Využití hybridní techniky modelování pro predikce chování technologických procesů

technological process control. When designing optimal
control philosophies, it is advantageous to use accurate, but
on the other hand very mathematically complex models that
give the results of the solution with a long time lag. their off-
-line use can show of driving directions, but it is not possible
to use them for real-time control of technological processes.
some existing technological process control technologies
allow for the use of simpler models and artificial neural
networks (aNN), for which a simple and time-consuming algorithm for typing is typical. the success of using aNN-
-based models is conditional on their effective learning on
data that includes all possible options that are realistic in the
management of the chosen technology. applying both types
of models, we can get a so-called hybrid model, where a demanding,
accurate model is used to teach a simple neural
model in off-line mode. a simple neural model can then be
implemented into real-process control algorithms.