For the modeling issue of non-uniform period refresh and sampling system, the subspace identification method is used to deal with the state space model. First of all, Hankel matrix created by the sampled input and output data is employed to consist with the extended state space equation. Then, the subspace identification with the oblique projection principle and sigular value decomposition is used to determine the augmented observation matrix and state vector. The least squares
method is proposed to confirm the parameters of the model. Finally, the simulation example demonstrates the effectiveness of the proposed algorithm.