This two-day workshop will provide attendees with a detailed and in-depth overview of methods and technology pertinent to active control of vibrations and acoustics. The course will also encompass applications involving structural-acoustic interaction. An important feature of the course is its balance between theoretical issues and technological implementation. The first day of the workshop is devoted to the physics of vibrations and acoustics, design aspects of practical sensors and actuators for active vibration suppression applications, and both fixed-gain and adaptive control methods.
The second day of the workshop focuses on relevant technology including hardware implementation issues, system identification techniques, and programming of DSP's. The workshop closes with hardware demonstrations of noise and vibration suppression techniques on two separate testbeds, one involving structural vibration and the other devoted to acoustic noise cancellation. This workshop is designed to be of value to attendees with diverse backgrounds. Researchers seeking to construct a controls laboratory involving noise and vibration control experiments will be exposed to fundamental hardware and equipment issues. In addition, industrial practitioners will obtain an overview of the latest advances in vibration suppression technology relating to design and implementation techniques.
SCHEDULE: JUNE 22 - Theory and Methods
08:30-10:30 Vibrations, acoustics, and structural-acoustic interaction
for control engineers
11:00-12:00 Control engineering design aspects of sensors and actuators
13:00-15:00 Fixed-gain control methods for vibrations and acoustics
15:30-17:30 Adaptive control methods for vibrations and acoustics
SCHEDULE: JUNE 23 - Implementation Issues
08:30-09:30 Hardware for vibrations and acoustics
09:30-10:30 Sampling and discretization effects in data acquisition and control
11:00-12:00 Time-domain identification for vibrations and acoustics
13:00-14:00 Frequency-domain identification for vibrations and acoustics
14:00-15:30 Programming essentials for DSP hardware
16:00-17:30 Hardware demonstrations of active vibration and acoustic control
SCHEDULE: JUNE 22
08:30-10:00 Fundamentals of MATLAB and Simulink
10:30-12:00 Modeling in MATLAB and sequential languages
13:00-14:30 Development of Hierarchical Simulink Models
15:00-17:00 Selected Examples
SCHEDULE: JUNE 23
08:30-10:00 Time Domain and Frequency Domain Simulations
10:30-12:00 Analysis of Dynamic Systems
13:00-14:30 Analysis and design of Control System
15:00-17:00 Final Project: Magnetic Suspension Seismometer
This workshop discusses fundamental limitations in achieving performance in control and filtering systems. An important part of any design process is characterization of the best achievable performance from the system in terms of its intrinsic dynamics and structure. This analysis reveals what is, and is not, possible prior to application of any specific design technique. This knowledge allows rational trade offs of desirable, but often incompatible features in design specifications; or, when feasible, to alter the system configuration to alleviate the most critical constraints, e.g. by changing actuator/sensor arrangements.
A well-known framework for analyzing design limitations uses frequency domain sensitivity functions which describe system performance and robustness. Limitations are then expressed as integral formulas (Bode and Poisson integrals) that arise from the analyticity of the sensitivity functions. An alternative, state-space approach is to study the best achievable performance of optimal control when the cost of control tends to zero (cheap control). This line of research, which has insightful connections with Bode integrals, has been recently extended to the analysis of performance of nonlinear systems. This workshop provides a comprehensive overview using both approaches. It includes classical results as well as a selection of the most recent contemporary extensions to multivariable systems, sampled-data, filtering, and nonlinear problems. Use of these techniques in engineering practice is illustrated with several industrial applications, including control of a single strand rolling mill.
SCHEDULE: JUNE 22
08:30-09:15 Motivation and history.
09:15-10:00 Time-domain analysis of tradeoffs. Application: Rolling mill.
10:30-11:15 Frequency domain limitations in SISO control.
11:15-12:00 Limitations in filtering: frequency domain and optimal approach.
13:00-15:00 Limitations in MIMO control. Automotive applications.
15:30-16:15 Limitations in sampled-data control.
16:15-17:00 Limitations in achievable L_2 peformance in linear and nonlinear control.
This workshop illustrates recently developed Set Membership (SM) methods for identifying models in a suitable form for robust control design. Robust control methods aim to design controllers which guarantee performance not for a single nominal model, but for a set of models (obtained by perturbations of the nominal model). Such perturbations recognize that models derived by any identification method are always affected by uncertainty. If the design methods are to be used in real world problems, the uncertainty model set must be provided by suitable SM identification methods, operating on measurements of the actual plant and on available prior information about the plant and the noise affecting its measurements. For example, H_infinity identification estimates a model and an identification error (the H_infinity norm of the error transfer function) which produces uncertainty model sets in a form well suited for the H_infinity control methodologies.
This workshop emphasizes two main aspects of the identification-control problem: (1) Obtaining "tight" uncertainty models, essential for obtaining "high" guaranteed control performances, and (2) Evaluating the effect of low complexity (order) models and controllers on guaranteed control performances.
The overall identification-control procedure presented in this workshop consists of,
SCHEDULE: JUNE 23
08:30-09:00 Preliminaries and motivations for a SM approach.
09:00-10:00 General SM identification theory.
10:15-11:00 SM identification from time open loop data.
11:00-12:00 SM identification from frequency open loop data.
13:00-13:30 Validation of prior assumptions.
13:30-14:15 Identification of "soft" uncertainty models
14:30-15:30 SM identification from closed loop data.
15:30-16:30 Robust control design and robust performances evaluation.
16:30-17:30 Applications (flexible structure, CD player, wafer stepper)
This workshop focuses on four SDRE methods: SDRE nonlinear regulation, SDRE "nonlinear H_2" control, SDRE "nonlinear H_infinity" control, and SDRE nonlinear filtering.
All of the theory developed to date on SDRE regulation is presented starting with conditions that guarantee local asymptotic stability. Optimality and suboptimality conditions and properties are covered, the nonuniqueness of the SDC parameterization is described, and it is shown how the SDC parameterization itself can be parameterized. Design approaches are reviewed, and the capabilities of SDRE design methodology are highlighted, including the ability to impose hard bounds on control, control rate, or even control acceleration and the capability to directly handle unstable non-minimum phase systems. Numerous examples are used to illustrate the methodology. Conditions and special cases that yield global and semi-global asymptotic stability of the closed-loop system will be covered. Controllability issues are then presented including differences and similarities between SDC factored pointwise controllability and true nonlinear controllability.
The remaining SDRE methods are treated similarly. For the full state information case, it is shown that the SDRE "nonlinear H_infinity" controller produces a local solution of the suboptimal nonlinear H_infinity control problem. The effectiveness of SDRE "nonlinear H_2 control" is illustrated through a full envelope missile pitch autopilot design. Finally, the filtering aspect of the SDRE methodology is illustrated using several examples; it is shown how the nonuniqueness of the SDC parameterization can be used to avoid singularities or avoid loss of observability.
SCHEDULE: JUNE 23
09:00-10:15 SDRE Nonlinear Regulation: Stability, Optimality, Suboptimality
10:30-12:00 SDRE Nonlinear Regulation: Examples
13:00-14:45 SDRE Nonlinear Regulation: Sampled Data, Controllability
15:00-15:45 SDRE "Nonlinear H_infinity" Control
15:45-16:30 SDRE "Nonlinear H_2 Control
16:30-17:00 SDRE Nonlinear Filtering
The design of modern controlled engineering systems requires integration of mechanical and structural components, sensors and actuators, and finite precision controllers to produce total systems with desired performance characteristics. However, universities and textbooks traditionally treat design, structural dynamics, signal processing, and controls as separate, independent disciplines. Thus we know how to predict closed loop performance, given COMPONENT properties, but we do not know how to determine COMPONENT requirements from SYSTEM requirements. To find true limits of performance, and to design optimal total systems, an integrated approach to system design is required, where all resources (materials, structures, signal processing and controls) are optimized (or at least "suboptimized") together rather than individually.
This workshop presents a scientific approach for integrating modelling, structural design, control synthesis and signal processing aspects in the design of controlled engineering systems. Issues include:
SCHEDULE: JUNE 23
08:30-09:00 Introduction and basic concepts - need for a theory for
09:00-10:00 Modeling, identification and design for control
10:15-11:00 Unified formulation of robust control problems
11:00-12:00 Design parametrization for achievable performance
13:00-13:30 Optimal mix of passive and active control (OMPAC software)
13:30-14:15 Design for controller implementation, LMI approaches
14:30-15:30 Finite signal-to-noise ratio models and control
15:30-16:30 Computational tools and iterative redesign methods
16:30-17:30 Design cases and examples - Discussion
This half-day workshop highlights technology transfer and application issues, discusses current control research trends from an industry perspective, and suggests promising avenues for industry-relevant controls research. The material is directed towards researchers interested in transferring technology from a research environment to industrial practice. Several of the non-technological factors involved in successful commercialization and broad-based application are discussed, including: marketing of new technology, support for maintenance and upgrades, and the necessary skill levels of users. Technology itself is just one piece of the puzzle, albeit a critical one. The importance of other technologies in control system solutions, including sensors, human interfaces, computing platforms, and communications are emphasized. These peripheral technologies are enablers for facilitating development of advanced control algorithms, but at the same time algorithmic research must take into account limitations of the existing infrastructure. The importance of domain knowledge will also be emphasized--general-purpose control science may be appropriate for academic coursework, but its application to real problems inevitably requires customization.
Based on insights gained from these big-picture considerations, the next topic reviews several major areas in control, focusing on those research and development issues that stand in the way of practical impact. These areas include: nonlinear control, adaptive control, system identification, and intelligent control; the discussion refers to domain-specific considerations in process control, environmental control, and other fields. Some particularly promising opportunities for control technology will also be discussed; there is considerable excitement within industry in areas such as hybrid models, large-scale optimization, and intelligent data analysis. Technical approaches that can furnish acceptable, fieldable solutions are needed.
Other topics, with suggestions and recommendations given as appropriate include: the structuring of mutually beneficial industry-university collaboration; the importance of, and avenues for, intellectual property protection; and consequences for the research community of the re-engineered corporation.
SCHEDULE: JUNE 27 (ONE-HALF DAY)
08:30-09:00 Motivation and overview: The role of technology in commercial
products and services
09:00-09:30 The system consideration in control system developments
09:30-10:00 Industry perspectives on topics in control: nonlinear control, adaptive control, system identification, intelligent control
10:15-11:00 New opportunities for control technology: hybrid modeling, large-scale optimization, data mining
11:00-11:30 Some other topics: industry-university cooperation, intellectual property protection, impact of re-engineering
11:30-12:00 Summary and discussion
This workshop presents a practical guide for control or structural engineers working on flexible vehicles and structures. The workshop begins with an introduction to dynamic modeling: differential equations, modal models, state space matrices, and distributed matrices. Computer programs for such models will be covered, with emphasis on Finite Element Modeling using NASTRAN or equivalent programs. Model reduction techniques include: residualization, truncation, reduced modal models, and state space reduction methods. The models will also be transformed for use in control system design. The control methods include: classical frequency domain methods and root locus, eigenvalue and eigenvector placement, optimal/robust control, and fuzzy control.
Following modeling and control system methodology, the workshop presents a practical design/validation process. It starts with stick model descriptions of the system that results in differential equations or other equivalent descriptions. Then finite element models are constructed which are linearized to generate a modal description. (Another possible approach is to generate a model using the assumed modes for the system.) This plant description is then reduced for use in the control system design process. A simple, linear representation of the control system is then implemented in the full structural dynamic model, for example in NASTRAN or the equivalent, for final verification of the design.
Example aeroservoelastic and large space structure applications are explored in the final part of the workshop. The aeroservoelastic example will be of an advanced supersonic transport (AST) with a long, slender fuselage resulting in relatively low frequency first and second fuselage bending modes. The large spacecraft example will be of an large radar satellite , which shows significant structural dynamics due to its large diameter antenna. Also, an example of a large antenna structure and of the flexible link of a robot will be shown.
NOTE: The textbook "Balanced Control of Flexible Structures" by Wodek Gawronski will be available for purchase at the Springer-Verlag booth in the exhibits area and is recommended to workshop participants.
SCHEDULE: JUNE 27
08:00-08:30 Welcome and Introduction
08:30-10:00 Dynamic Modeling
10:15-12:00 Control Design Methods
13:00-14:30 Design/Validation Process
14:45-16:00 Example Applications
16:00-16:30 Conclusion and Wrap-up