Coarsest Controllability-Preserving Plant Minimization

We define a controllability-preserving plant minimization and investigate its feasibility with respect to supervisor synthesis. A recent process-theoretic approach to supervisory control theory identified a so-called partial bisimulation preorder as a suitable behavioral relation to capture the notion of controllability for (nondeterministic) discrete-event systems. The equivalence relation, induced by the partial bisimulation preorder, can then be employed to minimize the unsupervised system, referred to as plant, by obtaining the coarsest quotient and optimize the process of supervisor synthesis. We present an efficient minimization algorithm for computing the partial bisimulation equivalence by partitioning the state space of the plant. We apply the algorithm to several experimental case studies and study the obtained gain.