A Practical Data-Driven Multi-Model Approach to Model Predictive Control: Results from Implementation in an Institutional Building
Bibliographic record
Abstract
Model-based Predictive Control (MPC) is an effective solution to improve building controls. It consists of the use of weather and occupancy forecasts along with a control-oriented model to predict the behaviour of the building a few hours or days ahead, and thus optimize the operation of its systems. Although the potential of MPC is widely recognized, and plentiful operational data is often available, the development of a model requires a great deal of effort, significant technical expertise and knowledge of building systems. The challenge of creating a model is a hurdle that makes the on-site implementation of MPC in buildings relatively rare. This study tackles the development of a multi-model approach to optimize the operation of electric and natural gas boilers in an institutional building to reduce greenhouse gas (GHG) emissions while maintaining the required level of comfort. This methodology leverages Machine Learning techniques to rapidly develop and calibrate control-oriented models using a limited number of input variables (indoor air temperature and temperature set-points, weather conditions, power meter data). The proposed multi-model approach consists of five models used to estimate the building total heating demand, the electric baseload, the natural gas boiler power, and the indoor air temperature under free floating conditions and during warming-up periods in the morning. The models are calibrated and validated with operational data and they are then used to optimize the transition between nighttime and daytime indoor air temperature. Since these are black-box models that require only a basic understanding of the building system and a few inputs, the model development was considerably reduced while the modularity of the proposed method makes it flexible. Such an approach could therefore be easily replicated in other buildings equipped with similar pieces of equipment. This methodology has been implemented in a Canadian institutional building, located in Varennes (QC). Results in 2020-21 showed that the COVID-19 pandemic has significantly impacted building performance and reduced energy use, thus creating a new baseline. The MPC strategy allowed to achieve an additional 20.2% GHG emission reduction compared to this new baseline while thermal comfort was improved. Nevertheless, energy costs increased, which was mainly due to the impact of the pandemic, which eventually made the pre-COVID-19 model and optimization parameters outdated; lower costs are expected with model recalibration, currently ongoing.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 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 source (direct Gemma or distilled Codex), 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".