Tuesday  |  Salon 8  |  09:00 AM–09:20 AM

#18643–Greedy Source Placement Optimization for MIMO Control of Lab-based Vibration Tests

In this work, we investigate the effectiveness of greedy algorithms in optimizing the placement of electrodynamic shakers to achieve optimal control in Multiple-Input Multiple-Output (MIMO) vibration tests. In the context of ground-based vibration tests, control algorithms are used to estimate excitation signals that will produce a specific vibration response in the test object. The challenges in control lie in estimating the drive signal, which is inherently an inverse problem, and in satisfying the limited input energy available from the shaker’s power. To this end, we investigate an inverse problem formulation that allows the user to weight their preference for matching target response data versus minimizing the demand on input power. The success of the proposed algorithm depends critically on the shaker locations, where poor placement results in diminished controllability resulting in unachievable drive powers and(or) lack of proper control of the system.

Optimizing the placement of sources is a Mixed Integer Programming (MIP) optimization problem, known for its computational expense and complexity. Greedy algorithms present a potential solution by trading optimality for enhanced computational performance. In our research, we conduct a comparative analysis between a greedy algorithm and Gurobi, a powerful optimization tool for solving MIP problems, to evaluate the greedy algorithm's capability in identifying effective shaker locations.

Our results demonstrate that the greedy algorithm often matches solutions to the MIP program. In addition, our study indicates that the ideal placement of the shakers is influenced by the user’s preference for minimizing input energy versus achieving target response accuracy. This highlights the importance of customizable parameters in achieving optimal system control through strategic source placement.


SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525

Chandler Smith   Sandia National Laboratories
Timothy Walsh   Sandia National Laboratories
Wilkins Aquino   Duke University
Ryan Schultz   Sandia National Laboratories
Drew Kouri   Sandia National Laboratories