Authors: Gasper Music
The paper deals with Petri net based modelling and optimizationof scheduling problems. Timed Petri net modelsare derived and used in conjunction with various optimizationstrategies. Standard Petri net scheduling approachesare applied, which include heuristic dispatching rules aswell as heuristic based search through the reachabilitytree. As an alternative a simulation-based optimizationis implemented to optimize the input sequences. Variousconflict resolution strategies are used in order to compareand evaluate possible operation schedules in the modelledsystem. The strategies are based on predefined orderingrules, i.e. sequences of transition firings, which arechanged during optimization procedure. The optimizationproblem is then solved by heuristic algorithms, includinggenetic algorithms, simulated annealing and thresholdaccepting. All these methods are implemented in the socalled MATLAB PetriSimM toolbox which offers, amongothers, an implementation of an automated model buildingfor the scheduling purposes. A number of benchmark testsis performed in order to compare various optimizationstrategies and to illustrate the suitability of the relatedapproaches for solving practical problems.