Evolutionary synthesis of hybrid controllers

Abstract

Automatic synthesis of control systems for hybrid systems is the main goal of the current approach. Such applications are found in the cyber-physical systems. These systems usually include some hardware components endowed with sensors and actuators that integrate software components.Each of the software components provides verified control competences.The research had the goal to conceive a method capable to automatically synthesize the software that implements specified competences and join them to compose the necessary hybrid control system. Enhanced Time Petri Nets (ETPNs) are used to model the different control components that react accordingly to expected competences. The ETPNs can model concurrent structures as well as their implementation behaviors. These models can be described by an ETPN based language (ETPNL). The capability descriptions are transformed into Lisp expressions that allow the use of Genetic Programming (GP) to guide the software evolution according to some performance criteria. The evolutionary system is used to perform the controller adaptations to their specific environments. For the general purpose, the control system is composed of Discrete Event Controller (DEC) and/or Discrete Time Controller (DTC), each of them providing different kinds of competences. The control system synthesis of an electric hydro power plant composed of a lake and two generators is used to show the application of this approach.

Modelling Business Processes Using Evolutionary Generated Petri Nets

“…Books have been published on theoretical and practical aspects of using DE in parallel computing, multi objective optimization, constrained optimization, and the books also contain surveys of application areas. Excellent surveys on the multi-faceted research aspects of DE can be found in journal articles like [11,17,18,19]. -NP, F, CR, N: described in section 6 -X: initial population (vector) -fcost: function returning fitness of current solution For our purposes the DE/rand/1 algorithm mutation has been choosen.…”