In: Performance Evaluation Review, Vol. 26, No. 2, pages 5-14. 1998.
Abstract: This paper examines the (semi) automatic generation of a hierarchical structure for generalized stochastic Petri nets (GSPNs). The idea is to partition a GSPN automatically into a set of components with asynchronous communication. Net level results obtained by invariant computation for these subnets are used to define a macro description of the internal state. This yields a hierarchical structure which is exploited is several efficient analysis algorithms. These algorithms include reachability set/graph generation, structured numerical analysis techniques and approximation techniques based on decomposition and aggregation. A GSPN model of an existing production cell and its digital control is analyzed to demonstrate usefulness of the approach.
Keywords: Kronecker algebra, Markov chains, hierarchical analysis, stochastic Petri nets.