The modeling of the system is discussed on Stochastic Petri net (SPN) with the transition subjected to Markov process. The determination of state space, expectation probability and duration of each state, the reliability of the system and the working intensity of maintainers are analyzed through examples. Compared with traditional modeling methods, in the course of analyzing the state space and logic relation among the states and the dynamic process of the system, SPN is discovered to be more effective than and superior to the traditional methods.