Decentralized supervisory control of discrete event systems without loss of information
Ahmed Khoumsi · 2017
Conventional decentralized control of discrete event systems consists in using local supervisors that observe locally the plant and compute local enabling/disabling decisions; the latter are transmitted to fusion modules that combine them to generate global decisions that are applied to the plant. We propose an unconventional approach where the local supervisors do not compute local decisions, they rather transmit all their local observations to fusion modules that combine them to generate global decisions. The proposed architecture is qualified as information loss-free (ILF), because all the information observed by the local supervisors is available to the fusion modules for making their global decisions. We justify the ILF control approach and prove that if the plant is slower than the control system, then: ILF control is equivalent to a combination of several centralized control architectures, and ILF control is more general than any conventional decentralized architecture. We also show that the slowness of the plant is necessary in any control architecture.