Estimation of parameters in a lumped mass, spring and damper model of a single stage mechanical system

Estimation of parameters in a lumped mass, spring and damper model of a single stage mechanical system

Parthasarathy, H.

Van Wingerden, J.W. (mentor)
Dietz, S. (mentor)

Mechanical, Maritime and Materials Engineering

Delft Center for Systems and Control

2016-09-23

Photo-lithographic stepper systems are used for manufacturing of integrated chips(IC). For incorporating more functionality and having more space chip manufacturing industries invest heavily to reduce the size of the structures on the chips. This directly requires the stepper system to operate in nanometer scale resolution. To achieve such high accuracy, the stepper system should be isolated from the external vibrations. Isolators are the auxiliary systems that protect the device from floor vibrations. In order to better understand how to tweak isolator components to boost the vibration isolation performance a parameterized multi body model of the system is the first step. From a system theoretical perspective, estimating the parameters of such a multi-body model is a problem on gray-box system identification. The state-of-the art approaches in this field estimates both the state transformation matrix and the system parameters. This increases the number of decision variables in the optimization problem. Very few results exist in which more than 6 system parameters have been estimated accurately. In this thesis we develop a framework for gray-box system identification comprising of PBSIDopt algorithm and H-infinity static output feedback synthesis. This approach bypasses the estimation of state transformation matrix. Also this optimization addresses the input output behavior directly which makes it conceptually simpler. We also explain how to judge whether the framework has estimated the parameters such that the input-output behavior of the system is captured. Our simulation results suggest that to estimate the 16 system parameters very accurately, precise knowledge of the inertial properties of the system is required. Nevertheless, there are other applications within vibration isolation where a multi body model which explains the data is sufficient. We expect our framework to aid the engineers at Zeiss for such applications.

PBSIDopt algorithm
static output feedback synthesis
model matching
Gray-box system identification
single stage mechanical system

http://resolver.tudelft.nl/uuid:c030e78e-62bb-4741-8e87-db3b6a8ace92

Student theses

master thesis

(c) 2016 Parthasarathy, H.