In: Jensen, K.: Lecture Notes in Computer Science, Vol. 616; 13th International Conference on Application and Theory of Petri Nets 1992, Sheffield, UK, pages 279-298. Springer-Verlag, June 1992.
Abstract: A methodology is presented for the approximate solution of large Stochastic Petri Nets that are structured into independent submodels. These subnets are aggregated to substitute nets of lower complexity to achieve a state space reduction. This is based on the estimated traffic process at a submodel's input in steady state and on a token's residence time distribution in the original submodel as equvalence criterion for matching the substitute network to the submodel. The Markov renewal process at the input of the submodel is approximated by a renewal process. Its moments and the arrival instant probabilities at the submodel are computed by means of a traffic set approach. The technique is applied to Generalized Stochastic Petri Nets and compared to Flow Equivalent Aggregation.