Ever tried wrapping a box in transparent plastic?
Wrinkles and uneven corners were likely the result, putting your patience to
the test. All the more astonishing are the machines from Sollas Holland – a
world player in the development of high-end packaging machines, which can wrap
160 boxes in plastic per minute – with perfectly smooth surfaces and crisp
corners every time. This attention to detail makes Sollas’ machines a favorite
among producers of high-end cosmetics. But the benchmark of 160 boxes per
minute is new; until recently, the machines were able to wrap “only” 120 boxes
per minute. To go from 120 to 160 boxes per minute was a challenge. The
machine’s mechanics could handle it, but not the packaging foil.
The objective of this Sollas–Siemens co-creation
project was to use AI to create a function block in TIA Portal to better
control the packaging machine.
Artificial
intelligence (AI) is used to optimize the control of a Sollas packaging
machine. But first, the AI was trained on a Digital Twin of the machine.
An
AI module independently and quickly built a logical function block to
optimize the control of a packaging machine.
The neural network that was trained by the AI can be
deployed on a Simatic S7-1500 through the TIA Portal directly. Simatic S7-1500
then sends the setpoints to the Sinamics Servo Drives that Sollas relies on.
We wanted to go from 120 to 160 products per minute.
That was when we started to wonder if AI might be able to help.
Dirk Verbeek, Engineering Manager, Sollas Holland
Thanks
to AI, the machine’s throughput was increased by 33 percent, without a single
adjustment to the physical hardware.
Much
faster than a human could ever do it, the AI developed a logical algorithm
to better control the machine.
Unwinding
of the packaging foil is now more consistent. This saves costs and extends
the life of the machine.
The
end product is guaranteed to be of consistently high quality
Sollas
has set up a roadmap to develop into an OEM of the future, and AI is part
of that.
This technology effectively automates the development
of complex automation applications and drastically reduces engineering effort
and commissioning time.