TY - CONF
T1 - Sequential Co-Design of an Artifact and Its Controller Via Control Proxy Functions
AU - Peters, Diane L.
AU - Papalambros, Panos Y.
AU - Ulsoy, A. Galip
N1 - Optimization of a 'smart' product requires optimizing the design of both the artifact and its controller. If the artifact and control optimization are coupled, then a combined approach is typically used in order to produce optimal solutions. The combined approach presents certain disadvantages, however.
PY - 2010/9/15
Y1 - 2010/9/15
N2 - Optimization of a 'smart' product requires optimizing the design of both the artifact and its controller. If the artifact and control optimization are coupled, then a combined approach is typically used in order to produce optimal solutions. The combined approach presents certain disadvantages, however. In particular, it precludes the decomposition of the problem into smaller functional sub-problems, and requires the formulation of both the artifact and control objectives and constraints before solving either optimization problem. In this paper, it is shown that a modified sequential approach utilizing a Control Proxy Function (CPF) can be used to produce optimal, or near-optimal, solutions while allowing this decomposition. Two physical bases for CPFs are presented, natural frequency and the controllability Grammian matrix, and their range of applicability is discussed. These concepts are demonstrated on a positioning gantry example.
AB - Optimization of a 'smart' product requires optimizing the design of both the artifact and its controller. If the artifact and control optimization are coupled, then a combined approach is typically used in order to produce optimal solutions. The combined approach presents certain disadvantages, however. In particular, it precludes the decomposition of the problem into smaller functional sub-problems, and requires the formulation of both the artifact and control objectives and constraints before solving either optimization problem. In this paper, it is shown that a modified sequential approach utilizing a Control Proxy Function (CPF) can be used to produce optimal, or near-optimal, solutions while allowing this decomposition. Two physical bases for CPFs are presented, natural frequency and the controllability Grammian matrix, and their range of applicability is discussed. These concepts are demonstrated on a positioning gantry example.
UR - https://www.sciencedirect.com/science/article/pii/S1474667015374711?via%3Dihub#!
U2 - 10.3182/20100913-3-US-2015.00041
DO - 10.3182/20100913-3-US-2015.00041
M3 - Presentation
T2 - 5th IFAC Symposium on Mechatronic Systems, 2010
Y2 - 15 September 2010
ER -