Fault Detection and Diagnosis (FDD)

Data-driven fault detection and diagnosis (FDD) for chiller systems.

Explainable AI for Chiller Fault-Detection Systems: Gaining Trust and Human Connect

XAI-FDD workflow.


Chillers are energy-intensive, prone-to-faults components used for space cooling in buildings. Data-driven Fault Detection and Diagnosis (DD-FDD) are widely used for chillers. However, field personnel are often not confident in DD-FDD mainly because no explanation is given for the results. EXplainable Artificial Intelligence (XAI) can help bridge this gap. We investigate XAI-FDD’s role in building trust in DD-FDD. We examine use-cases for XAI-FDD on a building in Singapore having 6 chillers.

Team: Seshadhri Srinivasan (GE Global Research), Pandarasamy Arjunan (BEARS), Baihong Jin (UC Berkeley), Alberto L. Sangiovanni-Vincentelli (UC Berkeley), Zuraimi Sultan (BEARS), and Kameshwar Poolla (UC Berkeley)

Causal 1-D convolutional neural networks based chiller fault detection and diagnosis

Modern buildings are densely equipped with smart energy meters, which periodically generate a massive amount of time-series data yielding a few million data points every day. This data can be leveraged to discover the underlying load and infer their energy consumption patterns, inter-dependencies on environmental factors, and the building’s operational properties. Furthermore, it allows us to simultaneously identify anomalies present in the electricity consumption profiles, which is a big step towards saving energy and achieving global sustainability. However, to date, the lack of large-scale annotated energy consumption datasets hinders the ongoing research in anomaly detection. We contribute to this effort by releasing a carefully annotated version of a publicly available ASHRAE Great Energy Predictor III data set containing 1,413 smart electricity meter time series spanning over one year. In addition, we benchmark the performance of eight state-of-the-art anomaly detection methods on our dataset and compare their performance.

Team: Pandarasamy Arjunan (BEARS), Seshadhri Srinivasan (GE Global Research), and Kameshwar Poolla (UC Berkeley)


Publications:

  1. ACM BuildSys CORE: A
    Causal 1-D convolutional neural networks based chiller fault detection and diagnosis
    Arjunan, Pandarasamy, Srinivasan, Seshadhri, and Poolla, Kameshwar
    In Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation 2022
    LINK PDF Code
  1. IEEE Computer IF: 3.564
    Explainable AI for chiller fault-detection systems: gaining human trust
    Srinivasan, Seshadhri, Arjunan, Pandarasamy, Jin, Baihong, Sangiovanni-Vincentelli, Alberto L, Sultan, Zuraimi, and Poolla, Kameshwar
    Computer 2021
    Abs LINK PDF