Abstract: A discrete time Markov chain model for the system by probabilistic model checking is established in this paper, and the failure rate is added to the subsequent state migration of the system. Compared with cloud platform flexible computing function, another distributed power is added in this model to verify its impact on the system process. The model describes the attribute of the system by probabilistic computation tree logic, and verifies the load-cost, economic cost and the number of state migration of the system quantitatively. The experimental results show that the failure rate of the distributed power-supply in microgrid has different effects on the systems revenue and consumption. In addition, the flexible application of the elastic computing idea of cloud platform in microgrid can improve the performance and production benefit of microgrid, which is of great significance for practical production and application