“21st PSCC 2020 papers submission and review platform

Full Program »

Data-Driven Learning and Load Ensemble Control

View File
PDF
0.8MB

Demand response (DR) programs aim to engage distributed small-scale flexible loads, such as thermostatically controllable loads (TCLs), to provide various grid support services. Linearly Solvable Markov Decision Process (LS-MDP), a variant of the traditional MDP, is used to model aggregated TCLs. Then, a model-free reinforcement learning technique called Z-learning is applied to learn the value function and derive the optimal policy for the DR aggregator to control TCLs. The learning process is robust against uncertainty that arises from estimating the passive dynamics of the aggregated TCLs. The efficiency of this data-driven learning is demonstrated through simulations on Heating, Cooling & Ventilation (HVAC) units in a testbed neighborhood of residential houses.

Author(s):

Ali Hassan    
New York University
United States

Deepjyoti Deka    
Los Alamos National Laboratory,
United States

Michael Chertkov    
University of Arizona
United States

Yury Dvorkin    
New York University
United States

 

Powered by OpenConf®
Copyright ©2002-2014 Zakon Group LLC