The decelerator tester twisting distortion caused angular measurement error calibration, based on the IBSCF-SGDPSO-RBF method

Abstract

Recently, robot decelerators have beenwidely applied in the automation industry. Thus, the robot decelerator feature measurement, especially rigidity measurement, is essential to estimating the decelerator. However, the angular measurement results of the instrument cannot be used as the decelerator's actual twisting distortion due to the influence of the distortion in the measurement chain. The instrument's twisted distortion characteristics are analyzed in this paper. Based on the analysis, a new method of calibrating the twisted-distortion-caused angular measurement error based on the improved B-spline curve fitting-stochastic gradient descent and particle swarm optimization-radial basis function neural network (IBSCF-SGDPSO-RBF) method is proposed. The steps of the calibration method are presented in this paper, and the measurement experiment from low load to high load is carried out on the full scale. The experiment results show that the instrument distortion caused angular measurement error after compensation is within ± two angular seconds. We can conclude that the problem that the change of instrument distortion affects the angular measurement accuracy of the instrument is solved. The calibration method can eliminate the influence of the instrument's twisting distortion. The authors' main contribution is that a CSF-iPSO-RBF-based angular measurement error calibration method is proposed. It provides a reference for measuring and evaluating the actual twisting rigidity of the Rotary Vector (RV) decelerator under any load.