacturing world of introducing
new high-quality products quickly to the
market – produced on new equipment, with
new worker skillset, new raw materials and
new suppliers – cannot wait to meet and
identify all potential issues at a “human”
ramp up speed, to know and place them
under control. Now, we must continuously
learn, predict and react at a digital speed to
constantly adjust the excellence practices to
achieve our short-term and long-term goals.
If you want to setup a new production line,
you now have a few months to deliver your
product and optimize your costs. While
before now, you had a few years.
Now, with new technologies of cheap data
capture, communication and recording,
associated with an almost “unlimited”
Figure 4: Example of prediction and machine learning analytics embedded in an Asset Monitoring platform
- 12 -
November 2017