Detection and treatment of faults in automated machines based on Petri nets and Bayesian networks

In this paper, a methodology for considering detection and treatment of faults in automated machines is introduced. This methodology is based on the integration of Petri nets for diagnosis (BPN) and Bayesian networks. After that, the integration among detection/treatment of faults and the "normal" processes (represented by Petri nets, PN) is possible. This integration allows us to develop a fault tolerant supervisor, which considers all the processes in the same structure. A case study of fault tolerant AGV is considered. Finally, a simulation tool for edition and analysis of models with these characteristics is introduced.