By measuring the hysteresis loop of shaft parts, the change of its feature can be used to describe the the surface hardness and case depth. It is one of the most promising technologies for nondestructive testing. The key of this technology is to develop the measuring devices and research the high precision identification method. This paper design a hysteresis loop measurement device for shaft parts based on the closed magnetic circuit. A Genetic Algorithm and Particle Swarm Optimization (GAPSO) hybrid algorithm is proposed to identify the parameters based on JA model, which can realize the fast and accurate identification of the global and local characteristic parameters of hysteresis loop. According to the measured hysteresis loops of three different kinds of steel material, the consuming time and accuracy of parameter identification are compared and analyzed among the proposed hybrid algorithm and other algorithms (genetic algorithm; particle swarm optimization; simulated annealing algorithm). The experimental results show that the minimum root mean square error of the global identification results of the hybrid algorithm is only 0.004 7, which is lower than the corresponding results of other algorithms. The relative error of the local feature parameters (coercivity and residual magnetic induction) identification results of the hybrid algorithm is less than 0.35%, which is smaller than other algorithms. The experimental measurements and parameters identification method can be expected to apply for the nondestructive testing for surface hardened layer of shaft component, e.g., dowel pins, bolts.