Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

BiTSA: Leveraging Time Series Foundation Model for Building Energy Analytics

Xiachong Lin, Arian Prabowo, Imran Razzak, Hao Xue, Matthew Amos, Sam Behrens, Flora D. Salim

TL;DR

BiTSA tackles the difficulty of turning rich but complex IoT energy data from buildings into actionable decisions for managers. It combines a browser-based interactive visualization UI with pre-trained time-series forecasting models to provide real-time insights and proactive energy management. The paper details the BiTSA framework, data preprocessing, model library, training setup, and an evaluation using BTS-B and BLDG datasets, illustrating when different models perform best. The approach offers a practical bridge between academic forecasting methods and real-world building operations, with potential to improve energy efficiency and advance net-zero goals through data-driven decision support.

Abstract

Incorporating AI technologies into digital infrastructure offers transformative potential for energy management, particularly in enhancing energy efficiency and supporting net-zero objectives. However, the complexity of IoT-generated datasets often poses a significant challenge, hindering the translation of research insights into practical, real-world applications. This paper presents the design of an interactive visualization tool, BiTSA. The tool enables building managers to interpret complex energy data quickly and take immediate, data-driven actions based on real-time insights. By integrating advanced forecasting models with an intuitive visual interface, our solution facilitates proactive decision-making, optimizes energy consumption, and promotes sustainable building management practices. BiTSA will empower building managers to optimize energy consumption, control demand-side energy usage, and achieve sustainability goals.

BiTSA: Leveraging Time Series Foundation Model for Building Energy Analytics

TL;DR

BiTSA tackles the difficulty of turning rich but complex IoT energy data from buildings into actionable decisions for managers. It combines a browser-based interactive visualization UI with pre-trained time-series forecasting models to provide real-time insights and proactive energy management. The paper details the BiTSA framework, data preprocessing, model library, training setup, and an evaluation using BTS-B and BLDG datasets, illustrating when different models perform best. The approach offers a practical bridge between academic forecasting methods and real-world building operations, with potential to improve energy efficiency and advance net-zero goals through data-driven decision support.

Abstract

Incorporating AI technologies into digital infrastructure offers transformative potential for energy management, particularly in enhancing energy efficiency and supporting net-zero objectives. However, the complexity of IoT-generated datasets often poses a significant challenge, hindering the translation of research insights into practical, real-world applications. This paper presents the design of an interactive visualization tool, BiTSA. The tool enables building managers to interpret complex energy data quickly and take immediate, data-driven actions based on real-time insights. By integrating advanced forecasting models with an intuitive visual interface, our solution facilitates proactive decision-making, optimizes energy consumption, and promotes sustainable building management practices. BiTSA will empower building managers to optimize energy consumption, control demand-side energy usage, and achieve sustainability goals.

Paper Structure

This paper contains 12 sections, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. BiTSA Framework Design
  3. Front-end UI Interface
  4. Back-end Service
  5. External Resources
  6. BiTSA Experimental Setup and Performance Evaluation
  7. Data Preprocessing
  8. Build-in Models
  9. Model Training
  10. Built-in Model Performance
  11. Strengths and Limitations
  12. Conclusion

Figures (4)

  • Figure 1: The general framework of BiTSA for assisting building managers in fetching accurate building activities based on pre-trained AI models.
  • Figure 2: This diagram presents the architecture of BiTSA. The front-end user interface will facilitate user interaction with the system by managing login credentials and handling requests. The back-end will be integrated with the Building Management System to access and process data, execute core functions, and respond to user inputs. Computation results from the back-end will be sent back to the UI for rendering and visualization, ensuring a seamless user experience.
  • Figure 3: A screenshot of BiTSA displaying interactive time series panel.
  • Figure 4: A screenshot of the Analytics page, displaying the predictions of Outside_Air_Humidity_Sensor generated by One-Fits-All. Building manager can filter the sensor through the dropdown to fetch the forecasting results of their interested sensor.