Structural optimization and dynamic calibration for load sharing dynamic force measurement platform based on MIGA and SVR

Abstract

In order to quantify dynamic forces/moments on the ground, a high-performance platform is necessary. This article improves the performance of the measurement platform from both structure and calibration. The platform’s loading capacity and base frequency have been significantly increased by the introduction of an innovative load sharing ring, which also allows the platform to have other important and practical characteristics. Further improvements in platform performance are then achieved by optimizing the platform parameters using a multi-island genetic algorithm (MIGA). In addition, the calibration method is improved. The maximum relative error and cross-coupling error for the novel non-linear support-vector-regression-based (SVR) method are 3.42% and 6.54%, respectively, which are much superior than the 9.98% and 15.39% for the traditional linear least-squares-based (LS) calibration method, indicating better accuracy and decoupling performance of the SVR-based dynamic calibration. This has important implications for measuring dynamic forces and improving the stability of equipment in spacecraft.