Identifying ground stress based on rock core discing can save a lot of surveying cost, so studying “rock core disicing” has great meaning. Aiming at the limitation of conventional methods, the particle swarm optimization (PSO) arithmetic is introduced to be combined with numerical simulation, so as to identify the ground stress based on rock core disicing observation. The objective function of optimization is constructed by combing Mohr-Coulomb criterion and tension criterion. Adopts dynamically adjusting momentum term method and steric compression method so as to improve the PSO in convergence velocity and globally searching property. The proposed method is mentioned and used to identify the ground stress of Bakken formation in USA. The arithmetic has good convergence speed and the identified results are coincident with the former reported results.