In this paper, we propose a method based on independent component analysis to represent the dependence structure of structural VAR model affected by latent variables in the case of non-Gaussian. The parameters of the model are estimated by a synthesis of least-squares method and independent component analysis. Based on the results of parameters estimation, the directed acyclic graph of structural VAR model with latent variables is constructed. Finally, simulation results demonstrate that the proposed method can correctly identify structural VAR model with latent variables.