Dynamic model updating of an asteroid probe considering experimental condition uncertainty

Abstract

A multi-body dynamics analysis model of an asteroid probe is established to validate the attachment performance. In the established dynamics model, the contact between the footpad and the contact surface is modeled by a normal nonlinear spring-damping and a tangential Coulomb friction. The dynamics performance of the buffer material is simulated with a force spline that changes with respect to deformation. To improve the fidelity of the simulation model, critical parameters of the attachment mechanism’s buffer components are modified. These include the equivalent rotational stiffness, the stiffness coefficient and the nonlinearity index in the collision model. These adjustments are informed by the outcomes of ground-based attachment tests, ensuring a more precise representation of the actual attachment process. Considering the inherent inaccuracy of the experimental equipment, the attachment terrain parameters in the test are refined using the generalized gradient descent method. This optimization process precedes the enhancement of the probe’s dynamics model, ensuring a more accurate simulation that accounts for real-world variations and measurement errors. Furthermore, the optimized Latin hypercube sampling method is used to construct a surrogate model based on Radial Basis Function (RBF) neural networks to reduce the computational cost. Finally, the multi-island genetic algorithm (MIGA) is employed to update the parameters of the probe by minimizing the discrepancy between the maximum contact force obtained from simulations and that measured in actual tests. The updated parameters are then integrated into the dynamic model for validation, resulting in a significant improvement in performance metrics, with an average error reduction from 50.6% to 11.8%.