Department of Mathematics, WUSTL - Talks List, Summer 2009

A list of lectures, seminars, colloquia, and other events hosted by
the Department of Mathematics at Washington University in St. Louis

JUNE 2009
Tuesday, June 9	Colloquium
Time: Tea: 4:00-4:30pm Talk: 4:30-5:30pm Location: Cupples I, Room 199 Host: Prof. Victor Wickerhauser	Speaker: Professor Peter F. Odgaard KK Electronic A/S, Denmark Title: Mathematically Model Based Control of Modern Wind Turbines - from an Industrial Point of View Abstract: Newly installed wind turbines of today are normally in the multi-megawatt range, with a blade diameter in the range from 80m to 110m. Consequently optimization of these turbines is of high interest, for a number of reasons and with different objectives. Some examples are: optimization of power production, minimization of cost price per energy unit, lifetime expansion, power quality, and reliability optimization. A model-based control strategy is an important path to take in achievement of these different objectives. This talk will start with a short presentation of modern wind turbines and their control systems. The attention will subsequently be turned on to specific problems recently suggested solved by model based schemes. An adaptive scheme is suggested to adjust the power control to the wind turbine operational point to the actual aerodynamics of the wind turbine blades. The blade aerodynamics changes slowly with time mainly due to debris build-up on the blades. An unknown input observer is designed based on a model of the wind turbine to estimate the variation in actually aerodynamics compared with the assumed one. Based on this estimate, the aerodynamic model can be updated with an interval of a month. The optimal point of operation of the wind turbine can subsequently be found based on this updated model. Fault detection and fault accommodation is at present handled in a simple and conservative way. Faults are directly detected based on the measurements, and are most often handled by a shutdown of the turbine and an unplanned service. This is costly both in lost production and the service itself, increasing the cost price per energy unit and lowering the reliability of the wind turbine. If model based methods are used, faults can be detected either faster or with less false positive detections. Secondly models can be used to design analytical redundancy such that the turbine can continue production at some level in a safe way. Some recently published results in detection of speed sensor faults and generator torque measurements are presented. In order to encourage the work on fault detection and accommodation on wind turbines, a benchmark model of this problem has been developed and will be introduced.
AUGUST 2009
Tuesday, August 4	Thesis Defense
Time: 10:00-12:00pm Location: Cupples I, Room 199 Host: Profs. Guido Weiss and Edward Wilson	Speaker: Bob Houska Department of Mathematics, Washington University in St. Louis Title: Frames, Composite Wavelets, and Shearlets Abstract: One-dimensional wavelet systems have enjoyed a great deal of success in applications. This success is due, in large part, to the plentiful existence of compactly supported and smooth one-dimensional scaling functions. Traditional wavelet systems have been used quite successfully in higher-dimensional applications as well. However, the geometric structure present in dimensions two and higher is significantly more complex than that present in dimension one, and there are several important multi-dimensional applications for which traditional wavelet systems are too geometrically simplistic. In response to this deficiency, the more geometrically diverse composite wavelet systems were recently introduced by Guo, Labate, Weiss, and Wilson. A particular type of composite wavelet system - the shearlet system - has been shown by the above mentioned authors to outperform traditional wavelet systems in several important multi-dimensional applications. Despite these positive results, however, shearlet systems have one major drawback - essentially no useful shearlet scaling functions exist. We will discuss several results of this variety.
SPRING 2010 Seminars Schedule
Thursday, April 22, 2010	Loeb Lecture
Time: Tea: 4:00-4:30pm Talk: 4:30-5:30pm Location: TBA Host: Prof. Ronald Freiwald	Speaker: Professor Martin Golubitsky Department of Mathematics, Ohio State University Title: TBA

Would you like to receive the current talk list?
Contact Marie Taris by e-mail:
marie(at)math(dot)wustl(dot)edu
to be added to our mailing list.


Last Updated 07/08/09

WUSTL Main Page
Math Department Main Page