Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Indoor thermal comfort management: A Bayesian machine-learning approach to data denoising and dynamics prediction of HVAC systems

Javier Penuela, Sahar Moghimian Hoosh, Ilia Kamyshev, Aldo Bischi, Henni Ouerdane

TL;DR

This work applied a method that merges several Bayesian machine learning and deep learning architectures that are suitable for predicting complex system dynamics, while relaxing the dataset quality requirements, and achieves state-of-the-art performance.

Abstract

The optimal management of a building's microclimate to satisfy the occupants' needs and objectives in terms of comfort, energy efficiency, and costs is particularly challenging. This complexity arises from the non-linear, time-dependent interactions among all the variables of the control problem and the changing internal and external constraints. Focusing on the accurate modeling of the indoor temperature, we propose a data-driven approach to address this challenge. We account for thermal inertia, non-linear effects, small perturbations of the indoor climate dynamics caused by ventilation and weather variations, as well as for the stochastic nature of the control system due to the observed noise in the input signal. Since the prohibitive cost of quality data acquisition and processing limits the implementation of data-driven approaches for real-life problems, we applied a method that merges several Bayesian machine learning and deep learning architectures that are suitable for predicting complex system dynamics, while relaxing the dataset quality requirements. Our framework includes a built-in deep Kalman filter, which makes it deployable even with low-accuracy temperature sensors. It achieves state-of-the-art performance, best performing with a 150-minute prediction horizon with an RMSE of 0.2455, an MAE of 0.162, and an $R^2$ of 0.926. The model's performance remains consistent even when exposed to highly noisy data. Finally, we show how our approach can be extended to other applications including demand response event duration prediction and equipment failure detection.

Indoor thermal comfort management: A Bayesian machine-learning approach to data denoising and dynamics prediction of HVAC systems

TL;DR

This work applied a method that merges several Bayesian machine learning and deep learning architectures that are suitable for predicting complex system dynamics, while relaxing the dataset quality requirements, and achieves state-of-the-art performance.

Abstract

The optimal management of a building's microclimate to satisfy the occupants' needs and objectives in terms of comfort, energy efficiency, and costs is particularly challenging. This complexity arises from the non-linear, time-dependent interactions among all the variables of the control problem and the changing internal and external constraints. Focusing on the accurate modeling of the indoor temperature, we propose a data-driven approach to address this challenge. We account for thermal inertia, non-linear effects, small perturbations of the indoor climate dynamics caused by ventilation and weather variations, as well as for the stochastic nature of the control system due to the observed noise in the input signal. Since the prohibitive cost of quality data acquisition and processing limits the implementation of data-driven approaches for real-life problems, we applied a method that merges several Bayesian machine learning and deep learning architectures that are suitable for predicting complex system dynamics, while relaxing the dataset quality requirements. Our framework includes a built-in deep Kalman filter, which makes it deployable even with low-accuracy temperature sensors. It achieves state-of-the-art performance, best performing with a 150-minute prediction horizon with an RMSE of 0.2455, an MAE of 0.162, and an of 0.926. The model's performance remains consistent even when exposed to highly noisy data. Finally, we show how our approach can be extended to other applications including demand response event duration prediction and equipment failure detection.

Paper Structure

This paper contains 26 sections, 11 equations, 11 figures, 8 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Methodology
  3. Indoor microclimate model and simulation
  4. Indoor microclimate probabilistic description
  5. Neural network architecture
  6. Architecture of CNN for denoising
  7. Architecture of RNN for dynamics prediction
  8. Training data
  9. Results
  10. Indoor temperature prediction
  11. Simulation validation
  12. Performance metrics
  13. Measurement noise
  14. Fine-tuning and specialization strategies
  15. Applications
  16. ...and 11 more sections

Figures (11)

  • Figure 1: Graphical representation of single thermal zone test case used for synthetic data generation.
  • Figure 2: Example of synthetically generated data using the model presented in Section \ref{['simulation']}.
  • Figure 3: Architecture of the denoiser.
  • Figure 4: Architecture of one RNN cell
  • Figure 5: Schematic illustration of the data flow at the prediction stage.
  • ...and 6 more figures