A Data-driven Study of Connected Residential Thermostats to Investigate user Behavior, Thermal Modelling, and Optimal Control of HVAC Systems
Bibliographic record
Abstract
Approximately a tenth of North America’s total energy usage is for space conditioning of residential buildings -- the majority of which is controlled by a thermostat. Despite these energy implications, thermostats have been difficult to study and their controls have remained reactive and heuristic-driven. Fortunately, with the latest generation of devices, connected thermostats, it is now possible to address these limitations. This thesis is among the first major bodies of research to exploit the emerging data from connected thermostats. Specifically, we seek to accomplish three main objectives: (1) extend the understanding of how users utilize these devices to manage their preferences, (2) develop predictive models of occupant behavior and thermal response of houses, and (3) optimize the control of residential HVAC systems using only existing available data.For objective (1), we determined that factors such as location and seasonality affected setpoint preferences, but that it was unclear if truly distinct user types emerged -- instead users appeared more on a spectrum of preferences. With schedule overrides, the behavior of users was found to be more complex than previously understood, with ultimately only a small group of users remaining in permanent and energy-intensive overrides. For objective (2), we found standard, well-tuned machine learning models (namely, random forest and ridge regression) were the most robust performers beating simple baseline and deep learning methods for predicting occupancy and thermal responses of the houses. Finally, for objective (3), we found that a data-driven model predictive controller outperforms a model-free reinforcement learning method and a standard deadband controller. In summary, this thesis makes multiple novel and significant contributions which improve the understanding of connected thermostat users, how to develop customized data-driven models, and how to improve the comfort and energy use associated with connected thermostats controlling the HVAC in our homes.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.003 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".