Session: MSNDC-02-01: Flexible Multibody Dynamics

Paper Number: 97837

97837 - Adjoint Method for Time-Optimal Control of Flexible Multibody Systems Using the Absolute Nodal Coordinate Formulation 

The increasing industrial relevance of innovative lightweight construction requires the consideration of flexible components in mechanical systems. Multibody systems with flexible components are usually underactuated systems and the optimal control problem becomes much more complicated in comparison to full actuated systems.

The present work discusses time-optimal control problems of an underactuated flexible multibody system and derives an adjoint gradient method considering a discrete control parametrization. The adjoint gradients will be used here for a direct optimization scheme, e.g. in the sense of a Sequential Quadratic Programming (SQP), to provide the analytically computed gradients in an efficient way concerning computational effort in order to avoid the elaborate numerical differentiation. The approach is open to various discrete control parametrization strategies.

The advantages of the proposed approach will be exploited by a highly flexible robotic system using the absolute nodal coordinate formulation (ANCF) for describing large deformations. The goal is to determine smooth drive signals such that the tool center point of a flexible two-arm robot reaches a predefined state in a minimal operation time. When comparing the usage of the numerical gradients and the analytically adjoint gradients in the optimization routine of the presented application, a tremendous reduction of function evaluations and computational time can be observed when using the adjoint gradients.

Presenting Author: Daniel Lichtenecker Technical University of Munich

Presenting Author Biography: 2014-2017 Bachelor degree in Mechanical Engineering (University of Applied Sciences Upper Austria, School of Engineering)<br/>2017-2019 Master degree in Mechanical Engineering (University of Applied Sciences Upper Austria, School of Engineering)<br/>2019-2020 Research Assistant (Johannes Kepler University, Institute of Robotics)<br/>since 2021 Research Assistant (Technical University of Munich, Chair of Applied Mechanics)<br/>since 2021 Doctoral candidate (Technical University of Munich, Chair of Applied Mechanics)

Authors:

Daniel Lichtenecker Technical University of Munich
Philipp Eichmeir University of Applied Sciences Upper Austria
Karin Nachbagauer University of Applied Sciences Upper Austria