Enhancing the time Petri nets for automatic hybrid control synthesis

Hybrid control refers to applications having a discrete event part and a continuous one that involves the interactions of different model types. Their control synthesis is difficult due to the fact that they contain models that belong to different approaches. In the current study the Time Petri Nets (TPNs) are used for discrete event controlled part model. For the continuous plant model is used a discrete time system (DTS) and the Fuzzy Logic Control (FLC) for its control. A new proposed method links the TPN models with FLC models. The formal Time Petri Net based Language (TPNL) (used for TPNs description) is enhanced to comprise the FLC functions and thus the entire hybrid control system can be described. The hybrid control descriptions constructed with the extended TPNL are transformed into trees that are further processed by Genetic Programming for control structure synthesis. The controller´s parameters are improved by Genetic Algorithm (GA). The evolutionary individuals are organized in species taking into account the isomorphic distance between them. The species evolution implemented by GP is alternated by its adaptation performed by GA. New methods are used for the bloat control leading to a higher speed of solution search.