EE ZOOM Seminars: Idan Amir & Zeev Kustanovich
Roi Livni is inviting you to a scheduled Zoom meeting.
Topic: Seminar
Time: Apr 5, 2020 03:00 PM Jerusalem
Join Zoom Meeting
https://zoom.us/j/961506585
Meeting ID: 961 506 585
One tap mobile
+13462487799,,961506585# US (Houston)
+16699006833,,961506585# US (San Jose)
Dial by your location
+1 346 248 7799 US (Houston)
+1 669 900 6833 US (San Jose)
+1 929 205 6099 US (New York)
+1 253 215 8782 US
+1 301 715 8592 US
+1 312 626 6799 US (Chicago)
Meeting ID: 961 506 585
Find your local number: https://zoom.us/u/aCcdfZ1DW
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Speaker: Idan Amir
M.Sc. student under the supervision of Dr. Roi Livni
Sunday, April 5th, 2020 at 15:00
ZOOM
Prediction with Corrupted Expert Advice
Abstract:
We revisit the fundamental problem of prediction with expert advice, in a setting where the environment is benign and generates losses stochastically, but the feedback observed by the learner is subject to a moderate adversarial corruption.
We prove that a variant of the classical Multiplicative Weights algorithm with decreasing step sizes achieves constant regret in this setting and performs optimally in a wide range of environments, regardless of the magnitude of the injected corruption.
Our results reveal a surprising disparity between the often comparable Follow the Regularized Leader (FTRL) and Online Mirror Descent (OMD) frameworks: we show that for experts in the corrupted stochastic regime, the regret performance of OMD is in fact strictly inferior to that of FTRL.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Speaker: Zeev Kustanovich
Ph.D. student under the supervision of Prof. George Weiss
Sunday, April 5th, 2020 at 15:30
ZOOM
Stability analysis of microgrid power systems
Abstract
Microgrids are key components of modern electric grids that facilitate the integration of renewable or distributed generation units. Power system stability analysis requires the development of advanced (usually model-based) techniques posing a big challenge to the control community, because the power system is complex and highly nonlinear, with an immense variety of actuators, controls, protections, and operational constraints that change in time and are subject to various faults. Although there are widely accepted (reduced) models for synchronous generators (SG), they are typically simplifications of the reality that may lead to non-accurate descriptions of the behavior of microgrids. The situation is getting worse due to the penetration of inverter based power sources (mostly from renewables) into the grid, because the lack of inertia of the non-SG power sources threatens the stability of the power grid. Most distributed generators are connected to the utility grid via inverters that rely on various control algorithms. Mostly they offer no inertia, and behave as controlled current sources that produce fluctuating power. Numerous researchers are investigating how the control of future power grids should look.
One of our studies concerns the stability analysis of a very simple system that contains only a SG with its prime mover and a resistive load. Inductive transmission lines connect between them. The model of the SG and its prime mover includes a frequency droop loop that acts through the prime mover and its governor, with its own dynamics. The SG model takes into account the variation of the inductances with the rotor angle and damper windings, as we contemplate to incorporate the effect of damper windings into the control algorithm of inverters.
