Stochastic Petri Net (SPN) are a powerful modeling and analyzing tool for system performance evaluation.But the problem of state space's explosion of SPN limits its ability to analyze complex and large-scale systems.So it is more feasible to simplify SPN model on the basis of performance equivalence before analyzing it.In this paper

the authors define a class of SPN-Elementary SPN (ESPN) which is composed of four elementary subnets including sequence subnet

parallel subnet

choice subnet and iteration subnet

and propose a group of performance equivalence formulas for the four elementary subnets

as well as a method of performance equivalence simplification and analysis for ESPN with linear time complexity.In addition

two transforming rules are put forward to transform non-elementary subnets into combined elementary subnets such that the proposed algorithm can be used in analysis of more general SPN.