Maximizing the time of invariance for large collections of switched systems

Motivated by the problem of capacity allocation in aggregate systems of thermostatically controlled loads, we seek to optimize the time of invariance - the maximal time that a scheduler can maintain local safety constraints without exceeding an aggregate capacity constraint. Having previously charac...

Full description

Saved in:
Bibliographic Details
Published in2017 IEEE 56th Annual Conference on Decision and Control (CDC) pp. 3618 - 3625
Main Authors Nilsson, Petter, Ozay, Necmiye
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.12.2017
Subjects
Aggregates
Optimal control
Safety
Silicon
Switches
Trajectory
Online AccessGet full text

Cover

More Information
Summary:Motivated by the problem of capacity allocation in aggregate systems of thermostatically controlled loads, we seek to optimize the time of invariance - the maximal time that a scheduler can maintain local safety constraints without exceeding an aggregate capacity constraint. Having previously characterized conditions when the time of invariance is infinite, this paper focuses on situations when no scheduler can satisfy all constraints indefinitely. We derive a lower bound for the optimal time of invariance, and leverage ideas from optimal control to propose control strategies with associated lower bounds on the resulting times of invariance. We also analyze the optimality gap. The results are illustrated on a 10,000-dimensional example.
DOI:10.1109/CDC.2017.8264191