Session Description:
System identification techniques build mathematical models of systems from experimental data. The models can be used to validate the analyses performed during the design of the system, to develop feedback control strategies, and to predict subsequent dynamic performance. When these techniques are performed over multiple nominally-identical systems, or in multiple operating states of the same system, they can be used to quantify expected levels of variation in dynamic response. This entry-level tutorial emphasizes mainly frequency domain techniques that yield transfer function measurements at a series of discrete frequencies. The resulting non-parametric model of the system, commonly known as a Bode plot, is essential to classical control techniques. A typical system identification experiment involves disturbing the system with some known excitation signal, recording measurements of the resulting behavior, and processing the measurements to yield the frequency response. The techniques appear to be simple, but the measurements are inevitably corrupted by noise, there are likely to be additional unmeasured disturbances influencing the system’s response, and most systems exhibit some degree of nonlinear behavior. Systems of interest to precision engineers often have multiple degrees of freedom, and operate under closed-loop control. Together these factors require an additional level of care in the planning of experiments, processing of data, and interpretation of results.Attendees to the course will practice techniques for designing an excitation signal, for choosing where and how to apply the excitation signal, and for processing the measurements into dynamic models. We will emphasize the use of periodic signal generation techniques (also known as multisines), and will extensively use experimental data and sample code to demonstrate the techniques, as well as likely problem areas. The focus of the discussion will be solidly on the practitioner, but references will be provided for those interested in developing a deeper understanding of the underlying mathematical techniques.
Dr. Stephen Ludwick leads the mechatronic research group at Aerotech, Inc., a manufacturer of precision automation systems. He joined Aerotech in 1999, and is currently responsible for developing motion control systems and feedback control algorithms with an emphasis on the interactions between mechanical, electrical, and algorithmic components of a design. He is also an adjunct associate professor in the Department of Mechanical Engineering and Materials Science at the University of Pittsburgh, and currently serves as an Editor-in-Chief for the journal Precision Engineering. Dr. Ludwick holds a B.S. degree in Mechanical Engineering & Engineering and Public Policy from Carnegie Mellon University, and S.M. and Ph.D. degrees in Mechanical Engineering from the Massachusetts Institute of Technology.