|
The interest in the issues of artificial intelligence of many different centers around the world has
brought specific results that have already found practical and widespread applications. "Artificial
intelligence" has become increasingly popular and more frequently used in recent years. The
rapid development of electronics and computer science is conducive to the development of this
field of science. "Intelligent machines" are needed by humans to create and discover new
relationships, and artificial intelligence has significantly influenced various areas of technology
and, above all, has achieved great implementation success in areas where a large amount of data
is available. It is also increasingly used in the processing of polymer materials. The designs of
screws of plasticizing systems of extruders are largely proprietary and there is very little specific
information available in the literature. Our team generated a set of many extrusion screw designs
using computer simulation software for the extrusion process, including the transport of solids,
melting and pumping of the melt. The parameters and results obtained were entered into four
machine learning algorithms. The performance of the four algorithms was evaluated by
comparing the predictions of each algorithm with the corresponding simulation results. For three
of the algorithms, we obtained satisfactory performance, and the best one was additionally
evaluated using a previously “unseen” dataset consisting of two screws with defined diameters.
It is argued that the same ML methodologies can be applied to datasets of existing real-world
screw designs. Our conclusion is that the data-driven methodology presented in this paper can
be used in the design of extruder screws, leveraging knowledge gained from the development of
existing designs.
|
