Loading...
Assessment of stochastic control methods for home energy management using a high-fidelity residential energy model and realistic levels of uncertainty, An
Blonsky, Michael
Blonsky, Michael
Citations
Altmetric:
Advisor
Editor
Date
Date Issued
2022
Date Submitted
Collections
Research Projects
Organizational Units
Journal Issue
Embargo Expires
Abstract
In the residential sector, there is considerable growth in smart, interactive devices including thermostats, water heaters, electric vehicle chargers, solar panels, and battery systems. These devices can all provide the electric grid with additional flexibility, which is a valuable resource for grid operators and helps reduce system costs and emissions. However, it is difficult to control these devices because there is considerable diversity in the residential housing stock and because the controls depend on stochastic variables such as weather and occupant behavior.
We present a stochastic control framework for home energy management systems that can control these devices while accounting for multiple sources of uncertainty. To do this, we first develop an integrated residential energy model that can simulate multiple controllable devices with high resolution and can interface with external controllers. We then design a control framework with a linear residential energy model, a forecast generator that provides realistic estimates of weather and occupancy variables, and a control objective that captures energy costs and occupant comfort. The control framework is used to evaluate the performance of heuristic, deterministic, and stochastic control methods, primarily model predictive control.
We show that the stochastic model predictive control performs best in scenarios with realistic levels of uncertainty. We also validate the residential energy models and show the benefits of high-fidelity modeling for building-to-grid co-simulation studies. The results shown in this dissertation provide a deeper understanding of residential load flexibility in uncertain conditions, and the frameworks developed enable future research for evaluating flexible loads in a broad set of applications.
Associated Publications
Rights
Copyright of the original work is retained by the author.