This repository contains the occupancy data from the paper "Occupancy Prediction for Building Energy Systems with Latent Force Models". They were collected at the Bosch Research Campus in Renningen, Germany. Additionally, an implementation of the Latent Force Model (LFM) predictor is provided
The LFM is provided by the utilities package inside this repository. Install the
package with
pip install .
and then run compare_prediction.py. This reproduces the prediction results from the
paper in section 4.5.
The occupancy dataset is found in the data folder inside the file Occupancy.mat.
The basic metadata for the whole dataset is found in the table below.
| Specification | Field |
|---|---|
| Start time | 01-Jan-2019 |
| End time | 15-Oct-2022 |
| Sample time | 15 minutes |
| Number of Zones | 9 |
The dataset consists out of every zone in the layout above. Their index and description is found in the table below
| Zone | Index | Description |
|---|---|---|
| 9 | 0 | Meeting Room |
| 10 | 1 | Meeting Room |
| 11 | 2 | Meeting Room |
| 12 | 3 | Kitchen |
| 13 | 4 | Copy Room |
| 21 | 5 | Open Space Office |
| 22 | 6 | Open Space Office |
| 23 | 7 | Open Space Office |
| 24 | 8 | Open Space Office |
Only the occupancy dataset is distributed under the CC-BY-NC-SA License. If using or distributing the data, please cite the following paper (preprint):
@misc{wietzke2024occupancy,
title = {Occupancy prediction for building energy systems with latent force models},
journal = {Energy and Buildings},
volume = {307},
pages = {113968},
year = {2024},
issn = {0378-7788},
doi = {https://doi.org/10.1016/j.enbuild.2024.113968},
url = {https://www.sciencedirect.com/science/article/pii/S0378778824000847},
author = {Thore Wietzke and Jan Gall and Knut Graichen},
abstract = {This paper presents a new approach to predict the occupancy for building energy systems (BES). A Gaussian Process (GP) is used to model the occupancy and is represented as a state space model that is equivalent to the full GP if Kalman filtering and smoothing is used. The combination of GPs and mechanistic models is called Latent Force Model (LFM). An LFM-based model predictive control (MPC) concept for BES is presented that benefits from the extrapolation capability of mechanistic models and the learning ability of GPs to predict the occupancy within the building. Simulations with EnergyPlus and a comparison with real-world data from the Bosch Research Campus in Renningen show that a reduced energy demand and thermal discomfort can be obtained with the LFM-based MPC scheme by accounting for the predicted stochastic occupancy.}
}