Home > Seminars > Learning-Based Indirect Adaptive Control for Nonlinear Systems

Learning-Based Indirect Adaptive Control for Nonlinear Systems

Start:

11/10/2014 at 2:00PM

End:

11/10/2014 at 3:00PM

Location:

258 Fitzpatrick Hall

Host:

College of Engineering close button
headerbottom

Panos Antsaklis

Panos Antsaklis

VIEW FULL PROFILE Email: antsaklis.1@nd.edu
Phone: 574-631-5792
Website: http://www.nd.edu/~pantsakl/
Office: 205A Cushing Hall

Affiliations

Department of Electrical Engineering H. Clifford and Evelyn A. Brosey Chair Professor
College of Engineering H. Clifford and Evelyn A. Brosey Chair Professor
Research Interests: My research group focuses on Cyber Physical Networked Embedded Systems and addresses problems in the interdisciplinary research area of Control, Computing and Communication Networks, and on Hybrid and Discrete Event Dynamical Systems. It addresses problems of control and ...
Click for more information about Panos
574-631-5792
Add to calendar:
iCal vCal

Classical adaptive control deals with controlling partially known systems based on their uncertain model, i.e., controlling plants with parameters uncertainties. Classical adaptive methods can be classified as `direct', where the controller is updated to adapt to the process, or ‘indirect’, where the model is updated to better reflect the actual process. The results in ‘classical’ adaptive control are mainly based on the structure of the model of the system, e.g. linear vs. nonlinear, with linear uncertainties parameterization vs. nonlinear parameterizations, etc. On the other hand, Extremum seeking (ES) is a well-known approach by which one can search for the extremum of a cost function associated with a given process performance without the need for a detailed model. Another worth mentioning paradigm is the one which uses ‘learning schemes’ to estimate the uncertain part of the process. Indeed, in this paradigm the learning-based controller, based either on machine learning theory, neural network, fuzzy systems, etc. is trying either to estimate the parameters of an uncertain model, or the structure of a deterministic or a stochastic function representing part or totality of the model. 
  
We want to concentrate in this presentation on the use of ES theory in the ‘learning-based’ adaptive control paradigm. We present some recent results on learning-based adaptive control for nonlinear systems. We first study the problem of adaptive trajectory tracking for general nonlinear systems, and show that for the class of nonlinear systems with parametric uncertainties which can be rendered Input-to-State stable w.r.t. the parameter estimation error, that it is possible to merge together the Input-to-State stabilizing feedback controller and a model-free extremum seeking (ES) algorithm to realize a learning-based adaptive controller. We investigate the performance of this approach in terms of tracking error's upper-bounds. Finally, we propose a learning-based approach to auto-tune feedback gains for nonlinear stabilizing controllers for nonlinear models affine in the control. 

Seminar Speaker:

Mouhacine Benosman

Mouhacine Benosman

Mitsubishi Electric Research Laboratories

Before coming to Mitsubishi Electric Research Laboratories (MERL) in 2010 as a Senior Member Research Staff, Dr. Benosman worked at universities in Rome, Italy, Reims, France and Glasgow, Scotland before spending 5 years as a Research Scientist with the Temasek Laboratories at the National University of Singapore. His research interests include modeling and control of flexible systems, non-linear robust and fault tolerant control, learning-based adaptive control, vibration suppression in industrial machines and multi-agent control with applications to smart-grid. Dr. Benosman is a Senior IEEE member and Associate Editor of the IEEE CSS Conference Editorial Board.