Share

Beyond average: evaluating indoor average temperature in grey box modeling.

Number: 3165

Author(s) : MULAYIM O. B., BERGES M.

Summary

Grey Box Modeling (GBM) strategies have proven effective in identifying the thermal behavior of buildings. The widespread adoption of Smart Thermostats (STs) has unveiled large datasets, which are now being harnessed for such modeling strategies. A notable limitation of earlier GBM approaches for STs was their sole focus on a single sensing point. With the proliferation of additional sensors in homes, STs have begun integrating more sophisticated control algorithms such as conditioning based on the average temperature of occupied rooms. When conducting GBM using the Indoor Average Temperature (IAT), the modeling assumptions can falter due to the inconsistent inclusion of sensors in averaging for the control temperature. In this study, we leverage a publicly available dataset from 1,000 homes across the United States, each outfitted with varying numbers of sensors, to identify grey box models. Beyond one-step ahead prediction models, we apply Model Predictive Control Relevant Identification to assess the effectiveness of room-level modeling in one-day ahead predictions. Our findings indicate that varying occupancy profiles typically result in a 0.5 to 3.3°F discrepancy in the accuracy of one-hour-ahead predictions when utilizing IAT for modeling. Also, distinctly modeling individual sensors yielded a 1.8 to 25.8% enhancement in the accuracy of one-hour and one-day ahead predictions, respectively.

Available documents

Format PDF

Pages: 10 p.

Available

Free

Download

Details

  • Original title: Beyond average: evaluating indoor average temperature in grey box modeling.
  • Record ID : 30032966
  • Languages: English
  • Source: 2024 Purdue Conferences. 8th International High Performance Buildings Conference at Purdue.
  • Publication date: 2024/07/15

Links


See other articles from the proceedings (63)
See the conference proceedings

Indexing