Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Improving Building Temperature Forecasting: A Data-driven Approach with System Scenario Clustering

Dafang Zhao, Zheng Chen, Zhengmao Li, Xiaolei Yuan, Ittetsu Taniguchi

TL;DR

The paper tackles the challenge of scalable, accurate room temperature forecasting for large-scale HVAC systems by introducing a data-driven, system-scenario framework. It clusters time-series operation data into distinct system scenarios using $k$-means, extracts rich time-series features, and trains interpretable symbolic regression models per cluster to enable one-day-ahead predictions; online data are assigned to the nearest cluster for immediate forecasting without re-training. The work provides a practical pipeline, including real-world data collection, cluster-based modeling, and SR-based prediction equations, demonstrating reduced modeling time while maintaining or improving accuracy relative to a CNN-LSTM baseline. The approach offers a scalable path for proactive HVAC control and energy management in buildings, validated with real operation data for cooling and heating scenarios.

Abstract

Heat, Ventilation and Air Conditioning (HVAC) systems play a critical role in maintaining a comfortable thermal environment and cost approximately 40% of primary energy usage in the building sector. For smart energy management in buildings, usage patterns and their resulting profiles allow the improvement of control systems with prediction capabilities. However, for large-scale HVAC system management, it is difficult to construct a detailed model for each subsystem. In this paper, a new data-driven room temperature prediction model is proposed based on the k-means clustering method. The proposed data-driven temperature prediction approach extracts the system operation feature through historical data analysis and further simplifies the system-level model to improve generalization and computational efficiency. We evaluate the proposed approach in the real world. The results demonstrated that our approach can significantly reduce modeling time without reducing prediction accuracy.

Improving Building Temperature Forecasting: A Data-driven Approach with System Scenario Clustering

TL;DR

The paper tackles the challenge of scalable, accurate room temperature forecasting for large-scale HVAC systems by introducing a data-driven, system-scenario framework. It clusters time-series operation data into distinct system scenarios using -means, extracts rich time-series features, and trains interpretable symbolic regression models per cluster to enable one-day-ahead predictions; online data are assigned to the nearest cluster for immediate forecasting without re-training. The work provides a practical pipeline, including real-world data collection, cluster-based modeling, and SR-based prediction equations, demonstrating reduced modeling time while maintaining or improving accuracy relative to a CNN-LSTM baseline. The approach offers a scalable path for proactive HVAC control and energy management in buildings, validated with real operation data for cooling and heating scenarios.

Abstract

Heat, Ventilation and Air Conditioning (HVAC) systems play a critical role in maintaining a comfortable thermal environment and cost approximately 40% of primary energy usage in the building sector. For smart energy management in buildings, usage patterns and their resulting profiles allow the improvement of control systems with prediction capabilities. However, for large-scale HVAC system management, it is difficult to construct a detailed model for each subsystem. In this paper, a new data-driven room temperature prediction model is proposed based on the k-means clustering method. The proposed data-driven temperature prediction approach extracts the system operation feature through historical data analysis and further simplifies the system-level model to improve generalization and computational efficiency. We evaluate the proposed approach in the real world. The results demonstrated that our approach can significantly reduce modeling time without reducing prediction accuracy.
Paper Structure (11 sections, 4 equations, 7 figures, 2 tables)

This paper contains 11 sections, 4 equations, 7 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Scenario-based Data-driven Temperature Prediction
  3. Data Collection and Preparation
  4. System Scenario Based Clustering
  5. Time-series Data Feature Extraction
  6. System Scenario Clustering
  7. Temperature Prediction with Symbolic Regression
  8. Simulation
  9. Temperature Prediction Under Cooling Operation
  10. Temperature Prediction Under Heating Operation
  11. Conclusion

Figures (7)

  • Figure 1: Diagram of temperature prediction workflow.
  • Figure 2: Room temperature distribution over week and day, along with the seasons.
  • Figure 3: (a) Silhouette coefficient for number of clusters, (b) Clustering results for all observation period.
  • Figure 4: symbolic regression-based on a binary expression tree.
  • Figure 5: Overview of creating temperature prediction function using symbolic regression.
  • ...and 2 more figures