Marc Bodson - US grants

The objective of the project is to study problems related to the design of robust and multivariable adaptive control systems. Robustness is defined as the ability of an algorithm to maintain stability in the presence of disturbances and unmoldeled dynamics. We plan to apply averaging methods to study the relationship between robustness and the error equation used for adaptation. The existence and location of the tuned values will be investigated. These values determine where the controller parameters converge in the presence of unmodeled dynamics. The analysis is expected to lead to a better understanding of robustness in adaptive control, and to suggest new approaches for design. We will also be concerned with the design of multivariable adaptive control systems. We plan to study parameterizations which guarantee parameter convergence to a unique point, and require a minimal number of parameters. These properties are desirable for convergence, robustness, and implementation considerations. The prior information needed to design multivariable adaptive control systems will also be investigated, together with ways to obtain this information from input/output measurements. Experiments will be carried out to validate the results of the theory, and motivate new problems. We expect that the research will contribute to improve design methodologies for adaptive control systems, and to a wider application of these methods.

ECS-9810048 Bodson The project will provide funds for students to travel to the 37th IEEE Conference on Decision and Control, which will be held on December 16-18, 1998 in Tampa, Florida. The goal is to attract engineering students to the field of control systems, and to give the opportunity to students who have chosen control systems as a specialty to be informed about the most recent developments in control systems theory and practice.

The object of this project is to develop control algorithms for the cancellation of periodic disturbances whose frequency content varies significantly over time. Typical applications include active noise and vibration control, although the problems under consideration are generic and the algorithms are applicable in many other contexts. Two classes of algorithms are considered. They have been studied before for restricted cases and have shown promising results. The algorithms are particularly valuable because of their simplicity and because of the tools available to design them for specific performance goals. In the project, the algorithms will be developed so that they will become applicable to the most general problems with multivariable system and with disturbances having multiple harmonic components. An important intermediate goal will be the analysis of the robustness properties of adaptive algorithms using little known but powerful results of linear time-invariant system equivalence. These results will enable a precise quantification of the robustness margins of the algorithms, and may lead to new strategies for robust design. Another goal is the development of new algorithms for the estimation and tracking of the parameters of periodic signals using multiple sensors. The algorithms will be tested in active noise control or active vibration control experiments.

This is a planning grant for department level reform at the University of Utah, Department of Electrical and Computer Engineering. In this project, the already strong laboratory component within the ECE program at the University of Utah will be enhanced by developing system-level design projects to be integrated within individual courses and also spanning multiple courses. This builds on a particular strength of the department, which already has a top-notch industrially-sponsored senior design sequence a few courses with strong system design projects. Through this change, it is expected to (1) increase student recruitment and retention (including diverse students who might otherwise not have chosen or stayed in engineering), (2) increase student motivation (system design projects are FUN), (3) increase knowledge acquisition and retention (because you can't forget something that you need to integrate at the end of the semester and perhaps next semester too), and (4) develop system-level design understanding (the formal training of which is very limited within ours and other traditional engineering curricula).

The novel aspects of this proposal include the projects themselves, the methods and concepts of integrating multiple classes within the curriculum, and the methods to assess and implement these projects. Undergraduate students, graduate students, and faculty will work together to build this new program within our department.