BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Isaac Newton Institute Seminar Series
SUMMARY:Data-Enabled Predictive Control of Autonomous Ener
gy Systems - Florian Doerfler (ETH Zürich)
DTSTART;TZID=Europe/London:20190503T113000
DTEND;TZID=Europe/London:20190503T123000
UID:TALK123631AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/123631
DESCRIPTION:We consider the problem of optimal and constrained
control for unknown systems. A novel data-enabled
predictive control (DeePC) algorithm is presented
that computes optimal and safe control policies u
sing real-time feedback driving the unknown system
along a desired trajectory while satisfying syste
m constraints. Using a finite number of data sampl
es from the unknown system\, our proposed algorith
m uses a behavioral systems theory approach to lea
rn a non-parametric system model used to predict f
uture trajectories. We show that\, in the case of
deterministic linear time-invariant systems\, the
DeePC algorithm is equivalent to the widely adopte
d Model Predictive Control (MPC)\, but it generall
y outperforms subsequent system identification and
model-based control. To cope with nonlinear and s
tochastic systems\, we propose salient regularizat
ions to the DeePC algorithm. Using techniques from
distributionally robust stochastic optimization\,
we prove that these regularization indeed robusti
fy DeePC against corrupted data. We illustrate our
results with nonlinear and noisy simulation case
studies from aerial robotics\, power electronics\,
and power systems.
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR