A novel error compensation method for a redundant parallel mechanism based on adaptive differential evolution algorithm and RBF neural network

Abstract

Accuracy plays an important role in the advanced industrial application of parallel mechanisms, and the related research is valuable. Taking a (6 + 3)-DOF kinematic redundant parallel mechanism (KRPM) as an example, a novel error compensation method is proposed. Firstly, the error model of the KRPM is established based on vector method. Secondly, adaptive differential evolution (ADE) algorithm and radial basis function neural network (RBFNN) are introduced, and they are combined as a novel error compensation method. Numerical simulation is carried out based on three cases, and compared with typical error compensation method, which verifies the effectiveness of the proposed method. Finally, the experiments are conducted based on laser tracker to verify the proposed method. Under the proposed method, the average effect of error prediction can reach 90.2 %, which is better than the typical error compensation method. After error compensation, the pose error of the KRPM is significantly reduced.