Coloured Petri Nets – Tool for Control Systems Learning

Abstract

Control systems become increasingly more complex, and sometimes the sequence of technological process so frequently changes, that the creation of the programs of control becomes problematic. One of the ways is to use learning systems. For purposes of learning the principles of multi-agent systems or the methods of artificial neural nets are used. The principles of multi-agent systems are most acceptable in the real devices, but the process of learning creates failures and damages the technological process. Artificial neural nets solve the same problem by probabilistic methods, so that during the learning the same problems appear. For purposes of learning coloured Petri nets are well suited if we use the global variables in them, which create the possibility to form, accumulate and preserve data of instruction. The model of system in the coloured Petri nets consists of two large parts; these are the model of the program of control and the model of technological process. The learning of system is conducted on the model and the already trained program of control is transferred to the control device. Ill. 4, bibl. 10 (in English; summaries in English, Russian and Lithuanian).