Dynamics and Controls Seminars

Abstracts 2002 -2003

* * * * * * * * * * * * *
Tuesday, June 10, 2003
10:00 a.m.
479 EBU-II

Ted Iwasaki
Mechanical and
Aerospace Engineering Department
University
of Virginia

“Toward Control Principles in Biological Locomotion”

The long term objective of our research is to use the knowledge from biological studies on animal locomotion and establish a new paradigm for control design that realizes robust, adaptive, and autonomous systems. As a very first step, we propose a new dynamical model of a single neuron, given by a specific class of the Lur'e systems, and show that the model makes a good trade-off between the simplicity and the accuracy when compared with the existing models. The Lur'e neuron model is intended for adoption as a basic unit in biologically inspired control systems.

 

The latter half of the talk will focus on investigating the potential of biological oscillators for use as a new feedback control architecture to achieve autonomous locomotion. In particular, we use the Lur'e neuron model to construct a biological oscillator and demonstrate by a simple pendulum example that the oscillator is capable of robustly exciting the natural motion of the mechanical system. Interestingly, an oscillator of the same architecture but with a simpler neuron model, similar to those used in artificial neural network literature, does not seem to have the robust self-excitation capability. Practical implications of the result will be discussed.

 

* * * * * * * * * * * * *

Tuesday, April 15, 2003
2:00 p.m.
479 EBU-II

Douglas MacMartin
Senior Research Fellow
California Institute of Technology

“Dynamics and Control of Shock Motion in a Near-Isentropic Inlet”

Inlet pressure recovery of supersonic aircraft could be improved using a near-isentropic inlet with only a weak normal shock aft of the throat, however, such an inlet is highly susceptible to unstart. Small perturbations can move the shock ahead of the throat, where it is unstable.  The dynamics of the inlet and shock are analyzed using a low order model that captures both the nonlinear shock motion and inlet acoustic propagation.  This model allows parametric exploration of both the potential and limitations of using control to actively stabilize the shock, including actuator authority as a function of location, actuator authority and bandwidth requirements, and sensor requirements.  A simple control law is shown to be sufficient to stabilize the shock motion.  If time permits, I will briefly discuss recent (preliminary) results on feedback control of separation dynamics.

 

* * * * * * * * * * * * *

Friday, March 7, 2003
10:30 a.m.
479 EBU-II

Ralph C. Smith
Department of
Mathematics
North Carolina State University


"Model Development and Control Design for High Performance Nonlinear Smart Material Systems"

High performance transducers utilizing piezoceramic, electrostrictive, magnetostrictive or shape memory elements offer novel control capabilities in applications ranging from flow control to precision placement for nanoconstruction.  To achieve the full potential of these materials, however, models and control designs which accommodate the constitutive nonlinearities and hysteresis inherent to the compounds must be employed.  Furthermore, it is advantageous to consider material characterization, model development, and control design in concert to fully exploit the novel sensor and actuator capabilities of these materials in coupled systems.

 

In this presentation, the speaker will discuss recent advances in the development of model-based control strategies for high performance smart material systems.  The presentation will focus primarily on the development of unified nonlinear hysteresis models, inverse compensators, and nonlinear control strategies for high precision or high drive regimes.  The range for which linear models and control methods are applicable will also be outlined.  Examples will be drawn from problems arising in structural acoustics, high-speed milling, deformable mirror design, artificial muscle development, tendon design to minimize earthquake damage, and atomic force microscopy.

 

* * * * * * * * * * * * *

Friday, February 14, 2003
11:00 a.m.
479 EBU-II

Martin Berggren
Sandia National Labs
Albuquerque, New Mexico

"Shape optimization of an acoustic horn"


We apply numerical shape optimization to alter the transmission properties of an acoustic horn. A finite-element approximation of the Helmholtz equation models the wave propagation, and a quasi-Newton algorithm solves the optimization problem. The adjoint-equation approach provide gradients, and a smoothing technique enforces smooth design updates.  Without smoothing, the optimization algorithm terminates at an unphysical, local minimum of the discrete objective function. The use of smoothing amounts to a change of variables and not to a regularization of the optimization problem. Single-frequency optimization yields convex horns narrowly optimized only for the design frequency. Multiple-frequency optimization yields horns with a slightly wavy shape and with good transmission properties throughout the optimized frequency band. A convexity constraint in the optimization inhibits waviness and produces a horn with smooth corners. The constraint causes only marginal performance degradation at the design frequencies but yields improved performance in a region above the design frequencies.


* * * * * * * * * * * * *

Wednesday, February 12, 2003
2:00 p.m.
479 EBU-II

Anna Stefanopoulou
Department of
Mechanical Engineering
University
of Michigan

"Control of Fuel Cell Systems"

Recent improvements in membrane material, fuel processing, and hydrogen storage enabled fuel cell (FC) power to move from the laboratory to experimental vehicles and commercial power units. Although steady-state FC behavior is considered the normal operating mode; start-up, shutdown, and sudden load changes are characteristic and ubiquitous to all power producing devices. During these critical periods, the viability, efficiency, and robustness of the FC systems depend on monitoring and controlling their  unique transient behavior.

 

In this talk, the critical control problems in Proton Exchange Membrane (PEM) FC operation will be introduced and a nonlinear dynamic model that was developed for control, estimation, and diagnostics will be presented. Also presented will be a few results on the airflow control design of a high-pressure direct hydrogen FC for an automotive application. A highlight of this work is the delineation of the tradeoff between fast oxygen (air) reactant supply that ensures long FC life and transient fuel cell net power response during rapid current (load) demands. A control architecture study for FC hydrogen starvation in the case of reformed natural gas will also be presented.

 

 

* * * * * * * * * * * * *

Thursday, January 23, 2003
1:00 p.m.
479 EBU-II

Carlos Canudas-de-Wit
Laboratoire d'Automatique de Grenoble
St. Martin d'Hčres, France

" Friction Tire/Road Modeling, Estimation and Optimal Braking Control "

Dr. Canudas-de-Wit will present a series of results concerning the problem of modeling, estimation of contact road/tire friction. In addition, some aspects related to the problem of optimal braking control will be presented.  The new dynamic friction force model for the longitudinal road/tire interaction for wheeled ground vehicles will also be discussed. The model is based on a dynamic friction model developed previously for contact-point friction problems.  By assuming a contact patch between the tire and the ground the model is indeed given by a PDE. It is, however, possible to approximate the PDE by an ODE  (the lumped model) for the friction force is developed based on the patch boundary conditions and the normal force distribution along the contact patch. Dr. Canudas-de-Wit will also the possible extension of longitudinal models to high dimensional models including lateral force and self-alignment torque. In addition, he will discuss how these models may be affected by other external factors such as: road conditions (roughness), road/tire interfaces (wet road), temperature variation, etc.  In the estimation part, Dr. Canudas-de-Wit, we discuss the problem of tire-road friction estimation using only angular wheel velocity which cannot always been computed from actual sensors. Tire forces information is relevant to problems like: optimization of Anti-look brake systems (ABS), traction system, diagnostic of the road friction conditions, etc.

 

These results may suggest alternative traction control methodologies, other than the current ones based on the use of tracking of the "optimal'' slip coefficient using, for example, sliding mode control. These aspects are presented in the last two section of Dr. Canudas-de-Wit's paper. First, he investigates optimal braking strategies for a simplified model of a wheeled vehicle. We derive the {\em exact} optimality conditions for the minimum braking distance problem by solving the associated optimal control problem. It is shown that the optimal control is singular and can be written in a state feedback form. Then, he will present a control scheme for emergency braking of vehicles.  The controller utilizes estimated state feedback control to achieve a near maximum.

 

* * * * * * * * * * * * *

Friday, January 10, 2003
11:00 a.m.
479 EBU-II

Dušan M. Stipanovic
Hybrid Systems Laboratory
Department of Aeronautics and Astronautics
Stanford University

"Overlapping Decentralized Approach in Control, Optimization and Computation of Reachable Sets"

A variety of natural and man-made applications, such as groups of vehicles, power system networks, networks of sensors and actuators (for example, a large segmented telescope) and fluid mechanical systems, can be treated as interconnected systems consisting of subsystems that are not completely decoupled, that is, the subsystems share common variables. This modeling process is based on the method of overlapping decomposition. In this talk, we present overlapping decentralized methods for control, optimization and computation of both backward and forward reachable sets for interconnected dynamic systems. The efficiency of the overlapping decentralized approach is demonstrated on the following examples: control of a formation of Unmanned Aerial Vehicles (UAVs), multiple vehicle coordination, and computation of the unsafe set for the two aircraft collision avoidance problem.

 

* * * * * * * * * * * * *

Thursday, December 19, 2002
11:00 a.m.
479 EBU-II

Rolf Isermann
Institute of
Automatic Control
Darmstadt University
of Technology

"Fault-Tolerant Drive-By-Wire Systems"

 

The development of automotive systems shows an increasing integration of electronic sensors, microcomputers and actuators for single components, engine, drive-chain, suspensions and brakes. After considering electronic driver assisting systems such as ABS, TCS, ASR, ESP, BA the developments towards drive-by-wire systems with and without mechanical or hydraulic backup are considered. Drive-by-wire systems consist of an operating unit (steering wheel, braking pedal) with an electrical output, a haptic feedback to the driver, bus systems, microcomputers, power electronics, and electronical actuators. For their design safety integrity methods like reliability, fault tree, hazard analysis and risk classification are required. Different fault tolerance principles with various forms of redundancy are considered resulting in fail-operational, fail-silent and fail-safe systems. Fault-detection methods are discussed for use in low-cost components. This is followed by some principles for fault-tolerant design of sensors and actuators.

 

A brake-by-wire system with electronic pedal and electrical brakes is then considered in more detail showing the design of the components and the overall architecture. An outlook then shows the further development of drive-by-wire-systems.

 


* * * * * * * * * * * * *

Monday, December 16, 2002
12:30 p.m.
479 EBU-II

Dr. Andrea Lecchini
CESAME - Centre for System Engineering and Applied Mechanics
Universite' Catholique de Louvain

"Convergence Analysis and Optimal Prefiltering in Iterative Feedback Tuning"

Iterative Feedback Tuning (IFT) is a widely used procedure for controller tuning. It consists in a sequence of iteratively performed special experiments on the plant interlaced with periods of data collection under normal operating conditions.   In this seminar the asymptotic convergence rate of IFT for disturbance rejection, which is one of the main fields of application, will be derived. Further it will be shown how to improve the convergence of IFT by prefiltering the input data for the special experiment. At each iteration step the optimal prefilter is computed from data collected under normal operating conditions of the plant.

 

* * * * * * * * * * * * *

Friday, December 6, 2002
11:00 a.m.
584 EBU-II

Bjorn Wittenmark
Department of Automatic Control
Lund Institute of
Technology, Sweden

"Extremal control of Wiener model processes"

Extremal control of Wiener type processes is considered. These models consists of a linear part followed by a static nonlinearity. We will consider nonlinearities having one extremum point. The purpose is to keep the output of the process as close as possible to the extremum point.  The main problem in the control of this kind of  processes is the non-uniqueness of the inverse of the nonlinearity. This causes problems, e.g., in the estimation of the states of the process and the identification in the adaptive case. A one-step-ahead controller combined with a probabilistic estimator is proposed and analyzed.

 

* * * * * * * * * * * * *

Wednesday, November 20, 2002
2:00 p.m.
479 EBU-II

Luca Brandt
Department of Mechanics
Royal Institute of Technology (KTH)
Stockholm, Sweden

"Breakdown of Streamwise Streaks in Blasius Boundary Layers"

In boundary layers with free-stream turbulence intensities of 1% or more, it is observed experimentally that transition occurs rapidly, bypassing the natural scenario triggered by unstable Tollmien-Schlichting waves. In the former scenario, denoted bypass transition, the turbulence is highly damped inside the boundary layer, but low frequency oscillations, associated with long streaky structures in the streamwise velocity component appear. The initial amplification of these structures can be explained by non-modal growth theory.  From the experimental observations, it was conjectured that, as the streaks reach a threshold amplitude, they undergo a secondary instability before the breakdown to turbulence occurs in the form of localized turbulent spots. In the present investigation, the instability and breakdown of a model streak, i.e. steady and  spanwise periodic, has first been studied.  The results obtained form the basis to understand the extensive direct numerical simulations of a transitional Blasius boundary layer under free-stream turbulence, which have been performed. It is shown that the streaks undergo a secondary instability either of sinuous (antisymmetric) or varicose (symmetric) type. Further, it is noticed how the interaction between streaks may play an important role in triggering the breakdown to turbulence.  Finally, recent experimental and numerical results obtained at the Department of Mechanics about possible control strategies for the transition scenario under consideration are presented.


* * * * * * * * * * * * *
Friday, November 8, 2002
12:00 noon
479 EBU-II

Mustafa Khammash
Department of Mechanical and Environmental
Engineering
University
of California, Santa Barbara

"Feedback Regulation in Biology: A Control Theory Perspective"

 

In the technological and engineering, sciences feedback control methods have been studied and applied extensively to man-made systems. In the biological sciences, the mathematical approach to studying feedback mechanisms has not been as common. Considering the ubiquity of feedback mechanisms in biology there is much to be gained by mathematical modeling and analysis of these systems.  In this talk we look at examples of dynamic systems in biology from a control theory perspective. We show that ideas from control theory and systems engineering may be used effectively to shed new light on known regulatory mechanisms, both at the molecular level and at the system level.

 

* * * * * * * * * * * * *

Monday, November 4, 2002
2:00 p.m.
479 EBU-II

 

Dan Henningson
Department of Mechanics, KTH
and
Swedish Defense Research Agency (FOI)


"Optimal Control and Estimation Applied to Three-Dimensional Boundary Layers"


An overview of the activities in flow control at Mechanics, KTH and FOI is given with emphasis on application to three-dimensional boundary layers.  First work is presented that extends previous research on linear controllers and estimators in temporal channel flow to spatially evolving boundary layer flow.  Falkner--Skan--Cooke velocity profiles are used as the base flow in the Orr--Sommerfeld--Squire equations to compute the optimal feedback control through blowing and suction at the wall utilizing linear optimal control theory. Estimators using only wall measurements are computed in a similar manner and used in compensators applied to the same flow. The controllers and compensators are used in direct numerical simulations of the Navier-Stokes equations and it is shown that random cross-flow disturbances and wavepackets can be stabilized. Second work is presented that describes the use of optimal control theory to extend the laminar region on aircraft wings by steady suction. An iteration procedure based on the boundary layer equations; parabolic stability equations and their adjoints are used to find the optimal suction distribution. Results for Fokker 100 and Airbus 310 wings are shown.

 

* * * * * * * * * * * * *

Monday, October 28, 2002
10:00 a.m.
479 EBU-II

 

Dr. Gaemus (Shamus) Collins
Department of
Mathematics
University
of California, San Diego

"Numerical Range (and stability analysis) of the Orr-Sommerfeld Equation"

 

Stability analysis of viscous channel flow often involves a consideration of the spectra of the Orr-Sommerfeld and Squire equations.  These spectra are contained in the closure of the numerical range of the Orr-Sommerfeld and Squire operators.  Bounds for the numerical range of both operators are computed, and the Squire operator is shown to generate stable evolution.  These numerical range bounds are then used to prove that the Orr-Sommerfeld and Squire operators generate C_0-semigroups.

 

* * * * * * * * * * * * *

For information: Sophia Bligh at (858) 822-1269