The end pose accuracy of parallel robots has a significant impact on their working performance, and establishing effective calibration algorithms is an important guarantee for improving the pose accuracy of robots. A 2TPR&2TPS parallel mechanism was taken as the research object. Firstly, the kinematics of the robot was analyzed, and the error model of the robot was obtained by using the total derivative method. According to the model, the quantitative relationship between the structural parameter error of the robot and the end pose error and the influence law of the error changes of each error item on the end pose error was obtained. Subsequently, a parameter identification model was established based on the improved particle swarm optimization algorithm. The effectiveness of the parameter identification model was verified by setting a set of error values for the identified variables, and comparing the identified values with the set values five times. At the same time, a calibration effect evaluation function was established. Finally, the structural parameter error of the robot was identified with the parameter identification model, and the kinematics model of the robot was modified with the identified error value, and the error calibration of the robot was completed. The calibration effect evaluation function established was used to analyze the calibration effect. The experimental results showed that the average position accuracy of the robot after calibration was improved by 68.62%, and the average distance error was reduced from 7.710mm to 2.350mm, with an accuracy improvement of 69.52%. The experimental results proved that the calibration algorithm was effective.