ACC 2011 will provide the following 1-day pre-conference workshops (workshop fee: $235)
Tuesday, June 28
Noon–5:00 pm
Room: Golden Gate 6
W. Messner (Carnegie Mellon University)
T. Astumi (Hitachi Ltd.)
Loop shaping remains a popular method for controller design for single-input/single-output systems (SISO) because it is powerful and intuitive. In spite of its long history, several developments over the last ten years or so have led to new tools for making loop shaping even more useful for designing high performance SISO controllers. These recent advances include new compensator structures (e.g. the complex lead compensator) and new visualization tools (e.g. the Robust Bode plot).
The new compensator structures use complex zeros and complex poles for decoupling the phase contribution of a compensator from its shape at a particular frequency. Among other advantages, these new compensator structures more readily provide the capability of designing phase stabilized systems with high gain (and therefore large disturbance rejection) even when the phase loss is greater than 180o.
The new visualization tools are enhancements of the Bode plot that depict closed-loop performance and robustness specifications on the open-loop Bode plot. These tools provide the capability for designing low order controllers for robust performance from frequency response data alone. That is, realizable transfer functions are needed neither for the plant model nor for the performance weighting function nor for the uncertainty weighting function. The actual disturbance spectrum rather than a transfer function approximation can be used to shape the loop. The collection of known actual plant frequency responses, rather an approximating transfer function and bound, can be used to design for robustness. Both of these features provide for higher performance, lower order, less conservative designs.
This afternoon-long workshop will teach the features of these tools and techniques to use them effectively. The structure of this workshop will be several 45 minute modules consisting of instruction and then examples for the participants to work on their laptops. Participants will be provided with Matlab® commands for the exercises which they can take with them. This workshop will be particularly valuable to the portion of the industrial controls community that uses loop shaping. Educators will find this valuable because the material on compensator structures is suitable for advanced undergraduates.
|
Compensator Structures |
Techniques and Visualization Tools |
Tuesday, June 28, 2011
8:30 am–5:00 pm
Room: Golden Gate 4
R. Russell Rhinehart
School of Chemical Engineering - 423EN
Oklahoma State University
Stillwater, OK 74078-5021
Office: (405)744-5280
Fax: (405)744-6338
Email: rrr@okstate.edu
Web: www.che.okstate.edu
Empirical models (models based on data) are often central for model-based control and for process forecasting. For many of these applications nonlinear models are preferred in order to capture the process behavior. But, even for seemingly linear models, an adjustable delay introduces a nonlinear model coefficient. Additionally, first-principles models are often used for operator training simulators, process analysis and diagnosis, mechanism validation, and supervisory optimization; but, since processes are neither ideal nor stationary, model coefficients need to be adjusted to make the models fit the process data.
The procedure of fitting models to data is regression. In nonlinear regression the adjustable model coefficients do not appear linearly within the model.
This workshop is intended to be a practical guide for nonlinear regression modeling. Although theoretical analysis behind techniques will be revealed, the takeaway will be techniques for defining the regression objective, optimization approach, design of experiments for data generation, data-based model validation/discrimination, and criteria for selecting an appropriate model order. Excel/VBA exercises and code will complement the set of workshop notes (a monograph of approximately 120-pages) and be used as in-workshop examples. Participants should being a laptop with Excel version 2007 or higher.
Topics will include equation structures, optimization of parameter values in the presence of constraints and local traps, choosing optimization stopping criteria based on model properties, data pre-processing and post-processing, data-based model validation, discrimination between models, design of experiments that support validation outcomes, propagation of uncertainty, and model utility evaluation.
This is not the standard linear regression approach to develop response surface model structures, which leads to classic DOE models such as Latin Squares and Box and Star experimental plans. This is intended to focus on nonlinear regression.
The intended audience is for those with an engineering degree needing to understand the process of developing nonlinear models from data.
Tuesday, June 28
9:00 am–5:00 pm
Room: Golden Gate 2
Eilyan Bitar, University of California, Berkeley
Anjan Bose, Washington State University, Pullman
Duncan Callaway, University of California, Berkeley
Pramod Khargonekar, University of Florida, Gainesville
Kameshwar Poolla, University of California, Berkeley
Pravin Varaiya, University of California, Berkeley
Target Audience
Graduate students and researchers interested in understanding the role of control, modeling, and optimization in meeting the challenges of the Smart Grid.
The Smart Grid
The Smart Grid is a vision of the future electric energy system. There are many visions of this future, but perhaps these are best summarized by US Energy Secretary Steven Chu who writes that ``the Smart Grid is the key enabler for: integration of renewable energy sources into the grid, management and deployment of energy storage, load management, system transparency, and cyber and physical security of the electric energy system''. Several drivers motivate this transformation of the electric grid: (1) The physical infrastructure of the electric grid is aging and over-burdened. Electricity demand continues to rise, while investments in transmission and distribution infrastructure remain insufficient. (2) Concerns over climate change and carbon emissions have led to aggressive goals for integration of large amounts of renewable generation – especially wind, and solar - to meet our electric energy needs. The confluence of these factors has led to increased stresses on the electric grid. The key idea is that the use of advanced communication, information, and control technologies in conjunction with advances in renewable generation, energy storage, materials, sensors, and power-electronics will allow us to reach the abovementioned vision of the Smart Grid.
Opportunities
Tools from systems and control theory have tremendous potential for the proper evaluation, deployment, and operation needed to enable the Smart Grid. Some of the challenging problems that researchers in the systems and control community are best equipped to address include, but are not limited to: optimal methods for integration of variable and uncertain renewable generation, optimal deployment and operation of distributed energy storage devices, scheduling of flexible demand such as PHEVs, enhanced power system stability using wide-area monitoring, intelligent distributed protection, cyber-security of the electricity grid.