Kinematics calibration of redundant parallel robot based on Sparrow search algorithm optimization neural network

In recent years, the redundant drive parallel mechanism has been widely considered by academia and industry because of its remarkable advantages in workspace, singular pose and stiffness. However, machine errors and assembly errors of parts will lead to motion errors and affect the accuracy of task completion. To improve the motion accuracy of this kind of mechanism, a kinematic calibration method based on Sparrow search algorithm (SSA) is proposed to optimize BP neural network. This method combines the global search ability of SSA and the learning ability of BP neural network and avoids the problem of local optimal solution by optimizing the initial weight and threshold of the neural network, to achieve more accurate error compensation. A three-degree-of-freedom redundant parallel mechanism is selected as the research object, and the kinematics calibration experiment is designed. The results show that the SSA-BP neural network can significantly reduce the mean error and maximum error of the mechanism, which proves the effectiveness of the proposed method. Moreover, this method does not need to establish a complex error model, reduces the calibration cost and improves the motion accuracy of the mechanism.