Abstract: Parallel robots often encounter motion singularities, which can lead to motor drive torque overload, causing overheating and reduced lifespan of the drive motors. Therefore, it is necessary to develop a method for accurately estimating the robot's drive torque based on the preset trajectory to reduce the failure rate of the drive motors. This paper takes the 6-PSU type parallel robot as the research object and proposes a solution for accurately estimating drive torque based on precise dynamic modeling and parameter identification. First, by considering the effects of ball screw modules and motor friction and inertia, and combining the Kane method, an inverse dynamics model for the resolution of the servo motor drive torque is established, resulting in an identifiable model corresponding to the minimum identifiable parameter set. To improve the accuracy and speed of the identification model for actual working conditions, in terms of end output trajectory optimization, viscous friction, and Coulomb friction are considered to reduce the condition number of the observation matrix of the identification model; in terms of structure, the symmetry of the linkage section is considered, and the linkage inertia that can be ignored is discovered, resulting in a simpler identification model. Using this model, a drive torque equation is constructed, which is used to estimate the driving force accurately. A physical simulation model of the robot was built using Simulink/Multibody, and inverse dynamics simulation experiments were conducted. The drive force estimation accuracy and noise resistance of the drive torque equations constructed by the identifiable model of the minimum identifiable parameter set and the simplified identification model obtained above were quantitatively analyzed. Finally, experiments were conducted on the 6-PSU type parallel robot platform, and the results show that the simplified drive torque equation proposed in this paper greatly simplifies the model complexity while ensuring the same estimation accuracy, which has a clear engineering value.