The motion characteristics recognition of the space objects is the basis for other on-orbit actions of space objects. Given the high costs associated with obtaining images labeled with motion features in space, we propose a method for constructing a dataset to recognize the poses of space objects in orbit through digital simulation. This approach aims to evaluate the feasibility of using intelligent recognition technology to identify the motion characteristics of space objects. The existing datasets utilized for estimating the kinematic attributes of celestial objects, exemplified by SwissCube and SpeedPlus, inadequately integrate the authentic reflective characteristics and illumination conditions of operational space targets, thereby yielding simulations of inferior realism. Furthermore, these datasets take into account solely the relative orientation between the target and the observational platform, without amalgamating it with the practical inertial J2000 coordinate framework, rendering them incompatible with actual on-orbit working scenarios. This method establishes the relationship between the Sun, the target, and the observation platform within the J2000 inertial system through dynamic simulation. The material of the target model is simulated by digital simulation, and the measured bidirectional reflectance distribution function (BRDF) reflection characteristic parameters are used to achieve the fidelity of the dataset. Finally, the relationship between the target and the observation platform in the inertial frame is transformed into the relative pose, enabling the digital image simulation and label production of space targets with different motion characteristics. The classical pose recognition methods Wide-Depth-Range 6D Object Pose Estimation (WDR) and Spacecraft Pose Network (SPN) are used to verify the dataset. The experimental results show that the construction method studied in this work is feasible and can serve as a reference for recognizing the motion postures of space targets.